I think teaching is one of the few roles that can't be replaced by AI. If you're a self-motivated learner, eager to gain new skills, then AI is perfect for you. Having a virtual Feynman coach you through a Physics course is perfect.
Most learners, the world over, are not self-motivated. The pandemic showed us exactly what children would prefer to do, when they don't have a physical teacher standing over them, which is bugger all. We send kids to school, in the hope they get some education, but the reality is that we use schools for free childcare while we work. If parents have to additionally monitor their child's learning, it breaks down pretty quickly.
I see AI being more of a teaching assistant, rather than a replacement for teachers. Having been in the education game for over twenty five years, I know the difference in impact when comparing virtual learning to in-person training.
AI certainly can't completely replace teachers, but the potential gains for personal tutoring from SOTA LLMs still seem enormous to me.
And I'm not trying to make a general argument against in person training. But I think the details of how virtual learning happens matters quite a lot. AI can make it much more personalized and make tutoring relatively affordable. Don't you think?
AI has personally tutored me about obscure, deep linear algebra concepts. It's so great to get applied examples and be able to ask why/how something works, rather than reading a stuffy Wikipedia article or math textbook.
It's been extremely effective for me, where reading a math textbook/wikipedia article seemed like too much effort, but a friendly conversation with my AI tutor was just fine.
How can you bring yourself to trust the AI? Just yesterday a friend and I asked Chat-GPT a physics question, and for some reason his assistant asserted that the speed of light was 3,000 m/s, which is off by two orders of magnitude. We know that's wrong so we can tell the AI to do it again but right this time, but if it was explaining a concept we didn't already understand, I can't see how the output would be any more meaningful than asking a random stranger and trusting their response.
Ever since the step(s) beyond ChatGPT 3.5 I haven't noticed any huge errors like that, personally. Are you sure you were on a new model?
Also, how can you trust anyone? People are wrong. Teachers can be wrong. Web pages can be wrong. Books can be wrong. I think LLMs will probably soon be the least likely to be wrong out of any of those.
A very easy way to get basically every current AI model to hallucinate:
1. Ask a highly non-trivial research question (in particular from math)
2. Ask the AI for paper and textbook references on the topic
At this point, already many of these references could be hallucinations.
3. If necessary ask the AI where in these papers/textbooks you can find explanations on the questions, and/or on which aspect of the question or research area the individual references focus.
This backs up what I mentioned in my other comment. My dad, an attorney, purchased both gpt-4o and Gemini Advanced to help write legal documents, which involves citing other legal cases. He says that he's found the legal cases that both models cite to almost always be completely fabricated.
I just asked chatgpt: "comparing 9.9 and 9.11, which is larger?"
and it responded:
9.11 is larger than 9.9.
When comparing these two numbers:
9.9 can be written as 9.90 to have the same number of decimal places.
9.11 remains 9.11.
Comparing digit by digit:
The integer part (9) is the same for both.
The first decimal place (9 vs. 1): 9 is larger.
The second decimal place (0 vs. 1): 1 is larger, which makes 9.11 larger overall.
My dad, a lawyer, has been trying to use gpt-4o to assist in writing legal documents. He has said that the documents are well written and convincing, but the cases that are cited by 4o to support the document are more often than not completely made up.
How can you bring yourself to trust a human teacher? Humans are wrong sometimes too, often with confidence.
The trick to learning effective timely (with both LLMs and human teachers) is to recognize that you should learn from more than one source. Think critically about the information you are being exposed to - if something doesn't quite feel right, check it elsewhere.
I genuinely believe that knowing that an information source is occasionally unreliable can help you learn MORE effectively, because it encourages you to think critically about the material and explore beyond just a single source of information.
I've been learning things with the assistance of LLMs for nearly two years now. I often catch them making mistakes, and yet I still find them really useful for learning.
If you can't trust a teacher or a textbook, then you are in big trouble. Especially if it is a brand new subject to yourself where you don't have an intuition about what is correct/incorrect. Part of a teacher/student relationship is obviously trust.
No, you aren't. You can listen to ideas and think about them and attribute them to the sources and come to (or not) your own conclusions.
The reason it's such a bad idea to "trust" the way you are suggesting is that many fields are quackery. Do you trust that fancy textbook and sophisticated sounding professor from first year macroeconomics?
>
How can you bring yourself to trust a human teacher? Humans are wrong sometimes too, often with confidence.
If humans/AIs are wrong about a topic (in particular wrong in a confident way) multiple times, I will stop trusting them to be experts in the topic. What I experienced is rather that many human experts in academia tend to be honest when they are not sure about the answer.
A human understands what they're saying. If a human teacher is working through a math problem and isn't sure of their work, they're able to stop and correct their mistake. An AI math teacher is trained on a corpus of data - probably very similar to the data that the human teacher was trained on, though I'm sure the AI was trained on far more data than any single human ever was - but can't do the introspective part. To put it another way, I think we agree that humans learn better by assessing multiple sources and thinking critically. An AI is very good at the former, but very bad at the latter, and I would rather have a teacher that can think critically about what it is saying to me.
Trust but verify. If you're doing your homework you should be able to notice things not lining up and ask the model about them. Human teachers can also make mistakes (though usually less than an AI hopefully) and it's the same process dealing with those.
In my opinion the best teachers just direct your questions in the direction where the answers you find give you the most useful information. I'm optimistic that AI could be an improvement to the average for scientifically minded learners, though I wouldn't expect it to be more effective than a 1 on 1 with a good teacher.
This problem isn't exclusive to current implementations of AI.
I had a US business professor explain in one of my business classes that making a bit more money might push you over into the next tax bracket and cost you more in taxes than you made.
This guy had a PhD, had been teaching for decades and apparently didn't understand the marginal tax system.
> I had a US business professor explain in one of my business classes that making a bit more money might push you over into the next tax bracket and cost you more in taxes than you made.
He's not wrong. You are correct if you consider only income taxes. But there are other tax benefits that lead to discontinuities with respect to income.
As an example, in my state you can deduct up to $5000 of contributions to a 529 plan if your income is under $250K. Go a penny above that threshold, and you can deduct only $2500. That extra penny just reduced your refund by a few hundred dollars.
In the Netherlands we have a marginal tax rate, so every Euro over X gets taxed 10%, everything over Y gets taxed 15% etc. (simplified numbers obviously).
However, often times it's better to stay in the top of a lower bracket because of tangentially-related benefits, such as healthcare subsidies, rent subsidies and other things like that. If you go from tax bracket 1 -> 2 because you get a 100 euro raise, sure you'll get 100 euros more (well, more like 95 but whatever), but you also lose out on more than that in the form of a loss in other benefits.
My partner went through this recently, she got a raise at work, but as a result she actually lost the subsidized rent money she got from the gov't. She had to request her workplace lower her wage so she was under the limit, because otherwise she couldn't have afforded rent on her own, and if the raise was even 2 euros/hr higher, she might've even been kicked out of her social housing situation.
That's because the benefits aren't marginal, they work on a hard cut-off limit. Anything over X amount and you're just cut off, you're not gradually weened off it until you're at a high enough income to not require gov't help.
Nitpick: Your number of orders of magnitude is off by a (binary) order of magnitude.
The speed of light is about 300,000,000 m/s. (In fact it's exactly 299,792,458 m/s, because that's how the metre is defined.) So 3,000 m/s is off by five (decimal) orders of magnitude, not two.
A likely truth no one wants talk about : LLMs will only help people who want to learn. Those people are likely already in very good shape in life. The amount of help from LLMs is likely very high for such people - as you note the ability to have a back and forth is very helpful.
For 99% of the population, they aren't going to do this. It is what it is.
Gotcha. So I guess the question is, can an AI run a Zoom meeting or interactive multiplayer learning game with a bunch of kids on it? Have to admit that might be a stretch.
> Most learners, the world over, are not self-motivated
this seems like a bizarre conclusion. In my experience, most people, the world over, are in fact self motivated. You won't see that if you have a very narrow definition of what is it that they're supposed to be learning
kids aren't motivated to do boring math drills, because they don't see why it matters to their life (the real answer is: it does not, they are not wrong).
I think it might be worth considering whether you’ve had a privileged upbringing. Thinking back on it, the majority of people probably would have been content to play games all day. You could argue that that’s learning, but unfortunately it’s not the kind of learning that society tends to reward.
I’ve heard that kids in upper middle class circles are totally different in this regard though. Maybe they want to do more on average.
I think this perspective is belied by the vast over-subscription of free public education in places where it has previously been paid only[1] (at this point, mainly in Africa). It does seem like there is strong evidence that most children and parents recognize the value of education and are self-motivated to pursue it where it is accessible to them. I believe it follows that lowering cost and barriers to quality education will improve outcomes without a need to otherwise coerce participation.
Not really. In my experience it is mostly effect of socio-legal pressure that kids can't be anywhere but school. In primary schools most kids are bored or miserable as hell while in school. And further parents keep pushing it because apparently education is key to future success / great career.
For higher education there is charade of education to get jobs. So for office manager job where grade 8 would be enough, we have MBAs now because we all need advanced education to survive in global economy blah..blah.
> For higher education there is charade of education to get jobs. So for office manager job where grade 8 would be enough, we have MBAs now because we all need advanced education to survive in global economy blah..blah.
This would actually be a good business opportunity: hire such "grade 8 educated" people as office managers, but pay them much less than MBAs. If they are nearly as good as MBAs, you save a lot of money on this group of employees, and thus your company has a strong economic advantage.
> hire such "grade 8 educated" people as office managers, but pay them much less than MBAs.
The trouble is it's performance all the way up and down. In the first place you're only going to get the weirdos / extreme gamblers, and then you'll struggle to attract investors, your clients/suppliers will wonder why your business development folks missed their classical references...
> In the first place you're only going to get the weirdos
You will (hopefully) nevertheless check whether an applicant has the necessary traits to be a decent office manager. On the other hand, I wouldn't claim that weirdos are necessarily bad office managers.
Yeah if anything, current education system is so garbage that it manages to completely demotivate curious kids who want to genuinely learn. It's designed around adults that need to run the place, runs at the wrong pace for most students and focuses on PTSD-inducing high anxiety testing constantly because it's easy to do for the teachers. Not to mention piles of pointless busywork as homework that's been proven to not help with learning at all.
> kids aren't motivated to do boring math drills, because they don't see why it matters to their life (the real answer is: it does not, they are not wrong).
Most kid athletes are also not self-motivated to run laps, or do boring repetitive drills, when they know from experience that these activities help them win games within the next few months. Usually need a coach to force them to do them. Same for young music players. Practicing scales endlessly does make you a better musician. But they won't do it till forced.
The primary reason kids don't like running laps or playing scales or doing math drills is because they are boring.
Reminds me of a recent podcast with Staša Gejo [1], a top competition climber. She basically says the same thing. At times she hated being told to do drills growing up, but really valued that later because as a kid she sometimes didn't feel like doing the hard work necessary for the outcomes she desired.
> Practicing scales endlessly does make you a better musician. But they won't do it till forced.
The value of such exercises, or any other drill-based curriculum, must be measured with its opportunity cost. If you practise scales for an hour a day, you can indeed reliably expect to be better at your musical instrument, but it could very well be that the same hour spent on improving another skill (sight-reading, articulation etc.) would make you considerably better still at your instrument.
I think it might be more generally useful to say that, in order to develop well-rounded competency in a given field, one should expect to sometimes have to perform boring drills.
Feedback is a critical part of education as well as motivation for learning. But the act of giving feedback is very hard to scale, even for virtual learning. Enter an LLM chatbot, which is imperfect but can fill a lot of gaps in expectation. Chatbots certainly aren’t for everybody, but the large gains in accuracy in years past make them on average more effective.
>kids aren't motivated to do boring math drills, because they don't see why it matters to their life (the real answer is: it does not, they are not wrong).
I think you are partially right in that the dryness of much of math teaching hides a lot of the underlying material's applicability to life. I think one thing AI could do is help design rich situational lessons that could are prompted, vetted, and updated by teachers and then taught to the class. It could be trivial to create incremental difficulty of problem materials tailored to each student's progress and goal.
> In my experience, most people, the world over, are in fact self motivated.
In your experience? The world over? Can you tell me your experience. I've been a teacher for a long time. I've worked in the UK, the USA, PNG, and Kenya.
The vast majority of kids in the developed world don't really care about education. A few do, and they get great grades. Most care more about social status, their cliques, or just surviving the jungle that is school.
School is important. It teaches you how to deal with other people. It teaches you how to deal with people in authority. You can't get that at home, in front of a screen. Learning stuff is secondary. I'm sure there are plenty of people here that are not working in whatever they majored at.
School shouldn't be primarily about experiencing social interaction. It's an artificial environment that disappears as soon as you graduate, and which you'll never find again anywhere else in society. You can learn social interaction in plenty of other settings, most of which are vastly more efficient and realistic. Admittedly, none of them function as daycare...
School should be (and used to be) about learning to learn, building mental discipline and a base of knowledge sufficient to bootstrap whatever other studies appeal to the student, even more so than memorizing a particular list of facts. But it seems that that position has been largely abandoned.
This is circular, how do you propose making school about that? If you’re only goal is to maximize the folks who like to obey authority then great, and maybe that’s all you care to do, and maybe you don’t care about losing the kids who don’t have the academics to make it, but you also lose a whole mess of kids at the top end of the spectrum too.
I'm not sure which part of my comment would result in maximizing folks who like to obey authority. I'm more focused on improving individual outcomes in terms of functional individuals, their quality of life, and the contributions they're able to make to society as a whole.
In any case, we homeschool.
I haven't really considered how to improve schooling at scale (particularly in an affordable way), but my proposal would be to introduce a _lot_ more granularity to schooling by eliminating the idea of grades and classes and focusing more on individual assessment.
Obviously this is likely cost prohibitive, but perhaps promoting and subsidizing homeschooling and homeschool co-ops is a good start in that direction, and could give rise to more cost-effective solutions over time. Not all parents are equipped to homeschool, but homeschooling does make use of resources which could be improved and which others could leverage as well.
I’m mixed, I definetly wouldn’t home school my kids and it doesn’t seem scalable and I do think there’s value in a population having a shared identity from education, but, at least from my own experience I suspect my kids will have their most valuable academic opportunities outside of school.
Community gardens, sports, religious or interest groups, collectives, contributing in a large household, early work experience, hobbies/interests. It probably is a fairly finite list because societies have optimized for the individual and people are often only active within of a community at work or in education facilities. So a part of a solution in my view would be establishing more communities that are separate from the family... they might look a lot like schools though, so maybe we should just focus on those? There's more need for new communities to be established for other age ranges.
I pretty much agree 100%. We need more, smaller communities – and we need them offline.
Notably, this is largely an American problem, since America is built around cars, which given the capitalist nature of American society proves to be antithetical to establishing local communities.
European and other countries, whose layouts and culture were established in pedestrian days, are much better off in this regard.
How does being an employee differ so much from being a student? You still get either good or bad grades for your work. You do assignments, get rules and processes you have to follow, play well with your fellow students/colleagues, etc.
I'm mostly talking about the rather artificial division of students into grades of equal ages without taking into account the individual's proclivities, abilities and achievements. This separation is entirely contrary to organic human self-organization (even in work places) from a tribal perspective and results in a great deal of social illnesses (bullying, cliques, etc.) that are, although found elsewhere, exacerbated by the artificiality of the group-making (which is necessary for the public school model as it currently exists today to function).
Not sure what you're referring to, tbh. Are you talking about the occasional student being promoted or held back a grade? If so, I would say that isn't a granular enough separation to be meaningful.
At least in Belgium, at 12, kids get divided into different schools and systems. You can go into trades, which expects you to be ready at 18 to go work as a plumber, electrician, mechanic, secretary, cook, etc. Or you go towards higher education and so get more focus on math and/or languages. At 18 you are expected to not be ready for the job market but study further.
For my own kids, you really see the differences at 11 to become more profound in their classroom, where some are running behind while others excel. So I saw that 12 is indeed the age where a split is necessary.
You don't have that?
Edit: After some Wikipeding, I see you really don't have that. Wow, that is crazy in my opinion. I have no clue how you could ever organize those differences among students. Here is the system of Belgium: https://en.m.wikipedia.org/wiki/Education_in_Belgium.
For example my daughter is 12 and she is in a class "STEM and moderns languages". Very focussed on programming, science, math, and they also get Dutch, French and English. Some of her classmates at the end of the schooyear had to switch because their grades weren't good enough.
So in Belgium, there are even different schools for different tracks.
That sounds so dystopian. What do kids even do at 12? How can you ask them to decide the course of their life at 12, generally they get segregated while choosing an undergrad major around 16 right!
At 12, there are already big differences among them.
Let's divide it into kids that like to learn vs kids that like to work with their hands. The 2nd group would learn how to work with metal, wood, .... Way more practical stuff, very little theory. The other group is the reverse of course.
At 16, some kids are really tired of sitting in a classroom. For example plumber track would have these 16 year olds already doing an internship with real plumbers. At 18 they can start their own company already.
It's also no secret that a 16 year old in Belgium learns math that in US you would only see at higher education.
I think it works great, and what I saw with my 2 oldest is that last year when they are all still together at age 11, the learning differences really start to show. It's neither fun for the smart ones nor for the slower ones.
Work environments tend to be class sorted. You also have recourse to handle people who behave horribly towards you. Disruptors are removed. Everyone is generally aligned towards the same goal. The two are vastly different.
Anyone who hits someone, says truly horrible shit about others, doesn’t do the job at all, constantly distracts others while doing a poor job themselves, blatantly sexually harasses people, et c, is highly likely to get fired from a job, and may go to prison.
The same person as a student, gets tons of chances before maybe having to leave. Depending on what they’re doing, you could just be stuck with them for north of a decade. No escape.
I mean, we can joke about how actually such people still exist at work, but it’s a far less widespread problem and manifests differently.
You can look for other jobs if you want out of a bad work environment. Probably, you’ll be able to find somewhere else to go. Getting out of a bad class is way harder.
As discussed in another thread already, it seems my own Belgian system differs very much from the US system. (I basically found out with this discussion :D)
And to be honest, I have no idea how you can make the US system work, so I probably agree with you. Come to Belgium and let your kids study here :D.
Actually, until the mid 20th century almost everyone agreed that school was about building character, which can only be done in a social environment. As a British government report put it in 1846, schools should be "a little artificial world of virtuous exertion".
Mmh, I have to read more about this, I'm not really familiar with British schooling models of the 19th century.
In any case, the problematic schooling model that persists to this day was introduced around the time of the Industrial Revolution, which predates your references.
Learning stuff is 100% secondary. If it wasn't these two below would have same-ish chances in life/career.
Student A: Went to College X and majored in Y. Finished all XXX number of credits and graduated with Bachelor's Degree
Student B: Went to same College X and majored in same Y. Finished all XXX-1 number of credits so is 1 credit short and never got a degree.
Student B is worthless even though she/he learned exactly the same thing as Student A. School (especially in USA) never was and never will be about learning ...
I agree that learning is secondary in practice, but in theory is the whole point of school, and society would be a lot better off if we managed to draw theory and practice closer together.
In your example, I would argue you haven't taken my position to its required logical extent. I don't believe in the value of college degrees at all, the way things are currently structured, and I would discourage my kids from going to college unless they had a very specific career path in mind for which the degree is required. The measurement of learning has become the goal of learning, unfortunately.
Yes teaching how to learn is the way for schools, but it is hard to explain to kids and lots of adults.
Just a nitpick that school enforcing memorizing particular list of facts or memorizing poems - is indeed teaching people how to learn, because how else will you explain to a child or an adult "hey you know if you read this thing 10 times and then try to repeat it another 20 times from memory - guess what !!! that is one trick to learn to memorize something."
But if they spend time on finding out how to memorize hand picked for them stuff and how to perform on exams on limited and picked topics - that sounds like they will be able to learn anything but still too many don't realize what the real lesson there is.
If the goal is learning how to memorize, I think it would be better to have specific courses or lessons on memorization rather than making all the subjects needlessly boring and stressful just for the sake of learning memorization techniques.
Knowing how to memorize is occasionally useful depending on one’s profession, but I don’t think it deserves such a heavy emphasis. People naturally remember what is interesting and useful to them. Force-feeding facts or dates or speeches or poems into memory is not fun and has very little educational value. That time and energy would be much better spent on projects that use and integrate the knowledge.
I won’t contest you entirely, but I do ask this: if society functions like a school environment, why do we need a school environment to learn social interaction? It seems at best a secondary or tertiary benefit of a structure that should primarily be focused on intellectual learning. It’s a facsimile of the real world, and facsimiles are always lacking.
Also, I think the biggest thing that present in school environments but missing from other environments is the forced segregation by age range, which prevents the more organic tribal self-organizational practices to which humans have adapted over tens or hundreds of thousands of years.
And I would also add that there are degrees of artificiality and what really matters is not so much whether we have constructed our environments so much as whether or not they have stood the test of time, and adapted to us as much as we’ve adapted to them.
I mean ... many people went to school as students for a good 12 years. That's likely tainted experience, less objective perhaps, but nontheless valid experience.
Those same people will say "I wish school taught me <X> (things like how to fill out a check book etc"
It did. You didn't pay attention. You want school to teach media literacy? It did, but you complained the whole time "when are we going to use this?". My school taught us how to interpret a "source", how to write well defended arguments (even if I don't always rise to that level), how to calculate mortgage interest rates and payments etc etc etc.
But people will swear up and down "school doesn't teach anything important"
because they didn't pay attention to what was taught!
The primary problem with education in america today is that a huge proportion of parents do not give a fuck about education, see school as just a thing you have to do instead of a constant opportunity. When a kid sees their parent complaining about education being "Liberal brain washing" every other day, why would they pay attention in class?
Education requires emotional and ideological buy in from parents and students for best results.
It didn't. I specifically remember the hole where my class should've explained what a court case actually determines, how the process runs, because I noticed things were missing / being swept under the carpet (we did some mock jury stuff but without the right context for it to teach us anything).
> You want school to teach media literacy? It did, but you complained the whole time "when are we going to use this?"
Nope. I specifically remember watching movie "making of" videos and my teacher saying "write this down!" about irrelevant technical details that distracted from the more important stuff the director was saying.
> But people will swear up and down "school doesn't teach anything important"
> because they didn't pay attention to what was taught!
Nope. I was an engaged, attentive student (indeed a kind of star pupil). I still learnt more in spite of school than because of it. Indeed some of the stuff that's most important for my education and career is stuff I was actively punished for doing at school (poking at the computers to see how they worked). Schools, at least the standard-ish state school that I went to (which had good official ratings) teach the wrong things and teach them badly.
While I'm thankful for schooling for teaching me a variety of things, occasionally recognizing my passions and helping to hone them, it is amusing the disruptive and annoying things I did with the school computers that ultimately became my carreer.
Agreed. There's a really specific form of nostalgia where a matured and adhd-managed person would simply LOVE to do school over again now they see the benefit and realise how absolutely fascinating and achievable learning is. Imagine how easy school would be for a mature adult! I wish being held back a couple of years was normal and encouraged, especially for boys. My life satisfaction would be way higher.
If I made be jaded here. I did extremely well in school(s). I was never a trouble maker and was a top athlete, student, and musician. I think education is supremely important. However, school didn't teach me any of what you mentioned. Because what life teaches is that unless you're Type A, you're not going anywhere. Life teaches that hard work and determination and doing the right things does not get you anywhere. Life teaches you that thinking differently is downright discouraged and rejected.
We mainly teach either boring material or outright lies about how life works. Because we try to put our education on rails, teaching students that that's how life and education works, that's why it's a complete failure. The fact that students in COVID didn't respond to remote lessons is partly because they weren't even engaged prior to COVID.
We don't teach students how to be present, to know nature, how to explore their thoughts and emotions, or how to collaborate.
> the real answer is: it does not, they are not wrong
The real real answer is that it probably does, but on a much longer timescale that we generally consider and it is really hard to explain why. Something like better math skills lead to better life outcomes. Maybe due to a better model of the world and sharper thinking, but I am just guessing.
I would tend to agree with your last (speculative) point. The breakdown lies in communicating this to students and ensuring that each student receives adequate support at their own pace and style of learning.
They very well may! Unfortunately, since our approach to teaching how to learn is flawed to the core, it results in peoples' ability to learn being compromised from the very beginning, requiring them to build their knowledge base and learning approach on shaky foundations.
The way to correct this is by imbuing students with the confidence and skills required to learn (according to their style of learning) correctly from the very beginning, so that they build on solid foundations instead.
There is evidence that a person’s ability to understand and succeed in algebra is mostly determined by whether or not they can do arithmetic with fractions. Number sense is important in my opinion. Relying always on the calculator or a CAS leaves students confused and befuddled. I see this all the time in calc classes that I teach. The CAS loving students just don’t understand as well.
> There is evidence that a person’s ability to understand and succeed in algebra is mostly determined by whether or not they can do arithmetic with fractions. Number sense is important in my opinion.
My opinion differs a lot here. I would not say that I have a good number sense (I guess that people who have to do "numeric calculations" or "back-of-the-envelope calculations" as daily part of their job have a much better number sense than me). On the other hand, I find it rather easy to learn really abstract algebraic concepts (think Grothendieck-style algebraic geometry or similarly abstract mathematical topics), which many people (most of them with much better number sense than me) tend to find insanely difficult.
The number sense I talk of is not being able to do numerical calculations easily or in your head but rather understanding how to operate with numbers and their different representations. A person who can understand algebraic geometry doesn’t have trouble understanding things like simplifying x + 5/3 x. People workout any number sense have a hard time with this. Knowing that 8/3 is just a different way of writing 1+5/3 is confusing to them.
Textbooks about "abstract nonsense" rarely require you to do such routine calculations/simplifications - they rather require you to be capable of making sense of definitions that are (at a first glance) insanely far removed from anything you have seen in your real life: I would rather liken it to taking strong, dangerous hallucinogenic drugs, and making sense of the world that you now see (which is something that only some people are capable of); by the way: I don't understand why hallucinogenic drugs are illegal, but textbooks about very abstract math are not. :-D
On the other hand, textbooks about, say, analysis and mathematical physics (both in a broader sense) - which can also be very complicated - have a tendency to demand a lot of (also long, tedious) "routine" calculations from the reader (often to do by his own). For these areas of mathematics your argument surely makes sense.
I studied commutative algebra in graduate school which is an adjacent subject to algebraic geometry. People capable of understanding Hartshorne have number sense.
Textbooks about particular areas, in particular specific topics in physics (including mathematical physics), teach me a lot about number sense (and let me feel that mine is not really good or perhaps badly trained). On the other hand, these very abstract topics feel like a quite different activity to me that is only barely related to number sense.
> People capable of understanding Hartshorne have number sense.
This can also be explained by the hypothesis that people with a strong number sense love to feel themselves challenged - thus they attempt to understand this nontrivial textbook (even though understanding it may in particular require different skills).
> There is evidence that a person’s ability to understand and succeed in algebra is mostly determined by whether or not they can do arithmetic with fractions.
Evidence that it's causative? That would be utterly bizarre and I'd love to see a citation, because doing algebra has nothing to do with fractions. I'd think it's far more likely that there's a strong correlation between the two because they're both determined by the ability to understand and follow the rules of an abstraction/notation system, and if you taught people algebra first and then fractions afterwards you'd say that ability to understand fractions was determined by whether they could do algebra.
Whether it is causative or not it is still the case that someone who doesn’t know fractions will have a hard time in algebra. It would be bizarre to teach someone how to add rational functions before they can add fractions.
> Whether it is causative or not it is still the case that someone who doesn’t know fractions will have a hard time in algebra.
Doubt. Do you have any evidence at all for this claim?
> It would be bizarre to teach someone how to add rational functions before they can add fractions.
Sure, rational functions obviously sit at the intersection of algebra and fractions and require both. But they're hardly some deep foundational piece of algebra; I'm not sure my classes even covered them.
Only anecdotal evidence. I’ve taught beginning algebra courses at a community college for 23 years. Students who don’t know fractions have a very hard time in algebra. Those who can’t understand that x + 5/3 x is 8/3 x have a hard time understanding that 2xy+ay is (2x + a)y.
Understanding rational functions helps to understand what vertical asymptotes are and as such are a fundamental source of examples when learning limits. They also aid in understanding why tan(x) has vertical asymptotes where cos is 0. Every complete algebra curriculum includes rational functions. I say complete because algebra is usually broken up into 3 courses (2 at the pre-college level).
> Those who can’t understand that x + 5/3 x is 8/3 x have a hard time understanding that 2xy+ay is (2x + a)y.
Sure - but that's just as true in reverse.
> Understanding rational functions helps to understand what vertical asymptotes are and as such are a fundamental source of examples when learning limits. They also aid in understanding why tan(x) has vertical asymptotes where cos is 0. Every complete algebra curriculum includes rational functions.
Meh. x^-1 is a good example of some things, sure, but I don't remember ever doing addition of rational functions which is what you originally talked about, and I went through an extremely reputable maths degree.
You learned about rational functions in high school or middle school (most likely given your use of “maths”). I can tell you have very little experience with teaching. Most students who know that x + 5/3 x is 8/3 x have trouble, initially, with understanding that 2xy+ay is (2x + a)y. There is a reason for the order in which topics are taught.
> You learned about rational functions in high school or middle school
No middle school, and I very much doubt it. Searching I can see them mentioned in a further maths GCSE (which is something most schools including the one I went to don't offer, and rather suggests they're not on the regular maths GCSE, which would match my memory).
> Most students who know that x + 5/3 x is 8/3 x have trouble, initially, with understanding that 2xy+ay is (2x + a)y.
Who know that first or who have been taught it? I genuinely would like to see any actual evidence that the latter is objectively more difficult than the former.
I can tell you have very little experience with teaching. But surely your thoughts on the topic must be on par or superior to those with training and experience. My wife is a doctor and lots of people like to tell her how the body works and why she must be wrong. They think reading a blog post on vaccines is equivalent to 4 years of med school. The same phenomenon occurs in education. Lots of people think that since they went to school they know about teaching and how it should be done.
An introspective person would wonder why it is so obvious to others that they have no experience with teaching in the classroom based solely on their views of teaching.
There's no deep mystery to that; anyone who questions the dogma in any of the fields I mentioned is also obviously an outsider. The fact you went so quickly to attacking my credentials rather than giving any real rationale is not a sign that your field is full of legitimate knowledge; quite the opposite.
There are quite a number of experts who would disagree with your conclusion.
Upon reaching a certain threshold of technological dependence, the need for rational thought (which includes calculation) is tied to the need for food. The actual yield may be low based on other factors, but it is absolutely necessary for survival.
The alternative you suggest, is where technology no longer advances.
Logically then, population growth hits a malthusian trap, the old crowd out the young since they have the most influence, and then a depopulation occurs as the old naturally die off, and replacement births cannot sustain those dependent systems used to feed the masses.
You get a dragon-king event where everyone its a free for all over food and bare necessities, farming no longer becomes possible (because of looters), and the world order collapses to pre-agrarian levels, assuming the environment isn't destroyed in the chaos (i.e. MAD and Nuclear Fallout).
There are much better ways to calculate than are currently taught in schools, Trachtenberg System and Vedic Maths have worked well in many places.
Mental math has been around for quite some time, and the principles of math are all about finding uncommon knowledge or information that is not immediately apparent (though it becomes so via various mathematical transformations).
The current pedagogy of math is all about sieving and exclusion, and rote-authority based teaching, since it is a requirement for any specialized area of science (and is only taught in relation to mathematical concepts, instead of intuitive approaches). This is why they adopted a burn-the-bridge strategy right around trigonometry at the grade school level (intended to cause PTSD/suffering/torture), to safeguard against disruptive innovators at the source.
Algebra -> Geometry -> Trig
1 -> 2 -> 3
What do you suppose happens when the passing grading criteria in 1 is changed from just following the process (but not correct answer) to 2 (separate unrelated material which is passed) to 3 (correct process and correct answer).
If they fail Trig, and the problems are from Algebra (not something a teacher paid bupkiss will bother to look at), how do they go back if they passed Geometry? The students not knowing why they are failing are simply told, well you maybe you are just not good at math and should consider other paths if you can't do it.
This structure is called burning the bridge because it makes it so you can't go back from a progression standpoint. Ironically, this structure was adopted at the request of representatives from the National Teachers Union in the late 80s/90s, and largely remains the same today.
There are several other progression sieves embedded in academia intended to make it almost impossible for us as a society to develop a large number of creative people who reach einstein-level achievements in math and science (outside-self study, or specific environments/private schools).
This broad push largely started in the 1970s in publishing, and expanded from there.
I imagine a world where a 19 year old takes a few courses in first aid, child psychology basics, and now they're a licensed "class supervisor". They aren't university educated but the AI is what offers personalized learning and expertise to the students.
Most teachers today aren't experts anyway, we just pretend they are. So I'm not sure "replaced by AI" is the right way to frame the conversation. Instead, it may change education.
Far better than many classrooms in North America today tho. Folks who disagree simply don't know anything about the huge waste of time that many kids are put through everyday.
In anything. Like math, or math education. I was a teacher for years and studied education and I've seen some shit. The acceptance criteria for education degrees is often the lowest of any field in colleges/universities. The pay is extremely low. Great teachers exist, but often teaching is just a backup career for non-experts that don't know what else to do.
Outside North America teachers are sometimes respected and paid as professionals like a doctor or lawyer. Here they're more likely the butt end of a joke. You don't need to be an expert in anything to be a teacher in NA, generally.
this place is filled with people who are motivated to learn for themselves which creates a huge sampling bias.
You will see this come up in all sorts of discussion and i find it enlightening as to how exactly the decisions behind modern software are made.
Too many here fail to realise that real life has all sorts of edge cases and exceptions, including bad teachers.
Claiming that most teachers aren't experts is just another example of this. One student learns more about one narrow topic and then dismisses the teacher's broader, but shallower knowledge as being that of a non-expert.
Typical of the general population, myself included.
> One student learns more about one narrow topic and then dismisses the teacher's broader, but shallower knowledge as being that of a non-expert.
Perhaps you're talking about me (I said teachers are generally not experts in anything). I was a teacher for a few years and got a master's degree in education. That doesn't mean I'm right but I don't think it's smart to dismiss people on the internet as "probably just some dismissive student with narrow expert knowledge". I don't think I'm "general population" on this subject.
Teaching in North America has low professional requirements and even lower pay. The profession gets almost no respect. As a result, experts in anything (including experts in education) are pushed out of the career because it's a shitshow.
Yes. Teachers in North America are not respected or paid as professionals. The typical university acceptance criteria to become a teacher is very low. The pay is rock bottom.
Great teachers exist too but generally teachers are not experts in anything, including teaching.
I think online courses and AI education need the kind of supervision you mentioned. But they should also be able to give career advice, not just watch the room and push students to focus.
I'm not talking about the perfect classroom. I'm talking about something that could realistically be better than the ludicrously underfunded garbage we have today.
My urban public school basically spent as much per year for me as it cost for me to go to Harvard. I am not convinced of this ‘underfunded’ as root of problem thesis.
Most highschool / gradeschool is being forced to sit in a chair being baby sat until 3PM each day, with no opportunity to select goals, and act towards them. My daughter transitioned to a Montessori jr high, and she went from enduring school to actively engaging in self-directed learning.
Weirdly enough I was not very curious in my schooling years, barely getting through classes. As I have grown up, I have so much more curiosity about the world and my willingness to actually learn has skyrocketed. I feel like this could be a great space for adults who are seeking to do the same. I always thought calculus would be daunting to learn(and I still do), but with AI tools I feel like I can approach it with a different mindset.
> I see AI being more of a teaching assistant, rather than a replacement for teachers
That is exactly what he says in the tweet.
I think the problem with traditional teaching, as in any skilled profession, is often in short supply and underpaid, not happy, and unable to keep up with 25+ kids in a class. The world needs orders of magnitude more teachers that are highly competent and more easily accessible.
AI could massively scale high quality teaching with still a teacher in the loop.
Most people that are academically inclined are self motivated and have a desire to learn more.
Most people aren't academically inclined so it follows that most people aren't academically self motivated. Therefore among those that are academically inclined it is important to provide them with all the tools necessary because they're the ones that will most likely excel in an academic environment.
It is odd that the curriculum tends to accommodate people that aren't academically inclined at the expense of those that actually want to learn.
People that aren't academically inclined should not be forced to learn, or at least forced only insofar as they're baseline literacy so that they function in today's world.
> Most people that are academically inclined are self motivated and have a desire to learn more.
Isn't that by definition? Most xes are x.
> Most people aren't academically inclined
Is that so?
> It is odd that the curriculum tends to accommodate people that aren't academically inclined at the expense of those that actually want to learn.
Well, if what you say is true, isn't it fair that the program is catered to the majority, who are apparently not academically inclined? For one size fits all mass education, catering to the largest mass is the best you can do.
> People that aren't academically inclined should not be forced to learn, or at least forced only insofar as they're baseline literacy so that they function in today's world.
Isn't that what the curriculum already accommodates then? Didn't you just say that?
>Isn't that what the curriculum already accommodates then? Didn't you just say that?
No. The current curriculum penalizes people that are academically incline. Fast track programs are difficult to access for example.
>For one size fits all mass education, catering to the largest mass is the best you can do.
Yes. But we now have other options.
> Most people aren't academically inclined
>>Is that so?
As OP pointed out most people need someone to guide them and give them directions. This is because a lot of kids are not interested in learning and do "bugger all" without supervision.
The kind of self-directed learning only benefits people that are already academically motivated.
>Isn't that by definition? Most xes are x.
Yes. This was in contrast to what OP was saying, which is "The pandemic showed us exactly what children would prefer to do, when they don't have a physical teacher standing over them, which is bugger all."
This isn't true for students that are academically inclined. Only true for those that aren't academically inclined.
It's not odd to me at all. The most "academically inclined" (although I don't think that's just one type of person) are people who have the ability to help themselves with very little advice from others. We shouldn't be going out of our way to provide anything for them; we should provide all levels of materials for everyone. It's the stupid people who need to be coaxed and trained to use them, whereas for the smart people, it's enough to make them available and give them advice when they ask.
Teachers like gifted kids because they'll be successful no matter what they do, and the teachers can test out all of their dingbat social and pedagogical theories with no consequences. They can start with elite kids, finish with elite kids, yet somehow take the credit. Not impressed. Make dumb kids smart, then I'm impressed. You might even be holding back the smart kids, but they're probably smart enough to see through you and do well anyway.
That being said, there are some people who are motivated to learn entirely by the desire to impress teachers and other authority figures. They need attention to develop. However, I do not think that most people are like this, and I honestly think those people should be in therapy.
Looking outside at places like China, society unfortunately does require more experts in specific domain fields than there are self-motivated workers.
You can't run a billion dollar chip industry on passion alone, you are going to pull in people who may be working for external reasons. What matters then I'd that the education they receive is effective regardless.
> I think teaching is one of the few roles that can't be replaced by AI
So far, AI can't replace good teachers. But there aren't that many good teachers. In my experience, GPT4 is better at explaining advanced concepts than 70% of college professors. Unfortunately, education is often oriented around this horrifyingly archaic method of instruction, which prevents people from imagining what an AI oriented system could look like.
I remember when the future was MOOCs. Let's get the top 30% (or 10% or whatever) of teachers to record high-quality videos, then everyone can have a top education. Even the rest of the professors might learn something!
AI based education might or might not be "MOOCs 2.0". Even for the less good teachers, having a real human in the room is one of the features that lots of people appear to be ready to pay lots of college fees for.
I really agree. And I think it's likely your detractors have not stepped foot in a classroom lately.
The issue is not engaging teachers. The teachers we have here in BC are excellent and love their subjects (my wife and many of my friends are teachers). The issue is behaviour, which has deteriorated significantly since COVID, though the changes have many other contributors.
Try asking an AI to engage with 30 kids who are on their phones with earbuds in. You absolutely need a human as a teacher.
That said, AI teaching could be a great teaching assistant.
A large amount of engagement is reading and responding to body language. There are also no social ramifications of ignoring your laptop.
Don't get me wrong, I think AI has a role in the future classroom, but that should be lead by professional educators used to dealing with children.
There is also the social side to education that goes beyond course content. Teachers are not just there to dole out information, but to act as role models and part time parents.
I don’t think children are the initial target of this company, but I get what you are saying.
The type of person who’s going to sign up for a course from this company are probably already autodidacts to some degree.
If I were teaching sixth grade mathematics, I wouldn’t be too worried yet. If I were running one of the many mathematics academies that have popped up throughout a lot of more affluent ‘burbs, I’d be very worried.
Yes, it looks like this project is starting with helping highly motivated adult learners go deep into a hard to teach/learn material. Contrast this with the Khan Academy approach at https://www.khanmigo.ai/ targeting young students and their teachers and parents with broad assistance across subjects. Maybe they converge?
> There are also no social ramifications of ignoring your laptop.
you could absolutely have a digital social credit system the way you have game scores and leaderboards. once you get a competitive system like that going, it would sustain itself. top students could get to visit cool stuff like grown up labs and get involved in museums, etc. bottom ones could be celebrated with a virtual dunce hat on their avatars.
the problem is how mediocrity is now valued over hard work.
> A large amount of engagement is reading and responding to body language.
I don't follow this assertion – it's possible to be engaged by something that doesn't even have a body. For example: the things currently engaging them in this scenario – their phones (or whatever's on them).
Engagement being a two way thing between the teacher and student. In this case I was referring to the teacher reading and responding to the student's body language.
> A large amount of engagement is reading and responding to body language.
That's due to a limitation of the current medium, don't you think? When I started going to school, I had to improve my social awareness not because it's an inherently, objectively better way to learn, but because that's what was needed as a result of how the classroom is structured.
> But social awareness is part of adult life, and school is the place we prepare our children for adult life
The school doesn't prepare kids for this. By most measures it does a rather poor job. There's a reason they say "A" grade students work for "B" grade managers who work in companies started by "C" grade students.
Other commenters have said it, but the social behavior kids are exposed to in schools doesn't match much with the "real" world. The way problematic people are handled is quite different. As are the metrics of what constitutes success.
I was walking to my classroom last Thursday, and a kid pushed another kid down the stairs, right into me. I went ballistic, and sorted it all out, but there is no way an online AI tutor can deal with that kind of behavior. So if you want social education, you need physically present teachers. If you want online education, then parents are going to have their work cut out.
It's not. The discussion is about behaviour control, which is part of a teacher's job, and not going to be provided through consistent support or multiple learning approaches. It's not about academic learning at all, in fact.
The only thing that this proves, is that asking google gemini to teach a kid socialization is hopeless. Kids dont just read cheesey platitudes and change their behavior, its something that requires constant and sustained teaching.
It is actually a good thing for kids to know that if they abuse somebody, then someone else, who is bigger and stronger will be very angry with them. That's how the adult world works. Or at least supposed to work, when police and other "bigger and stronger" men do their job properly.
We use teachers as enforcers. But in my experience going to a big urban public school, the best enforcers are often completely orthogonal to the best knowledge-conveyers and I think we could have specialization for each of these roles.
In my opinion, the best enforcers generally are charismatic yet firm and come from a similar community/background to that of the students. The best teachers have an infectious passion for their subject, but oftentimes that trades off with their ability to enforce.
Yes, a teacher's role is as much behaviour management as it is knowledge transfer. I think most teachers would love to shift the focus to the latter, but it seems to be shifting more to the former of late. Knowledge transfer is what teachers are actually trained to do, with very little of professional pedagogy training about behaviour management (at least, anecdotally from my teacher wife).
I don't know what the future of education looks like, but it sounds like there are significant behaviour problems in the classroom at the moment, with many teachers quitting or retiring early as a result of not being able to do what they are trained to do (teach).
> The pandemic showed us exactly what children would prefer to do, when they don't have a physical teacher standing over them, which is bugger all.
This is not true. The pandemic showed us exactly what children who are accustomed to being force-fed information and whose natural learning mechanisms and curiosity have been suppressed in favor of a generalized one-size-fits-all approach do when suddenly removed from the only learning paradigm they've ever been exposed to.
My kids (not yet old enough for school) are extremely self-motivated to learn and explore the world around them. So am I, and that never went away over the course of a full homeschool education.
Totally incorrect. The vast majority of the population are relying on schools and teachers to potty train, teach manners, instill excitement for learning and basically do everything a parent should be doing. Large number of kids have no real parent figure and thats from all types of backgrounds. We are not talking about kids who have strong households where learning and general manners are being taught.
While different teaching styles have their pros and cons, and a more hands-on approach might yield better results, the problem over the past 5-10 years is that many children lack parental support. Regardless of the pedagogy style chosen, without parents providing a strong foundation, meaningful progress in the classroom is unlikely.
Your kids clearly have a present parent who is engaged, it is a stark difference to many other kids.
It seems like that’s a societal problem that should be tackled at a societal level, rather than a fundamental failing of students.
It’s not something schools can fix, regardless of the style of pedagogy chosen; so why not improve school in its area of focus, and introduce solutions to students’ home life at the same time?
Easier said than done, I know, but focusing on the wrong area will get even less done.
>The pandemic showed us exactly what children would prefer to do, when they don't have a physical teacher standing over them, which is bugger all.
Amend to: The pandemic showed us exactly what children's parents would prefer to do, when they don't have a physical teacher standing over their children, which is bugger all.
I also noticed the material of an entire day can be learned or made in a few hours. So indeed I also realized it's mainly daycare with a bit of, or slow learning.
I worked on online learning for a bit. Turns out people are willing to pay for the inconvenience of in-person learning, even flying to another location. It's the only way most people can focus on a topic. Otherwise, work, kids, life interrupts and they can't stay on track. Replit's 100 Days of Python says only 0.4% of those who complete day 1 finish day 100.
yeah. When trying to learn, say algebra and getting stuck on a problem, what's better for learning? staring at the problem until you get bored and wander off, looking at the back of the book for the answer and then maybe going back to figure out why, or individual instruction where you're able to ask someone who knows that they're doing about why you're stuck, and have them give you hints until you get unstuck, and then give you another, similar, problem
for you to work through?
> We send kids to school, in the hope they get some education, but the reality is that we use schools for free childcare while we work
We also send kids to school to learn social skills they can't learn by themselves.
My kids sometimes watch science shows (on TV as well as online) and tell me all kinds of fascinating facts about black holes and the human immune system and {insert_huge_list_of_stuff_I_don't_fully_understand}. That's the easy bit.
"Getting along with other people" isn't something you learn ... by yourself.
If the technology is truly as capable as humans in many domains (and that might still take a while), it will not matter anymore whether it is a good teacher or not. The need for (and thus value of) human labor will depreciate and so will its "supply chain" the education sector.
> hope they get some education, but the reality is that we use schools for free childcare
Exactly, teachers will be less and less pedagogs and more and more wardens.
But how to harness "Bugger all" so that it results in educated students? Because my understanding is everyone likes to do stuff.. no one really does nothing, but often unproductive/consumptive things if not channelled.
i think kids would be self-motivated with the right system.
I got a lot more motivated to learn when i learned programming.
during the pandemic, the world was in shock, so of course kids are going to play video games when their parents are anxious and filled with cabin-feever.
kids will choose many different kinds of activities at any given time.a lot of kids really don’t like games, some do, some don’t.
i’m not trying to be pedantic, but anytime someone implies a human, particularly a kid will be at all predictable shows an incredible lack of understanding of people. the vast array of moods, time of day, quality of sleep the night before, are they hungry, mood of the parents when they drove them to school, how did their school/work day go, how was their social day, and on and on and on.
again, apologies, i’m not trying to be pedantic but i think in this particular topic it reeeeaalllly matters.
My broad sweeping generalization was primarily meant as a counterpoint to this from parent comment: "The pandemic showed us exactly what children would prefer to do, when they don't have a physical teacher standing over them, which is bugger all."
My point is more that kids, when left to their own devices (with basic needs met), will find ways to occupy themselves that they find interesting that are not outcome oriented (I call this playing).
And I personally have never met a kid that didn't like playing in some form or another, though the form of playing is highly, highly individualized.
Disagree slightly. I think AI can be used to generate average quality course material, which may be useful to below average teachers, or good teachers thrown into a subject they haven’t taught yet.
Obviously someone like Andrej will totally crush it.
Even as a self-motivated learner I fail to see the bigger impact of AI. For a “virtual Feynman” I would prefer the online video courses and books which exist without AI. The best I expect an AI to do is to answer my questions and confirm my understandings. At AI’s current state I can use it as a better search engine but due to hallucinations I can’t expect reliable answers yet.
Asking it “3.11和3.8哪个大” (meaning “Which one is larger, 3.11 or 3.8?” in Chinese) and it answers 3.11 more than half of the time. I assume it’s because Python 3.11 is larger than Python 3.8. While it does work in its native language English, this failure doesn’t give me much confidence in its reliability, as we don’t know why it works in one language but not the other yet.
There is an interesting physics education experiment. A random group of students are shown a lecture on a topic and take a quiz after watching the video. The students rate the lecture. Repeat with a different lecture on the same topic. The students did worse with the higher rated lecture.
There’s teaching students like. There’s teaching where students learn. Sometimes the two intersect. Will an AI education company optimize one that students enjoy or one where they learn better?
What is the baseline performance of the LLM in solving those programming tasks? And did you test the performance of the students in the Codex group at the end of the course without allowing them to use Codex? Essentially I'm asking how can you conclude that these students didn't just learn to call a LLM, but actually learned to code independently?
I don't think this is true at all. People failed to learn during covid because the technology is bad. I don't think most people are motivated very much at all by the disappointment of some stranger standing over them. I don't even see it as a desirable aspect of someone's personality that they can be extrinsically motivated by the approval of strangers.
What a teacher provides is a sometimes customized, sometimes flexible schedule, that (sometimes) pays individual attention to what aspects of a concept a student is falling behind at, and (sometimes) comes up with personal recommendations and alternative approaches to break down a student's involuntary resistance to a concept. This might be doable with A.I.. It's not doable with actual teaching anymore because class sizes are too large. A.I. will be cheaper.
And I'm not saying that teaching is so simple that A.I. can do it, I'm saying that teaching is so complicated that it might be that only A.I. is sufficient to largely replace it. I think that what I'm arguing against is that the idea that teachers could be replaced by glowering scarecrows, or fur-covered wire armatures like they once used in experiments to replace animals' mothers.
I don't think that teachers make as good parents as parents do teachers. I don't think most people are mostly motivated by the approval or judgement of their teachers.
What people need is constant, helpful, personalized guidance, and that is very expensive to get from employees.
Teaching is not the end goal of education though, the educated student is. Or so I was taught.
Part of the reason why teaching is considered noble is because it is an act of assured replacement, inspiring not dependency imparting skills of self-motivation and will power.
Completely disagree. ChatGPT has taught me more than I could ever learn from any lecture and I have a doctorate. A moderately motivated student will do wonders with AI.
For instance, I’ve had trouble understanding exactly how heat pumps worked. Sure I knew the basic concepts of condensation and evaporation but not the nuances of pressures and boiling points at various stages. I asked chatGPT to explain it to me from the perspective of the refrigerant. It started with “I am R-134a, a refrigerant just leaving the evaporator…”, and proceeded to give me the most thorough understanding of heat pumps I could imagine, complete with working pressures, boiling points, pressure differentials at the escape valve etc. Follow up questions led me down interesting paths where it came up with a brilliant comparison to quantify the greenhouse potential of the refrigerant R22 ie 1 pound of R22 has the same greenhouse potential as a human being breathing for 787 days in a row.
This is exactly the problem we have found in our research on generative AI for education [1]. We ran a pilot in a large high school in collaboration with math teachers, and found that students basically copy answers from ChatGPT, resulting in worse performance compared to students not given ChatGPT. If students don't want to learn, ChatGPT isn't going to fix anything.
You are just giving them ChatGPT with a bit of prompt engineering, and evaluating them on math problems, which we know LLMs make errors on because they are not calculators. You aren't putting in the effort needed to build a real tutor and learning assistant. I would not extrapolate from these results
There are also a lot of things that can come in before you build a full on tutor. One example is being able to tailor word problems (transform the nouns) to subjects interesting to the particular student. They could also be used to help understand where students are struggling. We are still at the early phases of useful AI, optimism is more appreciated, especially as contemporary times have become so pessimistic
> Most learners, the world over, are not self-motivated
This is the exact opposite conclusion and methodology of Maria Montessori (and her schools with the same name). Children are naturally curious and want to learn, but they may not want to use a poor education system designed to mark grades in a hyper specific focus.
Is the reason because montessori schools are often high-priced? Because that's definitely true (and you can easily say it's a privilege to attend one).
That doesn't, however, mean the methodologies don't work or don't apply. You can study the methods in any of her ~20ish books, or the more modernized recaps of them.
The teacher/tutor certainly won't be replaced by an AI. But those we call 'teachers' nowadays - possibly even for decades - are struggling glorified janitors because society treats schools as daycare. It's understandable that 9-6 parents need daycare. But why do we keep confusing daycaring with teaching? And why do we expect teachers to be perfect at both roles? And why all 30 students in the class should focus exactly on the same aspect of education simultaneously? That's the main issue with 'teaching'.
> The teacher still designs the course materials, but they are supported, leveraged and scaled with an AI Teaching Assistant who is optimized to help guide the students through them.
I assume that it would be an agent that the teacher prompts in a loop to build and refine the course material. Probably with an upload so if they want to bring the full thing ready to go they can and in that case the data just needs to be formatted by the LLM for their database, but also can have the AI fill things in. And then check what came out and refine it.
It sounds like Andrej's ambitions stretch beyond AI/LLM courses too - but they're a natural "first course" starting point because that's where his own teaching expertise is focused.
It is. Though they are also doing some really smart and thoughtful stuff with LLMs to power courses and learning environments (https://www.khanmigo.ai/ is a part of it).
that's a tokenization issue. every tool has strengths and weaknesses. why does it matter whether an LLM can compare numbers? that can be done trivially in any programming language
As an AI layman (downloaded Claude for android as a result of hn, just today) "why does it matter whether an LLM can compare numbers?", is rather important to me.
Probably others, also.
It doesn't seem super relevant to karpathy announcing he's created a company so that he can increase the production value of his AI YouTube videos
I mean, sure, the company is ostensibly going to also teach math at some point, but karpathy will not be using gpt 4o for that when it launches (what do you think his timelines are? Do you think he is going to be able to solve trivial things like "having the llm use something like function calling to do math"? If you're unfamiliar with his work, karpathy is a very good engineer, and this is a small problem that anybody working in building production apps on LLMs can easily deal with)
i can understand that perspective. as an end user, you would like your application to handle math questions correctly. it's true that llms are not the best at math
as an application developer, if we need an llm to be good at math, one solution is to give it access to a python interpret
I was going to say beware because there isn't an Anthropic Claude official app.... but I checked and I guess as of today there is one hah. https://www.anthropic.com/news/android-app
If I'm going to augment my education with AI, I'd at least want to know it could get basic numerical facts right. If a computer program struggles with the concept of a number being greater than another number, how do I have any confidence that it can teach physics?
your concern would be valid if an llm were the ONLY tool being used. applications use multiple tools so you can use the appropriate tool for the job. if you're doing math, you don't want a standalone llm
As a student, how would I know what things the LLM tutor can provide correct answers for, and what things I will need to "use appropriate tools" for? Should I rely on the LLM to help teach me spelling, or US history, or are there more appropriate tools for these, too?
if the product is great, hopefully you as the user would not have to worry about which tool is doing which task. the developers would worry about that. it's the same in any app. the user doesn't know or care what tool is used to render the frontend or store the data in the backend
Since we're all just speculating to the wind here, I can see multiple ways LLMs can be used. Maybe it'll help simplify TA triage, maybe it'll just be a Discord bot. Maybe a classifier will sample from multiple models.
I think if anyone can give this idea a fair shot it's Andrew Karpathy, an ML expert and a person known to be passionate about education.
applications can use multiple tools. an llm is just one tool, and it will not cover every use case, such as math. that does not detract from the utility of llms
This is the same issue I have with large language model as coding assistants, since you're effectively not in the driver seat - you're acting more like a code reviewer, and I think that that passivity eventually causes critical observational skills to atrophy.
The rest of the AI art in the course isn't any better. The thing is, it doesn't have to be like that. I do AI art and I follow a lot of AI artists, and you can fix all those little weird mistakes it makes.
The thing is, when the AI art generator makes a mistake and draws a person with 3 arms, that is obvious to the student and they can take the output with a grain of salt.
But when the AI physics tutor generates some physics result that's the equivalent of a person with 3 arms, that will not be obvious to the student. They will take the words of the AI credulously. I see it all the time in programming as well, where the AI just invents APIs, semantics, and syntax.
Don't use an algorithm which produces its response according to a probabilistic arrangement of tokens when solutions require accuracy / correctness? Most probable and most accurate are not the same thing. Hoping that we can get the errors down to something acceptable using an algorithm that is fundamentally inappropriate to solving the problem seems like a fool's errand to me.
Do you know what happens when you take VC money? Probably not. I assume OP wants to succeed and retain control. VC money is about the worst choice of funding for something like this.
its super easy to start ai companies, its difficult to make them meaningful. The ai generated image in the twitter post has an asian woman with 3 arms and some of the most horrific AI face placements I've seen in a while - Is this the quality control we can expect our future AI education systems to embody?
I appreciate attempts to disrupt things, but education seems to be one of those verticals that seems to be allergic to disruptive technologies. Education seems like it can either be very specialized, or very generalized, but at the end of the day it should be egalitarian. If this approach to education works, would we be able to have every teacher in every school in America adopt it? I have to imagine the resources needed to train the teachers, distribute the technology, acclimate the parents, and then do this all on a scale such that no one is left out or treated better if you didn’t happen to go to an “AI” school makes for a tough hill to climb.
I think a lot of the real issues with solving problems in education is that they have trouble applying to the larger picture of compulsory education.
This is the fundamental problem with education: everyone treats it as some problem to "solve" with tools, and technologies. The issues with education are human ones -- interpersonal and policy -- not because it's lacking some tool or technique.
They just installed some state of the art AI-enabled, "smart" mega drawing screens w at my daughter's schools touting all the supposed immeasurable benefits it will bring, and most of the parents, including myself, just rolled our eyes.
If delivering a good education can't be achieved because there are parts of our education system that continually resist adaptation, then whatever those parts are they ought to reach their breaking point and be pushed beyond it.
Those parts have reached their breaking point - the teachers are ready to quit from lack of resources and support, and parents can't do their part because they have to work far too hard to keep a roof and food for their families. AI doesn't solve that.
But if teachers could use AI to grade exams and automate the boring work, then they could take a second job and be able to afford housing. Sounds like an easy win for AI.
I think that 20 years later it is much harder make effective satire, so I rarely do it now. More extreme positions are taken pretty regularly by folks online, so even people like you in my target audience are never sure if something is satirical or just extreme.
That’s why most satire is pretty lame. The more obvious it is, the less funny; the less obvious it is, the less interesting. And the whole thing is built on an inside/outside dynamic that is pretty basic.
Realising the promise of AI, freeing teachers from the drudgery of 'boring work' so they can take on a second job as a food-server, burger-flipper, shelf-stacker or Uber driver (at least until robotics and self-driving tech eliminate those jobs too).
I got an excellent education from the public school system and had passionate, committed teachers throughout, and was well prepared to pursue an engineering degree at a top university. The solution to education is not to try and scale up some absurd and ineffective AI system that is worse than teachers, it's to pay teachers an actual proportionate salary that is in line with their impact on our society so we can retain good people. Just because people like Karpathy understand AI doesn't mean they understand education.
The incentives just are not properly aligned. I went to an urban school system that paid many teachers quite well (upwards of $130k 10 years ago for many), but the pay was just totally ubcorrelated with teacher performance.
Many of the worst teachers were the best paid because they had seniority and that was how the union had structured pay agreements, whereas many of the younger better teachers were paid very poorly (around $50k in a major metro) despite having left very high paying professional jobs to give back. The problem is that there is little-to-no economic incentive to do anything to improve student performance.
We shouldn't have to explain that it's plainly obvious that society massively benefits from an educated populace, especially one that claims to be a democracy. There are incredibly broad benefits from a reduction in crime to the expansion of the skilled labor pool.
I've noticed that many people in tech seem to disregard or disrespect educational institutions. So I'll turn it back on you. What draconian reason could you possibly have to make the argument that we shouldn't try to give every child an equal opportunity for a high quality education? Do you hate living in an educated society that much? Are you interested in living in a malthusian nightmare?
>Are you interested in living in a malthusian nightmare?
Some of them live in such an environment already. I don't know if it's a hallucination or not, but judging from what I've read here over the years a lot of tech people seem to live in the most cutthroat of environments and see everyone as competition to be eliminated or obstacles to be cleared. Some of them live in an environment where you can rely only on yourself, requesting help is seen as weak victim-like behavior, but giving out for free is worse - detrimental because that other person might see what you're doing and take credit for or steal your work; some say helping another with your skills/time and not charging money is peak cuck behavior, and some of the more organized (I'd like to say 'coeficient-driven') members of our community really believe that money is the greatest measurement tool ever invented and we should measure everything with it, including a person's worth.
That being said, generalizing is bad and there really are some truly golden individuals here who have done humanity a net benefit while charging nothing for their work. Like Fabien Sanglard, for example (you likely will never read this but I take my hat off to you and I hope if you get the chance, you should clone yourself in the future - humanity could use at least 10 of you).
I'd find quotes for all of these but I don't think I need them, you've seen these messages if you read the HN comments enough.
Edit: In the 10 or so minutes I took to write my comment, yours went from all black to almost unreadable. It shows better than any treatise would on the opinion HN denizens have on 'free' or 'equal' anything.
What draconian reason could you possibly have to make the argument that we shouldn't try to give every child an equal opportunity for a high quality education?
Opportunity, sure. But an opportunity to get a good education is not the same as actually getting good education. Because every child is different and we want to spend a lot more efforts to educate a child who shows genius level potential than on a child who has zero interest in anything. This is assuming limited educational resources - an assumption challenged by AI education initiatives.
If I have two sons, and one is bright and curious and hard working, while another is dim and lazy, the first one will get 95% of my attention.
The biggest predictor of whether or not a child is "gifted" is the amount of attention they receive as children from their parents and from the education system. Actual genetically driven genius is rare. The kids that gifted child programs benefit are largely the children of affluent parents and not the so called geniuses of our generation. They mostly serve to create a two tier education system within public schools. I'm not saying we tolerate disruptive kids or that we shouldn't reward merit, but this kind of rich kids get the resources and attention system is counterproductive to the outcome you seem to want.
Furthermore, there is zero proof that AI will give us the kind of system that will allow us to shore up the limited education system. The actual solutions to many of these problems are things like paying teachers more to retain the best people, giving kids free lunch, funding after school programs and one on one tutoring etc etc etc.
But doing those things is hard, so tech bros who believe in the myth of the gifted child, who don't have any background in education at all, come in with these systems that they think are silver bullets, then are shocked when they don't work, blaming the unruly children of the proletariat on their failure to fix anything.
And what will you do with the lazy one? Kill him? Let him die when in problems? Maintain it throughout his live? Ignore him and let him live a miserable life?
> I'm glad we as grown adults are sitting here and judging a child on their dedication to academic rigor.
In a conversation about academic performance, what were you expecting?
I'm not being facetious, I'd really rather like to know: in a conversation about resources being poured into academic outcomes, why is a child's athletic ability, or artistic ability, etc relevant?
We are comparing outcomes of investment into academic performance - do you expect this conversation about ROI to be completely without judgements?
We're talking about changing the entire trajectory of someone's life when they're a child because they find school boring when they're 8 years old. Talking about ROI and hard numbers about this makes your look like a ghoul. I think aggregate measures over years are required to accurately measure the impact educational investment, but we should pay teachers more and hire more of them to reduce class size because we have a moral obligation to do so, not just because we'd get an ROI.
> We're talking about changing the entire trajectory of someone's life when they're a child because they find school boring when they're 8 years old.
Maybe you are. We are not. No one is suggesting decreasing the resources made available to a cohort of children. We're suggesting increasing the resources for gifted children within that cohort.
> Talking about ROI and hard numbers about this makes your look like a ghoul.
How else are you going to discuss what is obviously a very important investment into humanity? Personal attacks, maybe?
Is it better to be egalitarian in education or is it better to focus on raising the floor, or raise the median, or providing equal opportunity? They aren't the same thing, and it's very possible that the education system that focuses on one will miss out on parts of the others. You can make moral arguments in favoring each of those options (or even focusing on raising the average).
You can optimize for economic benefit, innovation, fairness, or passions. Their is plenty of non-draconian reasons for preferring each.
It's better to be egalitarian. The clientele of hackernews is predominantly people working in high paying industries, who were lucky enough to get the resources and opportunities to get there, in addition to a considerable amount of hard work. I pushed hard to get where I am today and personally benefitted from gifted child programs. What terrifies me is the idea that being born to slightly different parents or in a slightly different area could have had drastic effects on my outcome. I believe we should make policy decisions assuming any one of us could have been born to the poorest and most negligent parents imaginable. Making any other assumption is being dishonest about the benefits afforded to you by your upbringing.
Wouldn't it be better to raise the floor? If given the choice of a higher floor but a higher ceiling of education quality vs equal but lower quality for everyone, I would prefer the former.
The human species has only existed as such for ~100,000 years. Almost all human societies have failed.
We should spread knowledge far and wide for the same reason adaptive mutations spread through a population: shit happens.
Rocks fall out of the sky.
The Earth is jelly with a thin "crust" of congealed goo on top. So-called "solid ground" is thinner relatively than the paint on a globe. It shakes.
(As a kid I lived through Loma Prieta[1], I've seen the earth roll like Santa's belly. We are small!)
We should go through life like "an old man crossing a river in winter", and things both precious and free, like knowledge, should be the treasure of every person, no matter how poor or weak, for tomorrow they may be all that's left.
> Education seems like it can either be very specialized, or very generalized, but at the end of the day it should be egalitarian.
This is the approach I'm taking.
There are a few psychological and technological tricks here or there, but in my opinion the hardest part is getting the incentives right.
It's very easy to build an educational system that incentivizes for the wrong things. And even if we all agree what the "right" things are, something like universal education is a collective action problem with all the difficulties that come with that.
It is self-evident to me that this is not true in the U.S.
There is very little about education that is egalitarian. An inner city grade school in Michigan is in no way comparable to a grade school in the Palo Alto suburbs.
> Such that no one is left out or treated better if you didn’t happen to go to an “AI” school makes for a tough hill to climb.
To drive this point home, in the real world we do not hold this standard for air conditioning. Tens of thousands of schools in the U.S. do not currently have adequate HVAC systems. If we are not currently applying this standard to A/C, I don't know why we would choose to selectively apply it to AI.
Attempting to gate innovation behind complete social conformity and universal adoption doesn't strike me as a rationale stance.
Nit: you’re replying to what “should be” with what is. Those two concepts are not the same. Our property-tax-based educational funding system is widely seen as a tragedy.
An inner city grade school in Detroit gets 16,500+ per child a regular child in Michigan gets 9,500. Palo Alto spends 25,143 per child. New York spends $29,873 per child. The states that spend the least Utah, Idaho are around 9,000.
Part of the cost is localized like salary, utilities costs.
Detroit isn't necessarily being under funded compared to Palo suburb when you normalize regional cost differences. It does affect the quality of teachers one can attract but so does the location in general. Bumping up Detroit to Palo Alto's spending wouldn't have much effect on the education for the average student.
Palo Alto is home to many immigrant from many places who self select and place education as extremely important for the next generation. For the average Detroit kid education isn't cherished and valued over everything else. Until you can equalize those believes things won't change. You won't be able to buy a higher educational rate by painting the walls twice as fast or by hiring more educational assistants.
there’s pretty good evidence that outcomes are largely in correlated with per pupil spend. would take me a little while to find since i haven’t looked at it for a while
I went to an inner city public school that was better funded than your Palo Alto numbers once you account for federal funds.
I agree that funding is not the only problem, the money could be spent much much better (and teacher unions should unfortunately probably not exist). Another large problem is that I do not think we as a society have agreed on what the actual goal of public education should be.
I think a good place to start is “anyone in the world who wants to learn and is hindered by their school and teachers”. Not “this must work in every American school from the start”.
A smart kid with a phone in Kenya should be able to have a shot. Wikipedia allows this. Khan Academy helps. This in the extreme is Khan Academy but potentially for every course a kid (or adult) could want, generated on demand from scratch if need be.
The next optimal study session generated for your current state of knowledge, energy level, learning goals, by the teaching agent who is provably the best at delivering that lesson for you, using the optimal mediums for your context. Video at home or in a class, audio when on a bus, printouts if you want. Add in spaced repetition and a few fun quizzes, gamify it. Let each person fulfil their potential unhindered by the resources of their school, parents or country.
Even if this ends up being nothing more than "Khan Academy but I can ask questions and get answers and it's all in my local dialect of Urdu" it is a huge step forward for educations AS LEARNING.
To me, the good place to start is "what do teachers want that they feel would make their teaching more effective" rather than "the entire community of teachers is a barrier and we must work around them."
I'm curious to see how this all plays out across different disciplines. The process of learning calculus is different from learning a language, which is different from learning the history of science, which is different from learning ethnographic methods.
There are all kinds of self-interested reasons educators might resist some of these technologies. But this also seems to be one of the areas where people from the tech world impose some idea that could potentially work for their limited domains of expertise but don't work at all for others.
Whew! Yes. Education is allergic to disruption because education shouldn’t be treated like a business.
Educational goals are to educate people not make money. Education requires countless failures. Business models are composed to be failure averse or reduce risk.
Education has and will likely continue to be a money and time sink because it is an area that requires constant failure from its participants to grow in knowledge.
One thing I've not understood about this is how do you create an AI course to teach people things... without creating an AI that can DO the very thing that will make that very same knowledge obsolete?
For example, how do you create an AI language teacher without creating an AI that can make learning languages obsolete? If you've got an AI that can, in real time, hear other languages and translate them (as you might have for an AI language teacher), then why would a human need to spend countless hours learning this other language? Just hold your phone up and let AI do the work for a fraction of the effort.
For a harder, non-solved problem, consider math. For an AI to do math will require something unknown at this point, if it can ever happen. But assuming it does, why would we want a human ever to "do math" ever again when we have the AI that can teach it just do it for us? The AI will almost certainly do it more cheaply and with more skill than a human IF it can be done at all.
It's this sobering realization that I struggle with. If someone can tell me where I'm wrong I'd be greatly pleased
In the language example, people still want to learn to read/write other languages despite many translators being available. The tech to teach might be less sensitive to latency than the skill in humans, or it might be very expensive, or it might be useful to non-tech-literate people.
Well, for instance, because hearing other languages and translating them is not even the primary use-case for knowing a language. It would take a suite of purpose-built AIs, and the knowledge and ability to use them in-situ, to replace knowing a language.
AIs can also teach us to do things they cannot themselves do, for instance you could have a driving-test tutor. It could teach you a lot of things, despite us not having full self-driving AIs.
Well I've never learned a language in order to be the best at that language, or the only person who can speak it.
If you only want some basic communication, sure use a translation app, but learning a language is also about learning a culture. Learning about new music and literature and poetry that you'd never otherwise get exposure to.
It's like asking what's the point of learning a musical instrument when I'll never be a great musician.
It's to benefit myself so what would be the point of just getting a translation. They're never the same.
Using a phone translator is fine for being a tourist or maybe for short business trips. It's super inconvenient if you want to live and work somewhere where they speak that language, though.
You need to have your phone on you all the time, otherwise you can't communicate. There is always going to be at least one clause worth of latency in each direction of translation due to differences in word order and semantics, so you'll be at least a sentence or two behind the conversation - whereas if you learn the language, you're interacting in real time. Also do not underestimate the goodwill that comes from being willing to learn a language - people recognize that it's hard, they're going to be much warmer to someone who spent over a thousand hours to reach B2 in their second language vs. someone who downloaded an app instead of putting the effort in.
Near real time speech-to-speech machine translation is super cool if you're a tourist visiting a country for a couple weeks, or an employee visiting a factory in a country you don't speak the language. It isn't a replacement for learning a language, though.
Some subjects are necessary building blocks to more sophisticated tasks. The AI can teach the building blocks and leave the task of building to us.
Coding, for example, is just a small subset of the tasks necessary for software engineering, which in turn is a small subset of the tasks necessary for SaaS company.
We can break down human endeavors into subjects that are "AI teachable" and without ever needing the AI to be able to use them in a sophisticated fashion.
I generally agree, in the long term there will be far less need for teachers, as there will be less need for human jobs to be taught.
However, creating many individual systems which are better than a human in a specific area each is very different from one unified and integrated system which combines all of them. The latter will take a lot longer to achieve. Until then there is value in teaching humans.
I think you're imagining something like chess. In order for an ai chessbot to teach me chess as efficiently as possible, that bot is probably superhuman at chess.
That's not necessarily the case. A good tutor can be more "curator" than "creator of course content"
I can learn math by reading existing math textbooks. Imagine having an ai that is able to judge my knowledge of that math and assign reading or problem sets accordingly. Imagine an AI that is compelling to talk to and keeps me on task given my stated learning goals.
None of that requires superhuman anything.
Would I still want to learn linear algebra even if the frontier of math is being advanced by some super intelligence? Sure, why not? Isn't the frontier already being advanced by many people smarter than I am?
> ... in the case of physics one could imagine working through very high quality course materials together with Feynman ... with recent progress in generative AI, this learning experience feels tractable.
Actually, this seems to be absurdly beyond any of the recent progress in generative AI. This sounds like the kind of thing people say when their only deep knowledge is in the field of AI engineering.
It's optimistic, but given the OP is one of the best-informed technical generative AI researchers, and has been passing on that knowledge to the rest of us for a decade +, I don't think we can just dismiss it as unfounded hype :)
My point is that he's a world expert on the engineering of AI systems. That shouldn't be mistaken for expertise, or even general knowledge, about anything else.
It's a good principle to bear in mind for people from any profession, but top AI engineers in particular seem to have an unusually significant habit of not being able to recognize where their expertise ends and expertise from another field (such as, say, education) begins. They also seem very prone to unfounded hype - which isn't to say they're not also good researchers.
Maybe Karpathy happens to be better on this than his peers, I wouldn't know.
I mean, the guy isn't saying that it's going to 100% happen. He's saying that the problem feels like it might be doable at all. As Andrej has a background in physics, the phrase of 'feels tractable' would mean that he thinks that a path might exist, possibly, but only a lot of work will reveal that.
This seems rather generous given that he was just a physics major. There's lots of physics majors who understand very little about physics and, crucially, nothing about physics education.
It's legitimate to call it a background in physics, but given the particular level of background and the context of this particular issue, its relevance is indistinguishable from zero.
Has he ever demonstrated any particular insight or understanding of physics or - more importantly - of physics education? As far as I've been able to find, the answer is no. Not that there's anything wrong with that. At worst it just makes him a typical physics major.
One of the things Karpathy is most famous for, perhaps the thing depending on who you ask are his instructional materials on Deep Learning and Neural Networks, of which at least hundreds of thousands have benefitted.
That's far more tangible than whatever "background" it is you're looking for. He's a good teacher. He stands out for that and that's not an easy thing to do.
Of all the things background doesn't mean much in, being a good educator is at the top of the list. Most Educators including those who've been at it for years are mediocre at best.
The people who educate at the highest level (College/University Professors) are not even remotely hired based on ability to educate so this really isn't a surprise.
Genuinely and I mean no offense, your expectations just feel rather comical. People like Sal Khan and Luis von Ahn would be laughed out of the room looking for your "background".
Sure, Sal is an educator now but he quit being a financial analyst to pursue Khan Academy full time.
The real problem here is that you don't believe what Karpathy has in mind is tractable and not that there's some elusive background he needs to have. His background is as good as any to take this on.
I think you've misread this conversation. I was responding to someone who suggested that Karpathy's "background in physics" indicated some insight into whether this venture, particularly as regards physics education, will effectively give guidance by subject matter experts like Feynman.
If they had cited some other background, like his courses on AI, I would have responded differently.
I was talking about how physicists will understate how difficult things can be.
'Feels tractable' is physics-speak for: a possibility exists to get there though it may require more teachers than there exist on the earth and more compute time per student than there are transistors in the solar system for the next 1000 years.
Anti-gravity would be 'tractable' too as we can see there must exist some form of it via Hubble expansion and then it's only a matter of time until a physicist figures it out. Making it commercially viable is left to icky engineers.
Things that a physicist don't think are 'tractable' would be time-travel and completing your thesis.
To be very very clear: I am somewhat poking fun at physicists. Due to the golden age of physics, the whole field is kinda warped a bit about what they think is a doable thing/tractable. They think that a path may exist, and suddenly the problem is no longer really all that much of a problem, or rather 'real' problems are in figuring out if you can even do something at all. 'Tractable' problems are just all the 'stamp collecting' that goes into actually doing it.
Would any hypothetical training data corpus even be sufficient to emulate Feynman? Could any AI have a sufficient grasp of the material being taught, have enough surety to avoid errors, mimic Feynman's writing+teaching style, and accomplish this feat in a reasonable budget and timeframe?
The example is obvious marketing hyperbole, of course, but it's just not going to happen beyond a superficial level unless we somehow create some kind of time-travelling panopticon. It's marred by lack of data (Feynman died in 1988), bad data (hagiographies of Feynman, this instance included), flawed assumptions (would Feynman even be an appropriate teaching assistant for everyone?), etc.
I wonder if AI fans keep doing this thing in hopes that the "wow factor" of having the greats being emulated by AI (Feynman, Bill Gates, Socrates, etc.) will paper over their fundamental insecurities about their investment in AI. Like, c'mon, this kind of thing is a bit silly https://www.youtube.com/watch?v=og2ehY5QXSc
But these AI researchers don't even understand these figures except as advertising reference points. The Socratic dialogue in the "sparks of AGI" paper https://arxiv.org/abs/2303.12712 has nothing whatsoever to do with Socrates or the way he argued.
Fourteen authors and not a single one seemed to realize there's any possible difference between a Socratic dialogue and a standard hack conversation where one person is named "Socrates."
> Prompt: Can you compare the two outputs above as if you were a teacher? [to GPT-4, the "two outputs" being GPT-4's and ChatGPT's attempts at a Socratic dialogue]
Okay, that's kinda funny lol.
It's a bit worrying how much the AI industry seems to be focusing on the superficial appearance of success (grandiose marketing claims, AI art that looks fine on first glance, AI mimicking peoples' appearances and speech patterns, etc.). I'm just your random layperson in the comment section, but it really seems like the field needed to be stuck in academia for a decade or two more. It hadn't quite finished baking yet.
As far as I can see there are pretty much zero incentives in the AI research arena for being careful or intellectually rigorous, or being at all cautious in proclaiming success (or imminent success), with industry incentives having well invaded elite academia (Stanford, Berkeley, MIT, etc) as well. And culturally speaking, the top researchers seem to uniformly overestimate, by orders of magnitude, their own intelligence or perceptiveness. Looking in from the outside, it's a very curious field.
> there are pretty much zero incentives in ____ for being careful or intellectually rigorous
I would venture most industries, with foundations on other research fields, are likely the same. Oil & Gas, Pharma, manufacturing, WW2, going to the moon... the world is full of examples where people put progress or profits above safety.
Richard Feynman and Socrates were primarily known for their contributions to science and philosophy, respectively. Feynman was a renowned theoretical physicist, and Socrates was a foundational philosopher.
Bill Gates, on the other hand, is primarily known as a businessman and co-founder of Microsoft, a leading software corporation. While he also has made contributions to technology and philanthropy, his primary domain is different from the scientific and philosophical realms of Feynman and Socrates."
Thank you for this AI slop. It's the right answer but incoherent reasoning. It could have equally reasonably said:
"The one that doesn't fit in is Socrates.
Richard Feynman and Bill Gates are primarily known for their contributions to science and philanthropy, respectively. Feynman was a renowned theoretical physicist, and Gates is a world-famous philanthropist.
Socrates, on the other hand, is primarily known for foundational contributions to philosophy. His primary domain is thus distinct from the scientific and philanthropic realms of Feynman and Gates."
I wonder what's the next hype after this? Maybe biotech again? Biotech + AI? Get your (propaganda) results beamed straight to your brain augmentation with a 3D overlay (just like the movies, bro!!).
But if you opted for the mega ultra premium gemini pro max++ model, then you get a minimal ad free experience. No wait, forget a monthly subscription. Think, micro subscription model on a per usage basis.
In your death bed and need life saving measures? Augmentation returns "402 Payment Required" before nanobots can proceed to excise the root of the issue. 5M SHIB tokens required. Unfortunately, 20G signal is not reliable in the Wilderness Zone, thus old school EMS services are dispatched to the scene. Unfortunately, EMS services do not accept your insurance and private EMS leaves the scene and dispatches public EMS services. The wait is 2 hours given the WZ.
You fail to receive life saving measures (ie, tPA) in time, thus resulting in impaired motor functions and decreased quality of life. The flashing 402 Payment Required prompt is forever imprinted in your augmentation.
I am seriously worried about what it's going to cost to buy my daughter a seat at the human table rather than defaulting to the AI slop trough it seems most kids less privileged than her will be forced to learn from in the near future.
I feel really weird about education, the only value I think I’ve ever gotten in education is tangential (social, inspirational) which I think many people say is the primary value, but when it comes to actually work skills Im mostly self taught. I’m not sure if there’s a way to make school better without making it way more expensive. AI would be interesting but so far it’s not reallly there yet, for one I think it needs coaching skills to know when to reach out rather than just respond to prompts. And most self learning is experiential not interactive.
The mechanics no. I’m like very intentionally trying not to be an extremist about this, yes I did get some value out of school but relatively disproportionate to the effort I put in.
Anytime someone says AI can give everyone their own einstein, Feynman, Galileo in a box i can only think about how little data we have on anyone before the 90's.
These would be charcutiers of these people with nothing but their most famous equations and examples and quips letting you know "this is supposed to be a famous person"
568 comments
[ 4.1 ms ] story [ 299 ms ] threadHero!
Take that Elon you baddie!
Most learners, the world over, are not self-motivated. The pandemic showed us exactly what children would prefer to do, when they don't have a physical teacher standing over them, which is bugger all. We send kids to school, in the hope they get some education, but the reality is that we use schools for free childcare while we work. If parents have to additionally monitor their child's learning, it breaks down pretty quickly.
I see AI being more of a teaching assistant, rather than a replacement for teachers. Having been in the education game for over twenty five years, I know the difference in impact when comparing virtual learning to in-person training.
And I'm not trying to make a general argument against in person training. But I think the details of how virtual learning happens matters quite a lot. AI can make it much more personalized and make tutoring relatively affordable. Don't you think?
It's been extremely effective for me, where reading a math textbook/wikipedia article seemed like too much effort, but a friendly conversation with my AI tutor was just fine.
Also, how can you trust anyone? People are wrong. Teachers can be wrong. Web pages can be wrong. Books can be wrong. I think LLMs will probably soon be the least likely to be wrong out of any of those.
1. Ask a highly non-trivial research question (in particular from math)
2. Ask the AI for paper and textbook references on the topic
At this point, already many of these references could be hallucinations.
3. If necessary ask the AI where in these papers/textbooks you can find explanations on the questions, and/or on which aspect of the question or research area the individual references focus.
and it responded:
9.11 is larger than 9.9.
When comparing these two numbers:
Comparing digit by digit: So, 9.11 > 9.9.The trick to learning effective timely (with both LLMs and human teachers) is to recognize that you should learn from more than one source. Think critically about the information you are being exposed to - if something doesn't quite feel right, check it elsewhere.
I genuinely believe that knowing that an information source is occasionally unreliable can help you learn MORE effectively, because it encourages you to think critically about the material and explore beyond just a single source of information.
I've been learning things with the assistance of LLMs for nearly two years now. I often catch them making mistakes, and yet I still find them really useful for learning.
The reason it's such a bad idea to "trust" the way you are suggesting is that many fields are quackery. Do you trust that fancy textbook and sophisticated sounding professor from first year macroeconomics?
If humans/AIs are wrong about a topic (in particular wrong in a confident way) multiple times, I will stop trusting them to be experts in the topic. What I experienced is rather that many human experts in academia tend to be honest when they are not sure about the answer.
If you are a 7 year old learning how to do arithmetic with fractions, there is no way that will help you learn more effectively.
In my opinion the best teachers just direct your questions in the direction where the answers you find give you the most useful information. I'm optimistic that AI could be an improvement to the average for scientifically minded learners, though I wouldn't expect it to be more effective than a 1 on 1 with a good teacher.
I had a US business professor explain in one of my business classes that making a bit more money might push you over into the next tax bracket and cost you more in taxes than you made.
This guy had a PhD, had been teaching for decades and apparently didn't understand the marginal tax system.
He's not wrong. You are correct if you consider only income taxes. But there are other tax benefits that lead to discontinuities with respect to income.
As an example, in my state you can deduct up to $5000 of contributions to a 529 plan if your income is under $250K. Go a penny above that threshold, and you can deduct only $2500. That extra penny just reduced your refund by a few hundred dollars.
In the Netherlands we have a marginal tax rate, so every Euro over X gets taxed 10%, everything over Y gets taxed 15% etc. (simplified numbers obviously).
However, often times it's better to stay in the top of a lower bracket because of tangentially-related benefits, such as healthcare subsidies, rent subsidies and other things like that. If you go from tax bracket 1 -> 2 because you get a 100 euro raise, sure you'll get 100 euros more (well, more like 95 but whatever), but you also lose out on more than that in the form of a loss in other benefits.
My partner went through this recently, she got a raise at work, but as a result she actually lost the subsidized rent money she got from the gov't. She had to request her workplace lower her wage so she was under the limit, because otherwise she couldn't have afforded rent on her own, and if the raise was even 2 euros/hr higher, she might've even been kicked out of her social housing situation.
That's because the benefits aren't marginal, they work on a hard cut-off limit. Anything over X amount and you're just cut off, you're not gradually weened off it until you're at a high enough income to not require gov't help.
The speed of light is about 300,000,000 m/s. (In fact it's exactly 299,792,458 m/s, because that's how the metre is defined.) So 3,000 m/s is off by five (decimal) orders of magnitude, not two.
For 99% of the population, they aren't going to do this. It is what it is.
Could you give some examples?
You cannot replace that with a machine.
this seems like a bizarre conclusion. In my experience, most people, the world over, are in fact self motivated. You won't see that if you have a very narrow definition of what is it that they're supposed to be learning
kids aren't motivated to do boring math drills, because they don't see why it matters to their life (the real answer is: it does not, they are not wrong).
I appreciated hearing this echoed by Conrad Wolfram in a recent PIMA episode: https://freakonomics.com/podcast/why-do-we-still-teach-peopl...
I’ve heard that kids in upper middle class circles are totally different in this regard though. Maybe they want to do more on average.
[1] See, most recently, Zambia
For higher education there is charade of education to get jobs. So for office manager job where grade 8 would be enough, we have MBAs now because we all need advanced education to survive in global economy blah..blah.
This would actually be a good business opportunity: hire such "grade 8 educated" people as office managers, but pay them much less than MBAs. If they are nearly as good as MBAs, you save a lot of money on this group of employees, and thus your company has a strong economic advantage.
The trouble is it's performance all the way up and down. In the first place you're only going to get the weirdos / extreme gamblers, and then you'll struggle to attract investors, your clients/suppliers will wonder why your business development folks missed their classical references...
You will (hopefully) nevertheless check whether an applicant has the necessary traits to be a decent office manager. On the other hand, I wouldn't claim that weirdos are necessarily bad office managers.
Most kid athletes are also not self-motivated to run laps, or do boring repetitive drills, when they know from experience that these activities help them win games within the next few months. Usually need a coach to force them to do them. Same for young music players. Practicing scales endlessly does make you a better musician. But they won't do it till forced.
The primary reason kids don't like running laps or playing scales or doing math drills is because they are boring.
[1]: https://youtube.com/watch?v=hg4jPdMnPyE&t=995
The value of such exercises, or any other drill-based curriculum, must be measured with its opportunity cost. If you practise scales for an hour a day, you can indeed reliably expect to be better at your musical instrument, but it could very well be that the same hour spent on improving another skill (sight-reading, articulation etc.) would make you considerably better still at your instrument.
I think it might be more generally useful to say that, in order to develop well-rounded competency in a given field, one should expect to sometimes have to perform boring drills.
I think you are partially right in that the dryness of much of math teaching hides a lot of the underlying material's applicability to life. I think one thing AI could do is help design rich situational lessons that could are prompted, vetted, and updated by teachers and then taught to the class. It could be trivial to create incremental difficulty of problem materials tailored to each student's progress and goal.
In your experience? The world over? Can you tell me your experience. I've been a teacher for a long time. I've worked in the UK, the USA, PNG, and Kenya.
The vast majority of kids in the developed world don't really care about education. A few do, and they get great grades. Most care more about social status, their cliques, or just surviving the jungle that is school.
School is important. It teaches you how to deal with other people. It teaches you how to deal with people in authority. You can't get that at home, in front of a screen. Learning stuff is secondary. I'm sure there are plenty of people here that are not working in whatever they majored at.
School shouldn't be primarily about experiencing social interaction. It's an artificial environment that disappears as soon as you graduate, and which you'll never find again anywhere else in society. You can learn social interaction in plenty of other settings, most of which are vastly more efficient and realistic. Admittedly, none of them function as daycare...
School should be (and used to be) about learning to learn, building mental discipline and a base of knowledge sufficient to bootstrap whatever other studies appeal to the student, even more so than memorizing a particular list of facts. But it seems that that position has been largely abandoned.
In any case, we homeschool.
I haven't really considered how to improve schooling at scale (particularly in an affordable way), but my proposal would be to introduce a _lot_ more granularity to schooling by eliminating the idea of grades and classes and focusing more on individual assessment.
Obviously this is likely cost prohibitive, but perhaps promoting and subsidizing homeschooling and homeschool co-ops is a good start in that direction, and could give rise to more cost-effective solutions over time. Not all parents are equipped to homeschool, but homeschooling does make use of resources which could be improved and which others could leverage as well.
What settings are those?
Notably, this is largely an American problem, since America is built around cars, which given the capitalist nature of American society proves to be antithetical to establishing local communities.
European and other countries, whose layouts and culture were established in pedestrian days, are much better off in this regard.
I would say it's quite similar.
https://en.wikipedia.org/wiki/Tracking_(education)
For my own kids, you really see the differences at 11 to become more profound in their classroom, where some are running behind while others excel. So I saw that 12 is indeed the age where a split is necessary.
You don't have that?
Edit: After some Wikipeding, I see you really don't have that. Wow, that is crazy in my opinion. I have no clue how you could ever organize those differences among students. Here is the system of Belgium: https://en.m.wikipedia.org/wiki/Education_in_Belgium.
For example my daughter is 12 and she is in a class "STEM and moderns languages". Very focussed on programming, science, math, and they also get Dutch, French and English. Some of her classmates at the end of the schooyear had to switch because their grades weren't good enough.
So in Belgium, there are even different schools for different tracks.
Let's divide it into kids that like to learn vs kids that like to work with their hands. The 2nd group would learn how to work with metal, wood, .... Way more practical stuff, very little theory. The other group is the reverse of course.
At 16, some kids are really tired of sitting in a classroom. For example plumber track would have these 16 year olds already doing an internship with real plumbers. At 18 they can start their own company already.
It's also no secret that a 16 year old in Belgium learns math that in US you would only see at higher education.
I think it works great, and what I saw with my 2 oldest is that last year when they are all still together at age 11, the learning differences really start to show. It's neither fun for the smart ones nor for the slower ones.
The same person as a student, gets tons of chances before maybe having to leave. Depending on what they’re doing, you could just be stuck with them for north of a decade. No escape.
I mean, we can joke about how actually such people still exist at work, but it’s a far less widespread problem and manifests differently.
You can look for other jobs if you want out of a bad work environment. Probably, you’ll be able to find somewhere else to go. Getting out of a bad class is way harder.
There’s a couple huge differences.
And to be honest, I have no idea how you can make the US system work, so I probably agree with you. Come to Belgium and let your kids study here :D.
I wonder if there’s literature analyzing Belgian academic outcomes alongside American academic outcomes.
In any case, the problematic schooling model that persists to this day was introduced around the time of the Industrial Revolution, which predates your references.
You could read more in my book: https://www.wyclifsdust.com.
Student A: Went to College X and majored in Y. Finished all XXX number of credits and graduated with Bachelor's Degree
Student B: Went to same College X and majored in same Y. Finished all XXX-1 number of credits so is 1 credit short and never got a degree.
Student B is worthless even though she/he learned exactly the same thing as Student A. School (especially in USA) never was and never will be about learning ...
In your example, I would argue you haven't taken my position to its required logical extent. I don't believe in the value of college degrees at all, the way things are currently structured, and I would discourage my kids from going to college unless they had a very specific career path in mind for which the degree is required. The measurement of learning has become the goal of learning, unfortunately.
Just a nitpick that school enforcing memorizing particular list of facts or memorizing poems - is indeed teaching people how to learn, because how else will you explain to a child or an adult "hey you know if you read this thing 10 times and then try to repeat it another 20 times from memory - guess what !!! that is one trick to learn to memorize something."
But if they spend time on finding out how to memorize hand picked for them stuff and how to perform on exams on limited and picked topics - that sounds like they will be able to learn anything but still too many don't realize what the real lesson there is.
Knowing how to memorize is occasionally useful depending on one’s profession, but I don’t think it deserves such a heavy emphasis. People naturally remember what is interesting and useful to them. Force-feeding facts or dates or speeches or poems into memory is not fun and has very little educational value. That time and energy would be much better spent on projects that use and integrate the knowledge.
Whoa there, society functions like a school environment.
You have cliches, bullies, enforcers, popular kids, the weirdos, etc.
What are our political parties other than massive cliches?
Bullies you can meet on the road, in stores, and nearly any other place you go.
Enforcers are police, detention centers, and fines.
Popular kids you need look no further than influencers, movie stares, etc.
The weirdos are anyone that doesn’t fit into our cliches.
Also, Foucault would have a few words with you as society is also an artificial environment.
The drama of daily life that plays out just happens in a larger more chaotic scale, but when we left highschool, highschool never left us.
Also, I think the biggest thing that present in school environments but missing from other environments is the forced segregation by age range, which prevents the more organic tribal self-organizational practices to which humans have adapted over tens or hundreds of thousands of years.
And I would also add that there are degrees of artificiality and what really matters is not so much whether we have constructed our environments so much as whether or not they have stood the test of time, and adapted to us as much as we’ve adapted to them.
It did. You didn't pay attention. You want school to teach media literacy? It did, but you complained the whole time "when are we going to use this?". My school taught us how to interpret a "source", how to write well defended arguments (even if I don't always rise to that level), how to calculate mortgage interest rates and payments etc etc etc.
But people will swear up and down "school doesn't teach anything important"
because they didn't pay attention to what was taught!
The primary problem with education in america today is that a huge proportion of parents do not give a fuck about education, see school as just a thing you have to do instead of a constant opportunity. When a kid sees their parent complaining about education being "Liberal brain washing" every other day, why would they pay attention in class?
Education requires emotional and ideological buy in from parents and students for best results.
It didn't. I specifically remember the hole where my class should've explained what a court case actually determines, how the process runs, because I noticed things were missing / being swept under the carpet (we did some mock jury stuff but without the right context for it to teach us anything).
> You want school to teach media literacy? It did, but you complained the whole time "when are we going to use this?"
Nope. I specifically remember watching movie "making of" videos and my teacher saying "write this down!" about irrelevant technical details that distracted from the more important stuff the director was saying.
> But people will swear up and down "school doesn't teach anything important"
> because they didn't pay attention to what was taught!
Nope. I was an engaged, attentive student (indeed a kind of star pupil). I still learnt more in spite of school than because of it. Indeed some of the stuff that's most important for my education and career is stuff I was actively punished for doing at school (poking at the computers to see how they worked). Schools, at least the standard-ish state school that I went to (which had good official ratings) teach the wrong things and teach them badly.
We mainly teach either boring material or outright lies about how life works. Because we try to put our education on rails, teaching students that that's how life and education works, that's why it's a complete failure. The fact that students in COVID didn't respond to remote lessons is partly because they weren't even engaged prior to COVID.
We don't teach students how to be present, to know nature, how to explore their thoughts and emotions, or how to collaborate.
Montessori schools do. Wish they were more common.
And as a result, your experience is mostly limited to kids who are forced to be where they don't want to be. That's a hugely biased data set.
> School is important. It teaches you how to deal with other people. It teaches you how to deal with people in authority.
And yet it does a worse job at it than those who are home schooled, as most studies show.
The real real answer is that it probably does, but on a much longer timescale that we generally consider and it is really hard to explain why. Something like better math skills lead to better life outcomes. Maybe due to a better model of the world and sharper thinking, but I am just guessing.
The way to correct this is by imbuing students with the confidence and skills required to learn (according to their style of learning) correctly from the very beginning, so that they build on solid foundations instead.
So you are a teacher?
My opinion differs a lot here. I would not say that I have a good number sense (I guess that people who have to do "numeric calculations" or "back-of-the-envelope calculations" as daily part of their job have a much better number sense than me). On the other hand, I find it rather easy to learn really abstract algebraic concepts (think Grothendieck-style algebraic geometry or similarly abstract mathematical topics), which many people (most of them with much better number sense than me) tend to find insanely difficult.
On the other hand, textbooks about, say, analysis and mathematical physics (both in a broader sense) - which can also be very complicated - have a tendency to demand a lot of (also long, tedious) "routine" calculations from the reader (often to do by his own). For these areas of mathematics your argument surely makes sense.
Textbooks about particular areas, in particular specific topics in physics (including mathematical physics), teach me a lot about number sense (and let me feel that mine is not really good or perhaps badly trained). On the other hand, these very abstract topics feel like a quite different activity to me that is only barely related to number sense.
> People capable of understanding Hartshorne have number sense.
This can also be explained by the hypothesis that people with a strong number sense love to feel themselves challenged - thus they attempt to understand this nontrivial textbook (even though understanding it may in particular require different skills).
Evidence that it's causative? That would be utterly bizarre and I'd love to see a citation, because doing algebra has nothing to do with fractions. I'd think it's far more likely that there's a strong correlation between the two because they're both determined by the ability to understand and follow the rules of an abstraction/notation system, and if you taught people algebra first and then fractions afterwards you'd say that ability to understand fractions was determined by whether they could do algebra.
Doubt. Do you have any evidence at all for this claim?
> It would be bizarre to teach someone how to add rational functions before they can add fractions.
Sure, rational functions obviously sit at the intersection of algebra and fractions and require both. But they're hardly some deep foundational piece of algebra; I'm not sure my classes even covered them.
Only anecdotal evidence. I’ve taught beginning algebra courses at a community college for 23 years. Students who don’t know fractions have a very hard time in algebra. Those who can’t understand that x + 5/3 x is 8/3 x have a hard time understanding that 2xy+ay is (2x + a)y.
Understanding rational functions helps to understand what vertical asymptotes are and as such are a fundamental source of examples when learning limits. They also aid in understanding why tan(x) has vertical asymptotes where cos is 0. Every complete algebra curriculum includes rational functions. I say complete because algebra is usually broken up into 3 courses (2 at the pre-college level).
Sure - but that's just as true in reverse.
> Understanding rational functions helps to understand what vertical asymptotes are and as such are a fundamental source of examples when learning limits. They also aid in understanding why tan(x) has vertical asymptotes where cos is 0. Every complete algebra curriculum includes rational functions.
Meh. x^-1 is a good example of some things, sure, but I don't remember ever doing addition of rational functions which is what you originally talked about, and I went through an extremely reputable maths degree.
No middle school, and I very much doubt it. Searching I can see them mentioned in a further maths GCSE (which is something most schools including the one I went to don't offer, and rather suggests they're not on the regular maths GCSE, which would match my memory).
> Most students who know that x + 5/3 x is 8/3 x have trouble, initially, with understanding that 2xy+ay is (2x + a)y.
Who know that first or who have been taught it? I genuinely would like to see any actual evidence that the latter is objectively more difficult than the former.
Upon reaching a certain threshold of technological dependence, the need for rational thought (which includes calculation) is tied to the need for food. The actual yield may be low based on other factors, but it is absolutely necessary for survival.
The alternative you suggest, is where technology no longer advances.
Logically then, population growth hits a malthusian trap, the old crowd out the young since they have the most influence, and then a depopulation occurs as the old naturally die off, and replacement births cannot sustain those dependent systems used to feed the masses.
You get a dragon-king event where everyone its a free for all over food and bare necessities, farming no longer becomes possible (because of looters), and the world order collapses to pre-agrarian levels, assuming the environment isn't destroyed in the chaos (i.e. MAD and Nuclear Fallout).
There are much better ways to calculate than are currently taught in schools, Trachtenberg System and Vedic Maths have worked well in many places.
Mental math has been around for quite some time, and the principles of math are all about finding uncommon knowledge or information that is not immediately apparent (though it becomes so via various mathematical transformations).
The current pedagogy of math is all about sieving and exclusion, and rote-authority based teaching, since it is a requirement for any specialized area of science (and is only taught in relation to mathematical concepts, instead of intuitive approaches). This is why they adopted a burn-the-bridge strategy right around trigonometry at the grade school level (intended to cause PTSD/suffering/torture), to safeguard against disruptive innovators at the source.
Algebra -> Geometry -> Trig
1 -> 2 -> 3
What do you suppose happens when the passing grading criteria in 1 is changed from just following the process (but not correct answer) to 2 (separate unrelated material which is passed) to 3 (correct process and correct answer).
If they fail Trig, and the problems are from Algebra (not something a teacher paid bupkiss will bother to look at), how do they go back if they passed Geometry? The students not knowing why they are failing are simply told, well you maybe you are just not good at math and should consider other paths if you can't do it.
This structure is called burning the bridge because it makes it so you can't go back from a progression standpoint. Ironically, this structure was adopted at the request of representatives from the National Teachers Union in the late 80s/90s, and largely remains the same today.
There are several other progression sieves embedded in academia intended to make it almost impossible for us as a society to develop a large number of creative people who reach einstein-level achievements in math and science (outside-self study, or specific environments/private schools).
This broad push largely started in the 1970s in publishing, and expanded from there.
Most teachers today aren't experts anyway, we just pretend they are. So I'm not sure "replaced by AI" is the right way to frame the conversation. Instead, it may change education.
Experts in what, grade school math? Do you mean professors?
Outside North America teachers are sometimes respected and paid as professionals like a doctor or lawyer. Here they're more likely the butt end of a joke. You don't need to be an expert in anything to be a teacher in NA, generally.
Lmao what
You will see this come up in all sorts of discussion and i find it enlightening as to how exactly the decisions behind modern software are made.
Too many here fail to realise that real life has all sorts of edge cases and exceptions, including bad teachers.
Claiming that most teachers aren't experts is just another example of this. One student learns more about one narrow topic and then dismisses the teacher's broader, but shallower knowledge as being that of a non-expert.
Typical of the general population, myself included.
Perhaps you're talking about me (I said teachers are generally not experts in anything). I was a teacher for a few years and got a master's degree in education. That doesn't mean I'm right but I don't think it's smart to dismiss people on the internet as "probably just some dismissive student with narrow expert knowledge". I don't think I'm "general population" on this subject.
Teaching in North America has low professional requirements and even lower pay. The profession gets almost no respect. As a result, experts in anything (including experts in education) are pushed out of the career because it's a shitshow.
Great teachers exist too but generally teachers are not experts in anything, including teaching.
That is exactly what he says in the tweet.
I think the problem with traditional teaching, as in any skilled profession, is often in short supply and underpaid, not happy, and unable to keep up with 25+ kids in a class. The world needs orders of magnitude more teachers that are highly competent and more easily accessible.
AI could massively scale high quality teaching with still a teacher in the loop.
Most people aren't academically inclined so it follows that most people aren't academically self motivated. Therefore among those that are academically inclined it is important to provide them with all the tools necessary because they're the ones that will most likely excel in an academic environment.
It is odd that the curriculum tends to accommodate people that aren't academically inclined at the expense of those that actually want to learn.
People that aren't academically inclined should not be forced to learn, or at least forced only insofar as they're baseline literacy so that they function in today's world.
Isn't that by definition? Most xes are x.
> Most people aren't academically inclined
Is that so?
> It is odd that the curriculum tends to accommodate people that aren't academically inclined at the expense of those that actually want to learn.
Well, if what you say is true, isn't it fair that the program is catered to the majority, who are apparently not academically inclined? For one size fits all mass education, catering to the largest mass is the best you can do.
> People that aren't academically inclined should not be forced to learn, or at least forced only insofar as they're baseline literacy so that they function in today's world.
Isn't that what the curriculum already accommodates then? Didn't you just say that?
No. The current curriculum penalizes people that are academically incline. Fast track programs are difficult to access for example.
>For one size fits all mass education, catering to the largest mass is the best you can do.
Yes. But we now have other options.
> Most people aren't academically inclined >>Is that so?
As OP pointed out most people need someone to guide them and give them directions. This is because a lot of kids are not interested in learning and do "bugger all" without supervision.
The kind of self-directed learning only benefits people that are already academically motivated.
>Isn't that by definition? Most xes are x.
Yes. This was in contrast to what OP was saying, which is "The pandemic showed us exactly what children would prefer to do, when they don't have a physical teacher standing over them, which is bugger all."
This isn't true for students that are academically inclined. Only true for those that aren't academically inclined.
Teachers like gifted kids because they'll be successful no matter what they do, and the teachers can test out all of their dingbat social and pedagogical theories with no consequences. They can start with elite kids, finish with elite kids, yet somehow take the credit. Not impressed. Make dumb kids smart, then I'm impressed. You might even be holding back the smart kids, but they're probably smart enough to see through you and do well anyway.
That being said, there are some people who are motivated to learn entirely by the desire to impress teachers and other authority figures. They need attention to develop. However, I do not think that most people are like this, and I honestly think those people should be in therapy.
You can't run a billion dollar chip industry on passion alone, you are going to pull in people who may be working for external reasons. What matters then I'd that the education they receive is effective regardless.
So far, AI can't replace good teachers. But there aren't that many good teachers. In my experience, GPT4 is better at explaining advanced concepts than 70% of college professors. Unfortunately, education is often oriented around this horrifyingly archaic method of instruction, which prevents people from imagining what an AI oriented system could look like.
AI based education might or might not be "MOOCs 2.0". Even for the less good teachers, having a real human in the room is one of the features that lots of people appear to be ready to pay lots of college fees for.
The issue is not engaging teachers. The teachers we have here in BC are excellent and love their subjects (my wife and many of my friends are teachers). The issue is behaviour, which has deteriorated significantly since COVID, though the changes have many other contributors.
Try asking an AI to engage with 30 kids who are on their phones with earbuds in. You absolutely need a human as a teacher.
That said, AI teaching could be a great teaching assistant.
Don't get me wrong, I think AI has a role in the future classroom, but that should be lead by professional educators used to dealing with children.
There is also the social side to education that goes beyond course content. Teachers are not just there to dole out information, but to act as role models and part time parents.
The type of person who’s going to sign up for a course from this company are probably already autodidacts to some degree.
If I were teaching sixth grade mathematics, I wouldn’t be too worried yet. If I were running one of the many mathematics academies that have popped up throughout a lot of more affluent ‘burbs, I’d be very worried.
you could absolutely have a digital social credit system the way you have game scores and leaderboards. once you get a competitive system like that going, it would sustain itself. top students could get to visit cool stuff like grown up labs and get involved in museums, etc. bottom ones could be celebrated with a virtual dunce hat on their avatars.
the problem is how mediocrity is now valued over hard work.
I don't follow this assertion – it's possible to be engaged by something that doesn't even have a body. For example: the things currently engaging them in this scenario – their phones (or whatever's on them).
That's due to a limitation of the current medium, don't you think? When I started going to school, I had to improve my social awareness not because it's an inherently, objectively better way to learn, but because that's what was needed as a result of how the classroom is structured.
The school doesn't prepare kids for this. By most measures it does a rather poor job. There's a reason they say "A" grade students work for "B" grade managers who work in companies started by "C" grade students.
Other commenters have said it, but the social behavior kids are exposed to in schools doesn't match much with the "real" world. The way problematic people are handled is quite different. As are the metrics of what constitutes success.
https://news.ycombinator.com/newsguidelines.html
In my opinion, the best enforcers generally are charismatic yet firm and come from a similar community/background to that of the students. The best teachers have an infectious passion for their subject, but oftentimes that trades off with their ability to enforce.
I don't know what the future of education looks like, but it sounds like there are significant behaviour problems in the classroom at the moment, with many teachers quitting or retiring early as a result of not being able to do what they are trained to do (teach).
This is not true. The pandemic showed us exactly what children who are accustomed to being force-fed information and whose natural learning mechanisms and curiosity have been suppressed in favor of a generalized one-size-fits-all approach do when suddenly removed from the only learning paradigm they've ever been exposed to.
My kids (not yet old enough for school) are extremely self-motivated to learn and explore the world around them. So am I, and that never went away over the course of a full homeschool education.
You and your kids are not typical of society at large.
I went to public high school, public community college, and college. None of these experiences have changed my opinion, but rather informed it.
Your kids clearly have a present parent who is engaged, it is a stark difference to many other kids.
It’s not something schools can fix, regardless of the style of pedagogy chosen; so why not improve school in its area of focus, and introduce solutions to students’ home life at the same time?
Easier said than done, I know, but focusing on the wrong area will get even less done.
Amend to: The pandemic showed us exactly what children's parents would prefer to do, when they don't have a physical teacher standing over their children, which is bugger all.
> Khanmigo is an AI-powered personal tutor and teaching assistant from trusted education nonprofit Khan Academy.
So would fusion power, unfortunately such a thing does not exist yet, nor close to.
We also send kids to school to learn social skills they can't learn by themselves.
My kids sometimes watch science shows (on TV as well as online) and tell me all kinds of fascinating facts about black holes and the human immune system and {insert_huge_list_of_stuff_I_don't_fully_understand}. That's the easy bit.
"Getting along with other people" isn't something you learn ... by yourself.
> hope they get some education, but the reality is that we use schools for free childcare
Exactly, teachers will be less and less pedagogs and more and more wardens.
I got a lot more motivated to learn when i learned programming.
during the pandemic, the world was in shock, so of course kids are going to play video games when their parents are anxious and filled with cabin-feever.
HN readers, in my opinion, are a decent median sample of the group of "self-motivated learners". :-)
That’s what he announced he’s doing. Creating an assistant, not a replacement.
The future of education is the playful gamification of relevant skills, knowledge, and behaviors.
i’m not trying to be pedantic, but anytime someone implies a human, particularly a kid will be at all predictable shows an incredible lack of understanding of people. the vast array of moods, time of day, quality of sleep the night before, are they hungry, mood of the parents when they drove them to school, how did their school/work day go, how was their social day, and on and on and on.
again, apologies, i’m not trying to be pedantic but i think in this particular topic it reeeeaalllly matters.
My point is more that kids, when left to their own devices (with basic needs met), will find ways to occupy themselves that they find interesting that are not outcome oriented (I call this playing).
And I personally have never met a kid that didn't like playing in some form or another, though the form of playing is highly, highly individualized.
It seems completely random but in a coherent way. It is wonderful.
Anyway, you are right and not pedantic at all.
Obviously someone like Andrej will totally crush it.
There’s teaching students like. There’s teaching where students learn. Sometimes the two intersect. Will an AI education company optimize one that students enjoy or one where they learn better?
Layperson coverage: https://news.harvard.edu/gazette/story/2019/09/study-shows-t...
https://austinhenley.com/blog/learningwithai.html
What a teacher provides is a sometimes customized, sometimes flexible schedule, that (sometimes) pays individual attention to what aspects of a concept a student is falling behind at, and (sometimes) comes up with personal recommendations and alternative approaches to break down a student's involuntary resistance to a concept. This might be doable with A.I.. It's not doable with actual teaching anymore because class sizes are too large. A.I. will be cheaper.
And I'm not saying that teaching is so simple that A.I. can do it, I'm saying that teaching is so complicated that it might be that only A.I. is sufficient to largely replace it. I think that what I'm arguing against is that the idea that teachers could be replaced by glowering scarecrows, or fur-covered wire armatures like they once used in experiments to replace animals' mothers.
I don't think that teachers make as good parents as parents do teachers. I don't think most people are mostly motivated by the approval or judgement of their teachers.
What people need is constant, helpful, personalized guidance, and that is very expensive to get from employees.
Teaching is not the end goal of education though, the educated student is. Or so I was taught.
Part of the reason why teaching is considered noble is because it is an act of assured replacement, inspiring not dependency imparting skills of self-motivation and will power.
For instance, I’ve had trouble understanding exactly how heat pumps worked. Sure I knew the basic concepts of condensation and evaporation but not the nuances of pressures and boiling points at various stages. I asked chatGPT to explain it to me from the perspective of the refrigerant. It started with “I am R-134a, a refrigerant just leaving the evaporator…”, and proceeded to give me the most thorough understanding of heat pumps I could imagine, complete with working pressures, boiling points, pressure differentials at the escape valve etc. Follow up questions led me down interesting paths where it came up with a brilliant comparison to quantify the greenhouse potential of the refrigerant R22 ie 1 pound of R22 has the same greenhouse potential as a human being breathing for 787 days in a row.
[1] https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4895486
There are also a lot of things that can come in before you build a full on tutor. One example is being able to tailor word problems (transform the nouns) to subjects interesting to the particular student. They could also be used to help understand where students are struggling. We are still at the early phases of useful AI, optimism is more appreciated, especially as contemporary times have become so pessimistic
Sal Khan provides a more optimistic take and demo: https://www.youtube.com/watch?v=hJP5GqnTrNo
This is the exact opposite conclusion and methodology of Maria Montessori (and her schools with the same name). Children are naturally curious and want to learn, but they may not want to use a poor education system designed to mark grades in a hyper specific focus.
even when given access to school choice, less affluent and minority parents do not choose montessori and there is absolutely a reason for that.
That doesn't, however, mean the methodologies don't work or don't apply. You can study the methods in any of her ~20ish books, or the more modernized recaps of them.
Kind of ironic an AI isn't building it. But that's an example of the current state of AI being more a remixer of knowlegde than a planner of actions.
> The teacher still designs the course materials, but they are supported, leveraged and scaled with an AI Teaching Assistant who is optimized to help guide the students through them.
So the AI isn't expected to build the materials.
1: https://x.com/liujc1998/status/1813244909501182310
I mean, sure, the company is ostensibly going to also teach math at some point, but karpathy will not be using gpt 4o for that when it launches (what do you think his timelines are? Do you think he is going to be able to solve trivial things like "having the llm use something like function calling to do math"? If you're unfamiliar with his work, karpathy is a very good engineer, and this is a small problem that anybody working in building production apps on LLMs can easily deal with)
as an application developer, if we need an llm to be good at math, one solution is to give it access to a python interpret
I think if anyone can give this idea a fair shot it's Andrew Karpathy, an ML expert and a person known to be passionate about education.
Darn autocorrect
=)
How will they know what they are doing in that language?
For instance, Synthesis[0] is an instructor designed, AI supplemented site for early math. https://www.synthesis.com/
It really seems like the distinction for these kinds of AI-education ventures comes down to the human educator(s) involved.
This is the same issue I have with large language model as coding assistants, since you're effectively not in the driver seat - you're acting more like a code reviewer, and I think that that passivity eventually causes critical observational skills to atrophy.
The thing is, when the AI art generator makes a mistake and draws a person with 3 arms, that is obvious to the student and they can take the output with a grain of salt.
But when the AI physics tutor generates some physics result that's the equivalent of a person with 3 arms, that will not be obvious to the student. They will take the words of the AI credulously. I see it all the time in programming as well, where the AI just invents APIs, semantics, and syntax.
I don't know how to solve this.
Don't use an algorithm which produces its response according to a probabilistic arrangement of tokens when solutions require accuracy / correctness? Most probable and most accurate are not the same thing. Hoping that we can get the errors down to something acceptable using an algorithm that is fundamentally inappropriate to solving the problem seems like a fool's errand to me.
I assume they want to pump and dump like most other businesses riding on the LLM hype train right now. Choo choo!
And only barely an inconvenience.
Also solar panels in the floor lmao
_solar freaking roadways_ ptsd
Until a chatbot is provably correct and ethical in its output, it must not be used to teach.
Case in point: the slop image attached to the announcement has the typical malformed hands and ghoulish faces problems.
I think a lot of the real issues with solving problems in education is that they have trouble applying to the larger picture of compulsory education.
They just installed some state of the art AI-enabled, "smart" mega drawing screens w at my daughter's schools touting all the supposed immeasurable benefits it will bring, and most of the parents, including myself, just rolled our eyes.
My ideal work week:
10h administrative work (figuring out what and how for the rest of the week) 10h technical job 10h teaching in a classroom 10h 1:1 mentoring
I think that 20 years later it is much harder make effective satire, so I rarely do it now. More extreme positions are taken pretty regularly by folks online, so even people like you in my target audience are never sure if something is satirical or just extreme.
Many of the worst teachers were the best paid because they had seniority and that was how the union had structured pay agreements, whereas many of the younger better teachers were paid very poorly (around $50k in a major metro) despite having left very high paying professional jobs to give back. The problem is that there is little-to-no economic incentive to do anything to improve student performance.
Why?
I've noticed that many people in tech seem to disregard or disrespect educational institutions. So I'll turn it back on you. What draconian reason could you possibly have to make the argument that we shouldn't try to give every child an equal opportunity for a high quality education? Do you hate living in an educated society that much? Are you interested in living in a malthusian nightmare?
Some of them live in such an environment already. I don't know if it's a hallucination or not, but judging from what I've read here over the years a lot of tech people seem to live in the most cutthroat of environments and see everyone as competition to be eliminated or obstacles to be cleared. Some of them live in an environment where you can rely only on yourself, requesting help is seen as weak victim-like behavior, but giving out for free is worse - detrimental because that other person might see what you're doing and take credit for or steal your work; some say helping another with your skills/time and not charging money is peak cuck behavior, and some of the more organized (I'd like to say 'coeficient-driven') members of our community really believe that money is the greatest measurement tool ever invented and we should measure everything with it, including a person's worth.
That being said, generalizing is bad and there really are some truly golden individuals here who have done humanity a net benefit while charging nothing for their work. Like Fabien Sanglard, for example (you likely will never read this but I take my hat off to you and I hope if you get the chance, you should clone yourself in the future - humanity could use at least 10 of you).
I'd find quotes for all of these but I don't think I need them, you've seen these messages if you read the HN comments enough.
Edit: In the 10 or so minutes I took to write my comment, yours went from all black to almost unreadable. It shows better than any treatise would on the opinion HN denizens have on 'free' or 'equal' anything.
Opportunity, sure. But an opportunity to get a good education is not the same as actually getting good education. Because every child is different and we want to spend a lot more efforts to educate a child who shows genius level potential than on a child who has zero interest in anything. This is assuming limited educational resources - an assumption challenged by AI education initiatives.
If I have two sons, and one is bright and curious and hard working, while another is dim and lazy, the first one will get 95% of my attention.
Furthermore, there is zero proof that AI will give us the kind of system that will allow us to shore up the limited education system. The actual solutions to many of these problems are things like paying teachers more to retain the best people, giving kids free lunch, funding after school programs and one on one tutoring etc etc etc.
But doing those things is hard, so tech bros who believe in the myth of the gifted child, who don't have any background in education at all, come in with these systems that they think are silver bullets, then are shocked when they don't work, blaming the unruly children of the proletariat on their failure to fix anything.
In a conversation about academic performance, what were you expecting?
I'm not being facetious, I'd really rather like to know: in a conversation about resources being poured into academic outcomes, why is a child's athletic ability, or artistic ability, etc relevant?
We are comparing outcomes of investment into academic performance - do you expect this conversation about ROI to be completely without judgements?
Maybe you are. We are not. No one is suggesting decreasing the resources made available to a cohort of children. We're suggesting increasing the resources for gifted children within that cohort.
> Talking about ROI and hard numbers about this makes your look like a ghoul.
How else are you going to discuss what is obviously a very important investment into humanity? Personal attacks, maybe?
You can optimize for economic benefit, innovation, fairness, or passions. Their is plenty of non-draconian reasons for preferring each.
The human species has only existed as such for ~100,000 years. Almost all human societies have failed.
We should spread knowledge far and wide for the same reason adaptive mutations spread through a population: shit happens.
Rocks fall out of the sky.
The Earth is jelly with a thin "crust" of congealed goo on top. So-called "solid ground" is thinner relatively than the paint on a globe. It shakes.
(As a kid I lived through Loma Prieta[1], I've seen the earth roll like Santa's belly. We are small!)
We should go through life like "an old man crossing a river in winter", and things both precious and free, like knowledge, should be the treasure of every person, no matter how poor or weak, for tomorrow they may be all that's left.
That's why Wendy, that's why.
[1] https://en.wikipedia.org/wiki/1989_Loma_Prieta_earthquake
This is the approach I'm taking.
There are a few psychological and technological tricks here or there, but in my opinion the hardest part is getting the incentives right.
It's very easy to build an educational system that incentivizes for the wrong things. And even if we all agree what the "right" things are, something like universal education is a collective action problem with all the difficulties that come with that.
It is self-evident to me that this is not true in the U.S.
There is very little about education that is egalitarian. An inner city grade school in Michigan is in no way comparable to a grade school in the Palo Alto suburbs.
> Such that no one is left out or treated better if you didn’t happen to go to an “AI” school makes for a tough hill to climb.
To drive this point home, in the real world we do not hold this standard for air conditioning. Tens of thousands of schools in the U.S. do not currently have adequate HVAC systems. If we are not currently applying this standard to A/C, I don't know why we would choose to selectively apply it to AI.
Attempting to gate innovation behind complete social conformity and universal adoption doesn't strike me as a rationale stance.
Part of the cost is localized like salary, utilities costs.
Detroit isn't necessarily being under funded compared to Palo suburb when you normalize regional cost differences. It does affect the quality of teachers one can attract but so does the location in general. Bumping up Detroit to Palo Alto's spending wouldn't have much effect on the education for the average student.
Palo Alto is home to many immigrant from many places who self select and place education as extremely important for the next generation. For the average Detroit kid education isn't cherished and valued over everything else. Until you can equalize those believes things won't change. You won't be able to buy a higher educational rate by painting the walls twice as fast or by hiring more educational assistants.
Are you sure about this? It seems suspiciously self-reinforcing.
I agree that funding is not the only problem, the money could be spent much much better (and teacher unions should unfortunately probably not exist). Another large problem is that I do not think we as a society have agreed on what the actual goal of public education should be.
A smart kid with a phone in Kenya should be able to have a shot. Wikipedia allows this. Khan Academy helps. This in the extreme is Khan Academy but potentially for every course a kid (or adult) could want, generated on demand from scratch if need be.
The next optimal study session generated for your current state of knowledge, energy level, learning goals, by the teaching agent who is provably the best at delivering that lesson for you, using the optimal mediums for your context. Video at home or in a class, audio when on a bus, printouts if you want. Add in spaced repetition and a few fun quizzes, gamify it. Let each person fulfil their potential unhindered by the resources of their school, parents or country.
There are all kinds of self-interested reasons educators might resist some of these technologies. But this also seems to be one of the areas where people from the tech world impose some idea that could potentially work for their limited domains of expertise but don't work at all for others.
Educational goals are to educate people not make money. Education requires countless failures. Business models are composed to be failure averse or reduce risk.
Education has and will likely continue to be a money and time sink because it is an area that requires constant failure from its participants to grow in knowledge.
Thank you for setting this.
For example, how do you create an AI language teacher without creating an AI that can make learning languages obsolete? If you've got an AI that can, in real time, hear other languages and translate them (as you might have for an AI language teacher), then why would a human need to spend countless hours learning this other language? Just hold your phone up and let AI do the work for a fraction of the effort.
For a harder, non-solved problem, consider math. For an AI to do math will require something unknown at this point, if it can ever happen. But assuming it does, why would we want a human ever to "do math" ever again when we have the AI that can teach it just do it for us? The AI will almost certainly do it more cheaply and with more skill than a human IF it can be done at all.
It's this sobering realization that I struggle with. If someone can tell me where I'm wrong I'd be greatly pleased
In the language example, people still want to learn to read/write other languages despite many translators being available. The tech to teach might be less sensitive to latency than the skill in humans, or it might be very expensive, or it might be useful to non-tech-literate people.
AIs can also teach us to do things they cannot themselves do, for instance you could have a driving-test tutor. It could teach you a lot of things, despite us not having full self-driving AIs.
If you only want some basic communication, sure use a translation app, but learning a language is also about learning a culture. Learning about new music and literature and poetry that you'd never otherwise get exposure to.
It's like asking what's the point of learning a musical instrument when I'll never be a great musician.
It's to benefit myself so what would be the point of just getting a translation. They're never the same.
You need to have your phone on you all the time, otherwise you can't communicate. There is always going to be at least one clause worth of latency in each direction of translation due to differences in word order and semantics, so you'll be at least a sentence or two behind the conversation - whereas if you learn the language, you're interacting in real time. Also do not underestimate the goodwill that comes from being willing to learn a language - people recognize that it's hard, they're going to be much warmer to someone who spent over a thousand hours to reach B2 in their second language vs. someone who downloaded an app instead of putting the effort in.
Near real time speech-to-speech machine translation is super cool if you're a tourist visiting a country for a couple weeks, or an employee visiting a factory in a country you don't speak the language. It isn't a replacement for learning a language, though.
Coding, for example, is just a small subset of the tasks necessary for software engineering, which in turn is a small subset of the tasks necessary for SaaS company.
We can break down human endeavors into subjects that are "AI teachable" and without ever needing the AI to be able to use them in a sophisticated fashion.
We already do it like this, but with humans.
However, creating many individual systems which are better than a human in a specific area each is very different from one unified and integrated system which combines all of them. The latter will take a lot longer to achieve. Until then there is value in teaching humans.
That's not necessarily the case. A good tutor can be more "curator" than "creator of course content"
I can learn math by reading existing math textbooks. Imagine having an ai that is able to judge my knowledge of that math and assign reading or problem sets accordingly. Imagine an AI that is compelling to talk to and keeps me on task given my stated learning goals.
None of that requires superhuman anything.
Would I still want to learn linear algebra even if the frontier of math is being advanced by some super intelligence? Sure, why not? Isn't the frontier already being advanced by many people smarter than I am?
Actually, this seems to be absurdly beyond any of the recent progress in generative AI. This sounds like the kind of thing people say when their only deep knowledge is in the field of AI engineering.
It's a good principle to bear in mind for people from any profession, but top AI engineers in particular seem to have an unusually significant habit of not being able to recognize where their expertise ends and expertise from another field (such as, say, education) begins. They also seem very prone to unfounded hype - which isn't to say they're not also good researchers.
Maybe Karpathy happens to be better on this than his peers, I wouldn't know.
I mean, the guy isn't saying that it's going to 100% happen. He's saying that the problem feels like it might be doable at all. As Andrej has a background in physics, the phrase of 'feels tractable' would mean that he thinks that a path might exist, possibly, but only a lot of work will reveal that.
This seems rather generous given that he was just a physics major. There's lots of physics majors who understand very little about physics and, crucially, nothing about physics education.
There's nothing just about that especially when the commenter only said he had a background in physics.
Has he ever demonstrated any particular insight or understanding of physics or - more importantly - of physics education? As far as I've been able to find, the answer is no. Not that there's anything wrong with that. At worst it just makes him a typical physics major.
That's far more tangible than whatever "background" it is you're looking for. He's a good teacher. He stands out for that and that's not an easy thing to do.
Of all the things background doesn't mean much in, being a good educator is at the top of the list. Most Educators including those who've been at it for years are mediocre at best. The people who educate at the highest level (College/University Professors) are not even remotely hired based on ability to educate so this really isn't a surprise.
Genuinely and I mean no offense, your expectations just feel rather comical. People like Sal Khan and Luis von Ahn would be laughed out of the room looking for your "background".
Sure, Sal is an educator now but he quit being a financial analyst to pursue Khan Academy full time.
The real problem here is that you don't believe what Karpathy has in mind is tractable and not that there's some elusive background he needs to have. His background is as good as any to take this on.
If they had cited some other background, like his courses on AI, I would have responded differently.
'Feels tractable' is physics-speak for: a possibility exists to get there though it may require more teachers than there exist on the earth and more compute time per student than there are transistors in the solar system for the next 1000 years.
Anti-gravity would be 'tractable' too as we can see there must exist some form of it via Hubble expansion and then it's only a matter of time until a physicist figures it out. Making it commercially viable is left to icky engineers.
Things that a physicist don't think are 'tractable' would be time-travel and completing your thesis.
To be very very clear: I am somewhat poking fun at physicists. Due to the golden age of physics, the whole field is kinda warped a bit about what they think is a doable thing/tractable. They think that a path may exist, and suddenly the problem is no longer really all that much of a problem, or rather 'real' problems are in figuring out if you can even do something at all. 'Tractable' problems are just all the 'stamp collecting' that goes into actually doing it.
The example is obvious marketing hyperbole, of course, but it's just not going to happen beyond a superficial level unless we somehow create some kind of time-travelling panopticon. It's marred by lack of data (Feynman died in 1988), bad data (hagiographies of Feynman, this instance included), flawed assumptions (would Feynman even be an appropriate teaching assistant for everyone?), etc.
I wonder if AI fans keep doing this thing in hopes that the "wow factor" of having the greats being emulated by AI (Feynman, Bill Gates, Socrates, etc.) will paper over their fundamental insecurities about their investment in AI. Like, c'mon, this kind of thing is a bit silly https://www.youtube.com/watch?v=og2ehY5QXSc
One of these doesn't quite belong ;)
But these AI researchers don't even understand these figures except as advertising reference points. The Socratic dialogue in the "sparks of AGI" paper https://arxiv.org/abs/2303.12712 has nothing whatsoever to do with Socrates or the way he argued.
Fourteen authors and not a single one seemed to realize there's any possible difference between a Socratic dialogue and a standard hack conversation where one person is named "Socrates."
Okay, that's kinda funny lol.
It's a bit worrying how much the AI industry seems to be focusing on the superficial appearance of success (grandiose marketing claims, AI art that looks fine on first glance, AI mimicking peoples' appearances and speech patterns, etc.). I'm just your random layperson in the comment section, but it really seems like the field needed to be stuck in academia for a decade or two more. It hadn't quite finished baking yet.
I would venture most industries, with foundations on other research fields, are likely the same. Oil & Gas, Pharma, manufacturing, WW2, going to the moon... the world is full of examples where people put progress or profits above safety.
It's human nature
> One of these doesn't quite belong ;)
I asked GPT to find which one:
"The one that doesn't fit in is Bill Gates.
Richard Feynman and Socrates were primarily known for their contributions to science and philosophy, respectively. Feynman was a renowned theoretical physicist, and Socrates was a foundational philosopher.
Bill Gates, on the other hand, is primarily known as a businessman and co-founder of Microsoft, a leading software corporation. While he also has made contributions to technology and philanthropy, his primary domain is different from the scientific and philosophical realms of Feynman and Socrates."
"The one that doesn't fit in is Socrates.
Richard Feynman and Bill Gates are primarily known for their contributions to science and philanthropy, respectively. Feynman was a renowned theoretical physicist, and Gates is a world-famous philanthropist.
Socrates, on the other hand, is primarily known for foundational contributions to philosophy. His primary domain is thus distinct from the scientific and philanthropic realms of Feynman and Gates."
But if you opted for the mega ultra premium gemini pro max++ model, then you get a minimal ad free experience. No wait, forget a monthly subscription. Think, micro subscription model on a per usage basis.
In your death bed and need life saving measures? Augmentation returns "402 Payment Required" before nanobots can proceed to excise the root of the issue. 5M SHIB tokens required. Unfortunately, 20G signal is not reliable in the Wilderness Zone, thus old school EMS services are dispatched to the scene. Unfortunately, EMS services do not accept your insurance and private EMS leaves the scene and dispatches public EMS services. The wait is 2 hours given the WZ.
You fail to receive life saving measures (ie, tPA) in time, thus resulting in impaired motor functions and decreased quality of life. The flashing 402 Payment Required prompt is forever imprinted in your augmentation.
Instead we just accept really bad education.
Kids everywhere will be learning nothing except how to hack these chatbots to get passing grades.
These would be charcutiers of these people with nothing but their most famous equations and examples and quips letting you know "this is supposed to be a famous person"
I think you mean caricature, but I do love the idea of Feynman making a mean charcuterie board