AI can help with content generation or scaffolding, but teaching is still a bidirectional feedback process.
When the model can’t adapt to misunderstanding or context, students immediately feel the gap. It’s a UX failure more than a “should AI be allowed” issue
It has been many years that most courses in most universities have inferior lectures than just watching a great series of YouTube videos. Many professors have no passion or training in teaching, they just want to do research. Or they have no time or pay to prepare a course. So of course they use AI slop wherever they can. Even if they record their lectures, that's almost never better than the best free ones out there.
Universities need to lean into the fact that for undergrads, they're only still good at one thing: proctured in person assessments. Also maybe community building.
Bad lectures delivered by rushed or apathetic professors is such a death march. Learning theatre.
Tangent: There's a short story I vaguely remember from when I was young about a kid being raised in a bunker after a nuclear war and the big twist was that all his friends he went to school with (virtually on the TV screen) were just AIs to keep him company. I could never find it again though, even when interrogating google or chatgpt I couldn't find it. Anyone else know of this story?
Outrageous at the least. These universities already became so commercial that they show photos of some Victorian era buildings as their campus, but most students never set foot in those buildings, as all classrooms are held in rented building outside of campus, and the main buildings are kept only as ornamental pieces.
Also, there are hardly any good teachers left. Most are hired on sharing basis, who shuttle between multiple colleges.
I just had a pretty amazing 4 hour session with gpt 5.1 going over my son's rare disease. Chat broke it all down for me in a really deep and clear way in the back and forth. Insights I've never gotten to from talking to docs, reading papers, reading bio textbooks etc.
I guess some small percentage of it was hallucinated, but if you want to call it a teacher/student relationship, it was pretty amazing.
Well the truth has finally been openly accepted by the universities themselves. They sell fancy pieces of paper with your name on them in nice calligraphy, not knowledge.
I can't imagine having so little respect for my own reputation that as a professor I'd throw out unreviewed AI slop as my own intellectual work, but I bet nobody is getting fired for it so that's just a sign of my own stupidity. A professor with no pride, working for a university with no pride, giving students with no pride certifications that they can use to get ahead in an economy with no pride.
I'm bullish on AI in education, because of the possibility of creating an individual student model that the machine can use to constantly target weaknesses in understanding. But that hasn't been invented yet. What you would get now is a teacher that hallucinates, simply lies to bridge gaps, forgets what it was supposed to be talking about, and constantly fabricates references.
In a case of the worst person you know makes a great point, Jordan Peterson was remarking a few years back about how Youtube and MOOC's were really the new universities of the modern age. If you want knowledge it's there for the taking.
The legacy institutions really are just a stamp / sorting hat for young people these days.
For the money people spend these days on education, you'd think there'd be grounds for refunds based on false advertising of the product.
This is the culmination of decades of cuts to education. I mean, what else was going to be the end point of having teachers buy supplies for their own kids, demonizing professors, demonizing higher education and the idea of education generally, not training enough teachers, and underpaying the teachers you already have.
In America we have to deal with school shootings, the latest religious group mandating the 10 commandments be put up or rainbows be taken down, irate parents mad that you failed their kid who didn't do work all semester and has severe behavioral problems no one is allowed to discipline. And now of course with AI, the students aren't doing their work, and if you call them out on it they call their parents, they sue, you get deposed and have to admit you can't 100% prove it's AI... so why bother? Who would ever want to grow up to be a teacher anymore?
So yeah, cut education, end up with AI students submitting AI papers to AI teachers. We have arrived.
The only question now is... what are we going to do about it?
Unpopular opinion, but I dislike how universities embraced vocational programs. These students were not there for academics, they were there for job and career training. The irony is that they complain about the high costs of tuition when they're just there to get more money. These professors have much better things to work on. I apologize if my opinion upsets you.
> But after a term of AI-generated slides being read, at times, by an AI voiceover, James said he had lost faith in the programme and the people running it, worrying he had “used up two years” of his life on a course that had been done “in the cheapest way possible”.
This is the future guys, get used to it.
The upside is Sam Altman will get really, really rich.
The typical student will get AI generated course content by the cheapest models.
Other children will spend $1000s/month on multimodal AI tutors spinning up Python code to check their math homework. Those students will easily surpass others without individualized support.
1) Two-Sector Economy: In Baumol and Bowen's observations, the economy is divided into two parts:
- A Progressive Sector: Productivity grows rapidly due to technology and automation (e.g., manufacturing, data processing).
- A Stagnant Sector: Productivity grows slowly, if at all, because the service is labor-intensive (e.g., a string quartet performance, a haircut, K-12 teaching).
2) Wage Linkage: Both sectors compete for labor from the same pool of workers. As productivity gains allow wages to rise in the progressive sector, the stagnant sector must also increase its wages to attract and retain employees.
3) Divergent Cost Impact:
- In the progressive sector, the higher wages are offset by the gains in productivity. The labor cost per unit of output can remain stable or even decrease.
- In the stagnant sector, there are no corresponding productivity gains to offset the higher wages. The labor cost per unit of service must therefore increase.
4) Resulting Price Trend: The prices for services in the stagnant sector (e.g., concert tickets, college tuition, healthcare) must therefore rise and faster than the prices for goods from the progressive sector (e.g., electronics, cars). https://en.wikipedia.org/wiki/Baumol_effect#/media/File:Pric...
5) A lot of European countries fund these expensive services through general taxation rather than direct user fees. In the US that's not going to fly, so cost pressure incentivizes orgs to cut corners and reduce quality and automate as much as possible.
Speaking as an instructor, everyone's goal should be to be better than the LLM. This includes teachers. And, yes, that requires work, but that's what we're getting paid for, and it's not particularly difficult to best an LLM in any field.
If you can't do it better than an LLM, you have no more value than one to an employer.
In short, it's shameful to provide a paid primarily-LLM-based product. People can just make those themselves with minimal effort and zero cost.
It is crazy to me the amount of people who are auto-completing through their education. "I don't need to learn how to code, the ai will do it for me" etc. If the only value you provide is pressing [ok] on content you don't understand, why do you think someone won't replace you with a bot? To me it's like going to pilot's school and saying you'll just have your copilot fly the plane for you.
I know at the UCs, anything you do on campus is the UCs copyright. There's even little brass plaques they put in the ground when you walk on campus delineating that line.
But when these PIs are sending all the student's data up to OpenAI (presumably via the free version), then OpenAI has that data and it's copyrights.
Likely, yet another essay on the causes of the French Revolution is not ever going to be cared about. But there is a chance that something a student does in an upper division class may be worth something.
28 comments
[ 2.8 ms ] story [ 53.9 ms ] threadUniversities need to lean into the fact that for undergrads, they're only still good at one thing: proctured in person assessments. Also maybe community building.
Bad lectures delivered by rushed or apathetic professors is such a death march. Learning theatre.
Also, there are hardly any good teachers left. Most are hired on sharing basis, who shuttle between multiple colleges.
I guess some small percentage of it was hallucinated, but if you want to call it a teacher/student relationship, it was pretty amazing.
I can't imagine having so little respect for my own reputation that as a professor I'd throw out unreviewed AI slop as my own intellectual work, but I bet nobody is getting fired for it so that's just a sign of my own stupidity. A professor with no pride, working for a university with no pride, giving students with no pride certifications that they can use to get ahead in an economy with no pride.
I'm bullish on AI in education, because of the possibility of creating an individual student model that the machine can use to constantly target weaknesses in understanding. But that hasn't been invented yet. What you would get now is a teacher that hallucinates, simply lies to bridge gaps, forgets what it was supposed to be talking about, and constantly fabricates references.
The legacy institutions really are just a stamp / sorting hat for young people these days.
For the money people spend these days on education, you'd think there'd be grounds for refunds based on false advertising of the product.
In America we have to deal with school shootings, the latest religious group mandating the 10 commandments be put up or rainbows be taken down, irate parents mad that you failed their kid who didn't do work all semester and has severe behavioral problems no one is allowed to discipline. And now of course with AI, the students aren't doing their work, and if you call them out on it they call their parents, they sue, you get deposed and have to admit you can't 100% prove it's AI... so why bother? Who would ever want to grow up to be a teacher anymore?
So yeah, cut education, end up with AI students submitting AI papers to AI teachers. We have arrived.
The only question now is... what are we going to do about it?
This is the future guys, get used to it.
The upside is Sam Altman will get really, really rich.
The typical student will get AI generated course content by the cheapest models.
Other children will spend $1000s/month on multimodal AI tutors spinning up Python code to check their math homework. Those students will easily surpass others without individualized support.
1) Two-Sector Economy: In Baumol and Bowen's observations, the economy is divided into two parts:
- A Progressive Sector: Productivity grows rapidly due to technology and automation (e.g., manufacturing, data processing).
- A Stagnant Sector: Productivity grows slowly, if at all, because the service is labor-intensive (e.g., a string quartet performance, a haircut, K-12 teaching).
2) Wage Linkage: Both sectors compete for labor from the same pool of workers. As productivity gains allow wages to rise in the progressive sector, the stagnant sector must also increase its wages to attract and retain employees.
3) Divergent Cost Impact:
- In the progressive sector, the higher wages are offset by the gains in productivity. The labor cost per unit of output can remain stable or even decrease.
- In the stagnant sector, there are no corresponding productivity gains to offset the higher wages. The labor cost per unit of service must therefore increase.
4) Resulting Price Trend: The prices for services in the stagnant sector (e.g., concert tickets, college tuition, healthcare) must therefore rise and faster than the prices for goods from the progressive sector (e.g., electronics, cars). https://en.wikipedia.org/wiki/Baumol_effect#/media/File:Pric...
5) A lot of European countries fund these expensive services through general taxation rather than direct user fees. In the US that's not going to fly, so cost pressure incentivizes orgs to cut corners and reduce quality and automate as much as possible.
If you can't do it better than an LLM, you have no more value than one to an employer.
In short, it's shameful to provide a paid primarily-LLM-based product. People can just make those themselves with minimal effort and zero cost.
I know at the UCs, anything you do on campus is the UCs copyright. There's even little brass plaques they put in the ground when you walk on campus delineating that line.
But when these PIs are sending all the student's data up to OpenAI (presumably via the free version), then OpenAI has that data and it's copyrights.
Likely, yet another essay on the causes of the French Revolution is not ever going to be cared about. But there is a chance that something a student does in an upper division class may be worth something.