- Work is shifting from building/doing to evaluating, judging, and steering — that's where human value will concentrate.
Other supporting points.
------
- No lab milestone or "RSI breakthrough" will suddenly eliminate jobs — economic impact unfolds gradually over decades.
- Reliability, not raw capability, is the real bottleneck holding back AI automation today.
- Historically, making work cheaper/faster (ATMs, radiology, coding) has grown employment, not destroyed it.
- Superintelligence claims misunderstand human intelligence, which is itself amplified by tools like AI ("co-superintelligence").
It is not a good idea to compress articles like this but there are many of these opinions to read and trying to get to the point quickly to uncover new viewpoints.
> A battle of two narratives
> Build wealth before AI obviates our skills
> Build skills, agency, taste, judgement
both narratives are portrayed as being odds with each other but, I can't come up with a single "build wealth" scenario that doesn't involve building skills, agency, taste and judgement.
But it's at my day job, and it's because I was able to write a prompt which automates having Copilot review uploaded scanned PDFs of invoices with checks (and the bank line obscured with a pen, so no PII) and then write a batch file which renames the files per a file-naming convention, removing the need to open them in batches of 50, find the Invoice ID, re-save using that filename, then quit and re-launch Adobe Acrobat (if left running, eventually I run into a bug where it stops saving files), then run a .bat file which renames based on Invoice ID as a filename.
Problem of course is I've been running into a limit of number of allowed files per 24 hr. period.
Even if it's not less work, it feels like less effort.
Extracting structured data from PDFs is something people have been doing forever, way before LLMs, there are state of the art OCR tools out there you can use as a lib, it’s hardly something that justify using a LLM
Yes it is. It's a poorly defined problem, and it needs to be correct, and existing OCR tools aren't perfect. Are the end of the day, it comes down to money and time, and throwing them at ChatGPT is cheap and fast enough to be used.
The position of (for instance) the check # on a physical check varies quite widely, as can date position and format, and a fair number of them are still written by hand.
On top of that, I'm scanning thousands of these each year, and the invoice underneath the check ranges from the gamut of: "pristine copy just re-printed 'cause the customer didn't include one" through "bad inkjet photocopy of a photograph taken w/ a potato phone and then printed" and includes variations such as "customer included half-a-dozen invoices to be paid w/ one check, and if arranging them and the check _just so_ all will fit in the document camera window, saving a trip to the sheet feeder scanner"
A co-worker actually worked on this for a different program, one where actually sending in paperwork in good condition was expected and customary, but his system had a reject/failure rate of ~10% --- that would be almost 1,000 invoices each year for the program I am handling.
I have the opposite, because I'm now getting things clients generated that they want me to implement. It's definitely more work and money for me, but it's concerning somewhat because ultimately it's not good for their business. It takes me longer to implement this kind of stuff than it would for me to code it from scratch. That is in a working way, the AI generated code has far more bells and whistles, but also layers and layers of needless complexity that quite literally add no value as in they aren't even a factor of the finished output.
The problem is they are now paying me more, plus paying for the cost of using the AI, and the needless complexity also slows down the employees. So more costs there as well, any future debugging is going to cost far more and at the end of the day they are getting less quality on the core function but far more presentation data that is essentially meaningless.
Yes, I have seen it happen on enterprise consulting.
For example, all the work like translations that used to be done by humans, now it is a CMS AI feature.
Secondly teams setup.
It used to be we did everything ourselves for development, then cloud, SaaS products and serverless decreased the teams size required for delivery.
Now with AI, there is an even greater push for low code/no code tooling, with agents, leaving the actual programming left for MCP tools that might not yet be available for the project.
Thus you get a team of five doing what used to be about 15 a decade ago.
Nope, but recently it does feel like I spend less time coding and more on prep work / requirements gathering. But that's n=1, I'm doing something very specific (building a design system with a small team of designers and developers) and recently my skill in writing requirements and doing research has been recognised.
But I'm trying to figure out if this is a thing in my organization. So far the channels discussing AI have been about how they use AI, how to build agents, and complaining they've run out of tokens - but nobody says anything about how much work they finished, how it impacted the product(s), the user, and the company's bottom line.
I don't know if they just don't know, don't care, don't measure, or they don't actually want to admit they're just burning through tokens with nothing to show for it.
That said, right now I'm using AI to extract some code from our app to create a minimal reproduction for an issue we're having with a 3rd party library, it's a huge time saver there - that is, I wouldn't have bothered making a reproduction because of how time consuming it can be.
So to answer your question, it's not that "I have less to do", it's "I do things I wouldn't have done otherwise because they cost less effort/time". Which is also the promise of e.g. automation - the automated loom didn't just reduce how long it takes to make cloth, it made it so people make a bajillion times more cloth.
I have been writing software for over 40 years and have had a long time interest in and some work in AI over that time. I wouldn’t say this gives me any more prognosticating power about how all of this is ultimately going to go, but I believe we're soon nearing an area of plateauing; whether that's because the science itself is plateauing or the intervention of governments is going to force plateau it.
So if things continue as they are today, I think in the near future, being a software developer is going to be more analogous to the medical field, where in the medical field you have different levels of professional expertise.
Some will be like nurses, and some will be closer to a medic and a smaller set will be like doctors. Each with increasingly required knowledge and experience to fulfill a needed role.
Those who used to be actual software developers are going to be (or have to become) more in the doctor role with years of internship and practical experience to be the architects guiding the overall AI implementation of software development in organizations.
The medics are going to be people who are semi-technical, where they have some technical understanding but they don't dedicate themselves to it, like say product managers, where they jump in to help development along, but don't need to have many years of experience or very deep technical knowledge.
At the nurse level, it's probably going to be similar to what people would do in the past with no code tools, where somebody in marketing who knows very little to nothing about coding at all is just going to directly converse with AI systems, but they'll never be likely to get anything more advanced than the tools they could think up for themselves.
Of course, it's so hard to tell what the next big discovery or changes to the nature of world society might push things in one direction or another.
> So if things continue as they are today, I think in the near future, being a software developer is going to be more analogous to the medical field, where in the medical field you have different levels of professional expertise.
The medical field is also going to change though. Massively. Because people are going to realize you don’t need to pay someone $400k per year to hand out advice about moderate exercise and which antibiotic is appropriate for a sneeze-cough with yellow mucus.
Regulation isn’t going to prevent this. AI is already way too easily accessible to ever rein it in again. Not to mention that the US now has serious competition from a hostile country, so they can regulate their own AIs all they want without it making a difference in practice.
> Because people are going to realize you don’t need to pay someone $400k per year to hand out advice about moderate exercise and which antibiotic is appropriate for a sneeze-cough with yellow mucus.
Who is going to realize that?
The same forces that prevent you from walking into a pharmacy and asking for antibiotics based on what you found on WebMD will prevent you from doing it with a ChatGPT printout in hand. Lawyers and doctors are the best-known examples of industries that are in control of who gets admitted to practice the profession.
It's basically a formality now to get a prescription for what you read on WebMD. Every Insurance has telemedicine, you just call, read the symptom list and get the prescription. Some even let you just email. There's a doctor or at least person with subscribing ability in the loop, but they are barely doing more than rubber stamping hundreds of requests per day.
Nothing effective about self-diagnosis. Quick and easy yes, but not effective. Most things that are worth diagnosing require a visit to a clinic.
No doctor is going to lose a job because people are self-diagnosing using the internet.
As long as doctors continue to hoard the sole ability to grant access to medications (currently a legal monopoly), they are safe from AI. Expect that to be a very, very long process. They'll continue to be the gatekeepers as long as they have the power, and they are still quite powerful.
My wife is a pediatric ER doctor. She’s board certified in pediatrics and pediatric emergency medicine. Here’s a scenario that regularly plays out.
Someone she’s knows calls her up late and says their child has a minor fever, upset stomach or some other mild symptom. They ask “should I go to the ER or can I just wait till tomorrow to go to the pediatrician?”
My wife says it’s totally fine to wait until tomorrow. That is not an emergency. And frequently she’ll tell them they don’t even need to go to their pediatrician unless it gets worse or continues for several days.
9/10 what happens is that the person goes to the ER or their pediatrician first thing in the morning and calls my wife to tell her “you were right, it wasn’t an emergency”.
Despite that my wife is more highly trained than either their pediatrician or a non-pediatric ER doctor, and they know and trust her, they still wanted to see someone in person.
It doesn’t matter how accurate AI diagnosis gets, that’s not going to change anytime soon.
That’s a good analogy, but I think the reason we currently need the "nurse" role is the need to interact with the physical world. Most software products don't require this step, so the demand for “nurses” or "doctors" will likely decrease.
If robotics technology continues to advance, the number of nurses in real world will also decrease in the foreseeable future.
The demand for nurses and doctors is only going to go up. Not just int their current roles, but everything adjacent to them as well. Everything involved with running trials of health care products. Humans aren't going away, and they need to be taken care of.
Nurses will likely be one of the last professions to be significantly impacted by robotics. Drivers, factory, warehouse, agriculture workers, cashiers etc will be automated away way before them.
It's just hard to design robots that can handle patients that may be simultaneously fragile, mentally handicapped and aggressive in a way that doesn't hurt them and respects their rights. It can take multiple human nurses to do a seemingly simple thing like changing a diaper.
I think what's actually happening is that the "threshold" of technical understanding required to be more productive than average is increasing, and other non-technical skills are becoming more important. Those below the threshold are not able to provide value over anyone else, even if they have a lot of technical experience. For example, I think today a strong PM with no technical ability can produce higher quality technical output compared to a mediocre frontend developer with no project management ability.
For example, I think today a strong PM with no technical ability can produce higher quality technical output compared to a mediocre frontend developer with no project management ability.
Wishful thinking by the managerial class. At best they can vibe code but they can’t verify that what was written is correct.
Heh. I wouldn't be so fast in calling this one, yet. It might go either way, or a combination of the two, who knows... Things are moving and progressing fast enough to at least be wary of, and keeping an eye on things. I wouldn't be surprised of any outcome, tbh.
On the one hand, PMs get to hone in a set of skills that include lots of juggling resources, bridging the gap between stake holders and people that execute, work with different layers of technical expertise, lots of back and forth and so on. Having to "call" a thing after talking to 5 people from 6 different PoVs is interestingly something that could make them quite good at using "AI". They're already used to working with "jagged expertise" so to say. And that's really close to what "agents" do today. You might get a session where claude/gpt/gemeni turns out a brilliant piece of technical artefact, or you might get an average piece of content that misses key important aspects that a "human" could see from a mile away.
On the other hand, LLMs do one thing quite well: they raise the floor of what "minimum effort" in a thing gets you. So things like language barriers, basic processes, procedures, etc. get elevated to a level where you can use these tools and be better off than not using these concepts at all. So a very talented engineer that previously had these issues, could now use LLMs to get better at auxiliary but "unimportant" aspects of integrated that technical experience into other places. In a "90% of the time people expect this to work like this" way, but with the floor being raised. One quick example of this would be a technically sound piece of software that now has bare-minimum ux/ui from this century, instead of "engineer GUI" aspect :)
Who knows where this all goes. Yesterday there was an article here about someone taking a bunch of languages, asking an agent to mix concepts from all of them and slop a "new uber language" design. Is it the ultimate language that many have tried before? 99% sure it's not. Or at least this one won't be. But there's a chance, with how many people can now prompt their way to a PoC quite quickly, that we eventually get something cool. No idea how that'd look like, but it's likely the "you'll know it when you see it" kind of thing.
Sure, but how confident can they be that it doesn't also do things they don't want?
Tests can't be exhaustive, while even a mediocre developer is likely to possess enough background knowledge to notice risks that a non-developer would not think to test.
People keep forgetting that we haven't reached AGI yet. These tools can still make serious and sometimes obvious mistakes. Not long ago, vibe-coded software could embed credentials directly. That particular blunder seems to have been addressed, but there is still no reliable way to tell an LLM to avoid every class of obvious blunder.
The un-hidden secret is that QA and real testing doesn't exist much anymore. People aren't surprised when things break, they just want it fixed quickly, and AI can do that.
Mmm... People may overlook a glitch here and there, but judging by reviews on app stores, it seems to me that people get very annoyed when things break, especially when it happens at the worst possible time.
As for AI fixing things... Sure, at least for a while. But in my experience, AI can hit a wall and start going in circles. It doesn't happen often, but it can happen. And when it does, what recourse will you have for fixing the tangled mess that vibe coding tends to produce?
P.S. Again, it all boils down to AGI not being with us yet. I guesstimate that AI writes at least 95% of my code, but given how often it comes up with inefficient or ineffective solutions, I'd never entrust my reputation to a vibe-coded app. Startups may have little to lose, but an established company?
Is that true anymore when anyone can vibe code? Seems to me that quality, correctness, and performance will be huge differentiators in a see of vibe coded slop.
Welcome to the current day difference between being an Accredited Engineer and being a Developer. The only thing that's happening is that the Developer side are getting a wake up call.
Just FYI, if you know much about the medical field, nurses tend to do most of the actual work, with highly experienced nurses actually taking up the mantle for many duties often done by doctors. Nurses are in far higher demand, than doctors.
Your analogy isn’t necessarily wrong, but it might ignore the extreme importance of nurses. Many medical facilities are only staffed with permanent nurses, with doctors helicoptering in, from time to time, to take care of specific duties that may require certain licenses, or provide specific advice.
> Many medical facilities are only staffed with permanent nurses, with doctors helicoptering in, from time to time, to take care of specific duties that may require certain licenses, or provide specific advice.
Maybe for a very loose definition of medical facilities that includes assisted living facilities.
But for example in an ER, nurses come and go with very rapid turnover and it’s common to staff with temporary travel nurses.
> nurses tend to do most of the actual work
Techs, environmental services, phlebotomists, respiratory therapists, CNAs etc. probably do more of the “work” than nurses.
> highly experienced nurses actually taking up the mantle for many duties often done by doctors
Only if they are receive extra training and become a Nurse Practitioners or an CRNA, but in that case they are no longer functioning as a nurse. But even then they are general operating under the direct supervision of a physician.
> Nurses are in far higher demand, than doctors.
Only in absolute numbers. It’s far harder to hire a doctor than it is a nurse. I know an ex-NFL player who works as a physician recruiter.
I think the only metric that matters is total medical care provided on a national / global scale.
It may take vastly more training but on average a full annual physical provides less benefit on average than a 30 second vaccination requiring minimal training. Value creation and skill are wildly different things in the medical profession.
There is no reason the medical field needs to gate keep injections. Anyone can inject, you don't need a nurse. Maybe as far as an IV with some very basic training.
Nurses may administer vaccines. (but you don’t need to be a nurse to do this. Medical assistants do it too. It’s a very easy procedure).
But nurses aren’t the driver of vaccination. Nurses are anti-vaxxers at a much higher rate than physicians, and if you removed physicians from the equation, vaccine rates would plummet.
>total medical care provided
At the end of the day a physician is responsible for every patient that comes through. Nurses are incredibly important, but all of the care they provide is at the direction of a physician, so it’s impossible to separate out “total medical care provided” into doctor vs nurse.
Yeah, I kinda feel like OP is missing the fact that it kinda already is the way they described. The majority of devs are mediocre at best and work on variations of mostly the same CRUD apps for different businesses. Only a relatively small number of people work on complicated stuff that requires CS knowledge.
There's a much bigger group of people building games with Unity than the number of people building complicated engines like Unity. Same for [insert Javascript framework here].
> Some will be like nurses, and some will be closer to a medic and a smaller set will be like doctors. Each with increasingly required knowledge and experience to fulfill a needed role.
Nursing and being a physician aren't really the same thing at all, and they require different skill sets, it's not just "having more knowledge". Just because someone is an amazing surgeon doesn't mean they would also make a good nurse.
> Those who used to be actual software developers are going to be (or have to become) more in the doctor role with years of internship and practical experience to be the architects guiding the overall AI implementation of software development in organizations.
I think you just described a staff swe
> The medics are going to be people who are semi-technical, where they have some technical understanding but they don't dedicate themselves to it, like say product managers, where they jump in to help development along, but don't need to have many years of experience or very deep technical knowledge.
These people already exist. They are the business analysts who know SQL and maybe Python, R, or VBA. Marketing people who work on Wordpress landing pages. People doing systems integration, the IT department, sales engineers, and on, and on, and on.
> At the nurse level, it's probably going to be similar to what people would do in the past with no code tools, where somebody in marketing who knows very little to nothing about coding at all is just going to directly converse with AI systems, but they'll never be likely to get anything more advanced than the tools they could think up for themselves.
You said it, no code/low-code has existed forever.
If LLMs reach a point of sufficiently high competence to the point that they're capable of producing usable software from vague, and sometimes contradictory instructions - the same software engineers have to regularly deal with, then at that point software will simply be an expectation of ability in any job.
So I think it would be more comparable to something like literacy. There was a time when that was a fairly uncommon and highly valued skill. Now the guy flipping burgers or pouring a cup of coffee is also almost certainly fully literate. And in fact many jobs have evolved in a way such that it became mandatory, but only because it was already ubiquitous. I expect to see the same thing with software. The industry of producing software that do fairly simple tasks will probably die, but in its place will be a vast array of heavily customized and oft iterated software for companies and people achieving their own stuff.
The mobile industry is a perfect example of where this will be massive shift. Right now there's a million mobile apps to execute extremely basic functionality on phones, but it's loaded with advertising, begging, and general annoyances. As are the app stores themselves. When you can make software that does that in a few minutes with a single prompt, and people realize this (as we're already practically at this point), then that will be the end of those apps. This is because the one thing LLMs have shown is that natural language interfaces are way less friction than using search, whether on the web or an app store. And so there will be a time when it will be lower friction to simply just quickly build your own app to do [whatever] than dealing with somebody trying to monetize an alarm clock.
> If LLMs reach a point of sufficiently high competence to the point that they're capable of producing usable software from vague, and sometimes contradictory instructions - the same software engineers have to regularly deal with, then at that point software will simply be an expectation of ability in any job.
I don’t understand how people can say this and then continue talking about software. So we’re saying machines can now casually do complex and cognitively demanding jobs like software development (or 90% of all white-collar jobs out there) and we’re NOT worried about the lynching mob going door to door and hanging IT people on lampposts? And I’m being serious, the impact this would have on societies would be unprecedented.
The only truly unprecedented thing is the white-collar part. Otherwise it happened many times over that entire regions where suddenly out of job and job prospects. Automation and offshoring have equivalent impacts on the affected job market.
If LLMs can handle 99% of software jobs, then they can “crack” robotics software and then we would have LLMs on robots to replace the rest of the workers
In such a future where LLMs hypothetically dominate the SDLC, the trade would be much more like pottery than medicine - i.e. a commodity in such obscene abundance that its functional utility would be entirely divorced from its creation.
In such a future there would be a handful of lucky well paid artisans, a healthy community of hobbyists, and the overwhelming population of the planet who would be perfectly content to delegate their entire software diet to generative superplatforms that script themselves to perform any arbitrary software function. The concept of individual bespoke software programs will become an anachronism for an era where software was so difficult to produce that entire teams spent years painstakingly tweaking programs to spec.
I think the big difference is that the only thing anyone ever needs a pot to do is hold fluids (or soil). It's a lot easier to spec out than a modern software product.
Yes, but this might also be a counter-point to your position. In a world of rising baselines, having a "pot but without all the bells and whistles" might be a thing people need. Instead of having a pot that holds fluids at every conceivable angle with 20 temperature thresholds, 5 versions that hold liquid in a cloud for you, with monthly subscriptions, having the ability to get a "like a pot but make it left handed and only holds sand because this is what I need" could turn out to be something that people want/need/end up chasing.
In other words spec down, not up, and base + my particular kind of a pot.
> a pot that holds fluids at every conceivable angle with 20 temperature thresholds, 5 versions that hold liquid in a cloud for you, with monthly subscriptions, having the ability to get a "like a pot but make it left handed and only holds sand because this is what I need"
Ah I see you're familiar with PowerPot by Atlassian(TM)
Very powerful potware xpity about how much soil it leaks.
This is why I mentioned the plateauing. If for some reason that just doesn't happen and things continue at a relatively linear or even exponential rate, then all bets are off.
I'd like to imagine that it's ultimately going to work out for humanity to be in a better condition overall if that happens.
However, if there reaches a level of near or full autonomy in all aspects of knowledge work, then there is a strong possibility that the field with the highest growth potential is going to be in "Private Security"; as those who have access to the most resources seek to defend their own positions in a societal return towards aristocracy and serfdom.
I think many people will wish it was like this. But when AI becomes so capable that a nobody who doesn't understands computers a well as a "doctor" uses AI to create something BETTER or superior or just as good then employers will think... why am I paying that "doctor" more than that nobody?
See right now everything is fine, because AI is not that good yet. But if it gets better. Well. The future will be one where we trust AI more and more.
Current discussions is like deja vu. One time there was hype around of no code/low code development using PowerApps, Zapier etc.
The sell was - Creating your apps was easier than before. You don't need to know any programming language. You don't need IT. You have a GUI and just drag and drop and visual your apps. Just fill the fields and voila your app is ready.
The criticism of the approach was that low codes apps might not be secure, unsuitable for large scale and critical apps, and lead to increase in unsupported apps by "shadow IT".
The same reasoning exists today for AI coding. Anyone can create apps, its not great for complex and mission critical apps, might not be secure etc. And lots of discussion like your post parsing the future too closely.
Low/no code apps have continue to grow. But many of the low/no code tool users are developers who use it to make their jobs easier.
While some might say - this time it is different. I believe we are currently we are the beginning of the cycle so everyone is excited to use their PowerApps shaped Claude Code/Codex to make their 100th budget tracking app but as time passes and edge cases are figured out the biggest users for AI are going to be software developers (and I believe they currently are the biggest users).
As for software engineering jobs just like before engineers will be expected to output more and faster. This has happened with every innovation from assembly compiler to IDE to low/no code ways of building.
The different levels of expertise exists even today and it will remain so.
i don't think it's the same exact conversation. with low code tools you still need to poke around, figure out what you want, what to do, etc.
with the latest agents, you can even be vague and they'll probably do an ok job if it's not something super complicated. some models are also pretty good at stopping to let you know about stuff you haven't thought about and ask for clarification.
so you can outsource thinking now, not just the doing part.
> you still need to poke around, figure out what you want, what to do, etc.
Is it different? because then:
> you can even be vague, and they'll probably do an ok job if it's not something super complicated
That's lot of qualifiers to say maybe, should be possible - how does someone who is not a software engineer know what is complex, vague or ok. There are lot of assumptions to say - This time it is different.
It is entirely possible it is different but I am yet to find a convincing argument.
The answer to this question after a lot of reflection: games.
AI can slop fork or clone existing software well, but a clone of an existing game is pointless, it's basically guaranteed to be derivative and worse than the original game, and games aren't so expensive that you can't just buy the original. AI can't know if new mechanics or angles to an existing genre will feel good to play, or if a new genre is fun, that requires a human to experience the game in its totality.
Games are also very resilient to sloppy AI coding, and if an indy game crashes nobody is getting paged.
> Games are also very resilient to sloppy AI coding
Depends on what kind of game you are building. I can tell you that even Claude Opus absolutely hates desync bugs and has a rather hard tracing them. Maybe Fable or chatgpt-5.6 are better?
Indy games markets have been flooded with human slop since long before AI was a thing, the marketing playbook hasn't changed at all, if you just fire games into the void you never had a shot.
Why is it guaranteed to be worse? A motivated fan of a game took issue with some part of the game and made their own one that didn't have that issue. Aka made it better in their eyes.
Games are complex systems, you actually have to playtest them to observe whether the changes you made increase or decrease the fun, despite your best guess. A lot of times things that seem like they should be fun actually aren't for reasons you couldn't predict. This is what makes them immune to being completely automated.
Remember when they created "COBOL" so that everyone could write programs? This is just round two.
If you think it is different, just think of how many people write books professionally, or even publish online.
Once the noise settles down a bit and boardroom shakes off their delusions as you can see in rehiring in Ford and Zuck who was very bull on AI remark about "not being it". It will be just the same, but different.
I like this article because it seems to go into decent depth on the “framework” that the author comes up with.
However, this following quote has a simple reason that I don’t see anywhere in the article or framework:
“””
Why is there a huge gap between what people in various occupations could be using AI for and what they’re actually using it for? One reason could be that people are slow to adopt technology, and that’s certainly part of our framework.
“””
I would like to add a reason: that the Silicon Valley companies who developed the LLMs are brigands: cognizant of their actions, they have stolen (and continue to steal) the world’s copyrighted material and are selling it back to the masses and the politicians as if they are the arbiters of information itself.
Specifically responding to the quoted question, I could be using Claude or ChatGPT or Grok or DeepSeek or any other to have come up with this comment, or to write emails, or to implement my Python for me, etc., but I use none of them for anything. Doing business with brigands is a choice, and a choice that I hope becomes less and less palatable so that the financial, political, social, and moral fever that is our zeitgeist finally breaks.
The other issue is Gen Z and Gen A are now very much opposed to AI. I'm wondering with those two sets of generations who already have a very negative view of AI, how AI can survive that coming tsunami of change.
According to WRITER’s 2026 Enterprise Adoption Survey, 44% of Gen Z employees admit to sabotaging their company's AI strategy in at least one way compared to 29% of employees overall.
Sabotage behaviours include entering proprietary information into AI tools, using non-approved AI tools, refusing to use AI tools or outputs, ignoring guidelines or best practices, intentionally generating low-quality outputs, refusing to take AI training and tampering with performance metrics to make AI appear to underperform.
Do "entering proprietary information into AI tools", "using non-approved AI tools", and "ignoring guidelines and best practices" really count as sabotaging an AI strategy because you are opposed to it? An over-enthusiastic early adopter would do all three.
Ironically, I think AI boosters are probably turning into AI's biggest obstacle right now. I see so much "adapt or die!" bullshit here (people literally saying that!), and yet when I talk to other professionals I know in real life everybody is just kind of sick of all this. It's utterly exhausting and the AI industry is looking and acting more and more desperate. They've basically put themselves in a stupid position where it has to replace everyone's job for the investment to even make sense, and it's just not happening. I'm longing for the day when the investment money runs out and the grifters go grift on something else.
The utility of AI is too high for the opposition to be real. It's like money or heroin. Someone who says money is the root of all evil will likely take a billion dollars if offered if for free.
Everybody says they hate heroin but once you try it, you can't get enough of it.
Are they really opposed to AI or are they just opposed to anything that old people do/like, just like every generation? The moment they get their own AI that us dinosaurs don't understand they'll be all over it.
Yeah I hear that all the time "someone needs to clean up the slop tech debt" yet in the before times there was plenty of tech debt no one ever cleaned up, why would that be different now.
Come on now: we translate vague ambitions into communications for non-living entities to do human bidding. Until we have recreated humanity as mythic gawds, there is a ton of work to do.
Mr. Narayanan seems to be trying to be a bit more positive than the vibe I am getting off of his presentation. Or maybe it is just sort of meant to make us all experience our shoved-down anxiety about being phased out with nowhere else to go. A lot of adaptation to do sounds not so fun. So I kind of think that is not a terrible point, if true.
I sort of worry about things like AI figuring out scripts so well that even multi-tier support work is gone. And learning how to write fiction or create foods so in accordance to our tastes (sugar, fat, etc with food, exactly what each of us is interested in, with writing) that we even lose those truly human creative jobs. Might not ever wanna leave those bubbles.
So much of the human drive is exploration and why and what if. Assuming everyone in the world can have no money problems, what will AI not be able to figure out?
AI could be a horrible jailor. And better at cancelling than any perhaps sager Gen Z or millenial. Bears some caution to be wary of this and where that power sinkhole will go.
But then, I still think the previous AI winters were more a result of sense and caution than most of us know, and we cannot fathom our species' ways of reasoning/thought processes the way we did as a species thirty, fifty, eighty years ago. Erring on the side of caution is not a terrible thing.
If we were more connected to all the problems that exist in the world, we’d become acutely aware of just how much work there is to do, and we’d eagerly reach for any tool that could help us do more, faster.
Exactly. For solving problems every help is a blessing. It's just clashing with our legacy society system. This system forces us to have jobs. But the real problems humanity is facing, like planetary boundaries or a decent base standard of living for all humans, are completely unimpressed by that human made model.
AI will also make us faster as exploitation, genocide, and weapons manufacturing.
Our technology has long since surpassed our capacity to understand its impact or to morally restrain its use - we've gone from chivalry to "oops, that was a school, oh well"
We will solve problems faster though, but we aren't showing signs that AI is making us better humans, and that is the problem.
If AI is used to solve the socialist calculation problem, most of you will die.
If AI is subject to private ownership in a competitive market between competing suppliers, it will be like better cars, we’ll just drive faster.
Power consumption will be a limiting factor in those countries relying on intermittent, weather dependent power generation with no base load. Especially if users prefer Apple’s privacy first AI on edge devices.
Hopefully in western countries it can encourage more young women to bear three children before they turn 35. Young men have to pick up their game and create an environment to redirect their suicidal empathy into more productive pursuits.
“Where can you find another non-linear servo-mechanism weighing only 150 pounds and having great adaptability, that can be produced so cheaply by completely un-skilled labour?” - Albert Crossfield 1954
You need to first address 1)What is work? 2)Why we need to work?
Animals don't "work". Not atleast for their own sake. If there is enough green pasture and water around, they don't even migrate to other places. So if work is meant to provide food and shelter and if machines can ensure that, humans don't need to "work".
Wealth is only a reserve capacity to help future generations so that they don't need to work for their basic needs. But if machines ensure that too, then wealth itself, as a reserve, is unnecessary.
This. Before we worry about how much work there is left, we have to define what work is "to be done." We're already at a point where an incredible amount of work done isn't strictly "necessary." It's not growing food, it's not making clothing, it's not building houses, or providing other basic needs and comforts...
How many man-hours go into various parts of the advertising distribution chain? Though a certain fraction of that energy goes to connecting people with goods and services they might find valuable, most of it goes into shifting numbers around for people that already don't personally have to worry about money.
We don't need to find endless ways for people to spin wheels, but as long as we're worried about "jobs," we will. We just need to find the social structures to provide people with basic needs and reserve "work" for things that are vital to society or truly inspired.
> We just need to find the social structures to provide people with basic needs and reserve "work" for things that are vital to society or truly inspired.
> How many man-hours go into various parts of the advertising distribution chain?
Not that much. AI is already heavily automating advertising and has been for a long time. And a lot of activity that was previously exposed to untargeted passive advertising - like TV - has shifted to mediums that don't have any, like Netflix (for most subscribers).
And as you note, advertising isn't useless. It's how people find out about things they didn't know they wanted.
Alphabet and Meta, two of the largest firms to ever exist by market capitalization, revenues and profits, are largely sustained by their online advertisement products.
I don't think your assessment that this is mostly automated stands in the face of the number of man-hours both of these companies purchase each year.
This is a very short-sighted, and exceedingly common, take.
Until the machines aren't owned by anyone (or owned by everyone, take your pick on the phrasing), the owners of the machine have no need to keep you alive.
This take is basically "Don't worry, people like Sam Altman are looking out for us"...
Disparity in ownership of machines is not the main factor that is driving the need for work. It is the un-ending desire (or selling pressure) to have things that require money to buy. Most people work to be able to pay their loans and have things that are perceived to be common needs in their geography and culture.
These "needs" are sometimes enforced by the systems and government so that people don't stay away from the work and "economy" keeps churning. The housing prices could be a way to keep the people working for loan payments.
Instant foods, nursing homes for elderly, creches, roads, commuter trains - are all ways to have more workers and make them focused on work.
Okay, lets say "the environment". The universe does not care whether you live or die, so why are you so sure that someone will paternalistically care for you?
On what are you basing this viewpoint on?
> Machine owners don't have the power, for example.
That's because the world doesn't care about them either. So why would the world care about you?
I don't buy this. Tons of people in the West live on welfare and don't contribute shit to the economy, they are net takers. No one is plotting to kill them, in fact they're (arguably) getting more than they did 50 years ago and definitely more a 100 years ago.
We live in democracies, not in a dystopian nightmare and I don't see why A.I is going to change that dynamic.
People vote, people control the government, the government controls the armed forces, private citizens are not allowed to gather unregulated weapons... billionaire or no billionaire I don't see how you can beat that.
And also, it's not like the billionaire class is some kind of a cohesive group that wants to work together to rule the world - they pretty much compete with and hate each other. I don't think Musk, Altman, Hassabis and Larry and Sergey are all going to agree to work together on "controling the machines" and killing all the other people.
> I don't buy this. Tons of people in the West live on welfare and don't contribute shit to the economy, they are net takers. No one is plotting to kill them, in fact they're (arguably) getting more than they did 50 years ago and definitely more a 100 years ago.
And people had to fight for it. Through voting, yes, but also through strikes, sometimes riots, and some died for it. People all the world fought and died for the right to strike (basically, the right to refuse to work). For the right to organise as groups. For the right to only work 8 hours a day, for the right to a have a weekly day off. For the right to pay for pensions. For the right to have some sort of health insurance.
All over the world. Some still do. Many still don’t have half those rights.
Now, I don’t know about any billionaires specifically. But welfare didn’t fall from the sky. Or anyone’s good grace.
If in doubt, looking back at the late 1800s and early 1900s work and life conditions across the Western world is enlightening.
Look at Civilian Conservation Corps, a very interesting solution to all of it, especially that there is a transition period for full automation. Perhaps our job will be to "care" for the outcome in the real world that affects people.
Of course, billionaires have other plans, and are the main obstacle in achieving any sort of social cohesion.
Sure. But either way we aren’t going to have to worry about it. We’ll have post scarcity utopia, or we’ll have died. But there is an argument that the billionaire need to pay attention to: an ASI that kills 7 billion people won’t mind killing a few more.
What happens in the space between when we think we have reached an equilibrium and an eventual incipient utopia, and whenever 9/10ths or whatever of the population is dead? What will AIs think of our fitness, anyway? Are we training them to be non-judgmental (and how is that safer or less safe?) and enforcing some weird idea of "equality"? Or are we teaching AIs to choose our choices for us, and our survivors, down the line? What will people be like when that time comes, anyway?
But this assumes the ultimate "want" is "stuff" like houses, cars etc. But that's all just a means to an end or at least secondary to the real want, which is sex. For now at least, that requires people.
>owners of the machine have no need to keep you alive
And what reason do they have for killing everyone else? Where is this abject nonsense even coming from? You can't just assert that the "owners of the machine" are all cartoonishly evil for some unknown reason.
All else being equal, at least Sam Altman et al. aren't constantly making up fantasies about exterminating people.
> And what reason do they have for killing everyone else?
Who said they would kill everyone else? Since the rest of your strawmanned caricatured argument is a response to an argument never made, there's no point in addressing it.
Nobody's saying they'd actively round up and kill everyone, they would just have no reason to keep the "unnecessary" people alive. The elite class is generally in favor of doing things that worsen human suffering and cause additional deaths if doing so helps their financial position. See the recent cuts to SNAP and Medicaid for an example. If large swaths of the population got laid off from their jobs because they were no longer economically necessary, what incentive would the people benefiting from AI have to make a little less money in exchange for funding programs to keep the former white collar workers fed and healthy?
Agree on this. Since we have this wonderful system called capitalism there are tendencies for, relatively speaking, a few human beings hoard / control most of the productive capacity and as such wealth.
Some may consider local LLMs to help with that as the power of LLMs would be more distributed. I think local LLMs would help marginally. Companies as an entity would still be better positioned to use these to their advantage vis a vis individuals.
So projecting into the future I still wonder if it won't become more challenging for individuals to make a living on average.
But animals and humans are different. Humans are an animal but to suggest a lion, dog, fish etc operate in the same manner seems a little odd to me.
My dog was perfectly happy to snooze most of the day, play with toys occasionally, go for walks and try to hunt rodents. I on the other hand would be incredibly bored with that.
Only if you buy the fairy tale that the machines will be our nursemaids and that the powerful will forget their obsession with controlling and exploiting others.
> the powerful will forget their obsession with controlling and exploiting others
I do find this line of reasoning overly simplistic
The drive some individuals have to control / exploit others isn't limited to the rich, they're just in a better position to exercise it
There are people who have absolutely no interest in what other people around them are doing; there are others who will move heaven and earth to help those around them whenever they can, even at the expense of their own interests; others still who can't help but compare themselves to everyone and make it everyone else's problem if they find themselves lacking; and yet others who at all costs want people to afford them "respect" which when you drill into it means stroking their ego
I'm sure we've all experienced petty individuals who've been granted just enough power that they can gain personal satisfaction from leveraging it to make themselves feel better at the expense of all other considerations
These are all just some of the many flavours of humanity
Since you keep being the top comment:
I agree with your philosophy, but that is not how Biology often works.
When enough green pastures are around animals usually reproduce till that is no longer the case and they need to move to other pastures. They continue doing this untill all green pastures are occupied. After this they start competing with one another to compete for the green pastures already occupied.
Some animals will be so succesfull that they take larger green pastures, letting others starve. If by some miraclelous event (ai?) suddenly a lot of green pastures arise, animals will simply reproduce again till they occupied all green pastures again.
Looking at us?
The agraric revolution, neither the Industrial revolution, nor this ai revolution have decreased how much we work on our 'jobs' (finding food indirectly).
All it did was increasing the population so that all green pastures became occupied.
He's basically saying that even though AI capability is high and rapidly increasing, it is not reliable, creative or tasteful enough to replace humans. Further he implies that it will take decades before this is the case.
But we already do have have some kind of measurement of most of these types of side factors, and they actually aren't at zero and are increasing rapidly. So the implication that they will not be human level until decades from now is just (hopeful?) speculation or fuzzy thinking.
To me this looks like a really academic and official sounding version of the same quasi-religious hopium that usually defends the sanctity of the human. He is essentially saying that there is just something so special about humans that it will never be reproduced in a machine. It's very similar to dualism (and in many people actually is religious dualism). No AI is going to have human creativity or judgement. Not anytime soon. Why? Well, we all just _know_ that's not possible. Okay, maybe in a couple of decades (but they don't necessarily believe that anyway). Why would that take decades? Well we all can just _tell_ it's no where close, right? Because AI of today just isn't special like humans.
Aside from that worldview issue, I think that people still are not taking seriously or internalizing the concept of exponential improvement.
Computing efficiency gains can actually level off. In fact, they have many, many times before. But they always tilt back up again when we invent the next approach to get beyond the current level. This is how it has been for 90 years.
There are multiple ways that we continue to see huge gains in AI software, architecture, and hardware. There are huge efficiency gains available still as we move towards more radical fully compute in memory and/or analog approaches and other options like models implemented in hardware.
It’s already the case that I’m no longer paid to type source code into a computer, but rather to control agents that do that. There’s still plenty of demand for human expertise and labor. It’s possible that this will change as well. What gives me hope that I won’t be completely useless in the future is Marx’s labor theory of value, which states that the value of a commodity is determined by the amount of human labor time invested in it. His reasoning as to why this is the case makes sense to me, even though other economists argue against it. Marx also argues that technological progress, which offers a market advantage here and there in the short term, tends to become widespread in the medium and long term due to competitive pressure, so that in the end, all that remains is human labor time. Seen this way, it doesn’t matter how much better AI becomes. In the end, human labor always floats on top of it like a layer of fat on soup. This does not rule out the possibility that the human labor that remains will be shit, but preventing that is a matter of political action.
No, of course, you can only ever be certain of that in hindsight, through a historical lens. We’re not there yet. The question is simply whether Marx’s analysis was abstract and general enough at that point to still hold up in our present day, despite the changed circumstances. It’s surprisingly common for Marx’s analysis to become even more accurate later on than it was in his own time. But I can’t say whether anything has actually changed so fundamentally that it violates the underlying assumptions of his theory.
I just don't see that borne out in the real world though.
My mother worked in an office in London as a shorthand typist in the 60s, along with thousands of other young women.
At some point computers began to enter offices, bosses typed their own letters and then emails, and this category of job simply evaporated. Of course there was still SOME secretarial work, and some workers retrained to do it, but most simply had to leave the sector.
Isn't that a common story with technology replacing workers?
Unfortunately, yes. One real-world example was the former Linotype typesetters. As far as I know, they had union-negotiated benefits such as job security until retirement, early retirement, or generous severance packages. It just requires the appropriate political organization. It doesn’t happen on its own.
I have been coding in one form or another since the 80s. Back then it was 68k assembly for me. People seem to forget that "software development" is really a big bloated abstraction between "I want or need" and the product that serves that want or need. It's hard to write and debug in ones and zeros so we invented languages. Software projects have been difficult to schedule so we invented all kinds of ceremony. What will be left for us to work on? The same stuff you already work on! People who write this drivel don't seem to actually be directly engaging with the tools to understand what is what. The reality at large companies right now, based on my experience at least, is that the product people who always cared about the product and the user want to move fast and skip the bloated inflamed stuff that stood in their way (i.e. "software development shrouded in magical mysticism reserved for the D&D introverted crowd"). And then, on the other side, the AI luddites try to hold onto SOPs from twenty years ago because they feel like that magical veil is being pulled up and they no longer have a smoke screen to hide behind. Back when we had boxed software that retailed for $1k a box, "professionals" used to freak out about some kid in Bangladesh pirating the software and thus immediately and directly stealing their jobs. I never lost a gig to a kid with pirated software. I won't lose a gig to AI either.
I think this is an important article that provides a framework how to think and navigate what’s happening.
What upset me a bit were phrases like “This is not a slogan. It’s a framework” which immediately devalued the work for me.
I have read so much Ai generated text recently, that I developed some AI-fatigue or AI-burnout, and I’m wondering if that might hit more fields - making more humans reject Ai work.
To be clear, I still like the text and I don’t know if it was written (partially) by Ai or not - but it’s this uncanny feeling I got reading it.
I am starting to seriously consider switching my pi-hole from blacklist to whitelist; block everything except few domains. I am convinced now most of the text I read is either ai generated, or people who wrote it have been influenced by AI so much that they write like it.
I cant read this shit anymore.
"creating cross functional AI evaluation team to keep the company honest"
Fucking garbage.
I have an old book between me and the keyboard, and just read it while the AI is thinking, so I can read non AI words.. otherwise its like reading books by the same author over and over and over again.
If it is a common AI pattern, I wonder why the developers haven't taken measures to circumvent tells like that. Unless they keep them there on purpose as like a signature or watermark.
210 comments
[ 2.9 ms ] story [ 60.0 ms ] thread- Work is shifting from building/doing to evaluating, judging, and steering — that's where human value will concentrate.
Other supporting points. ------
- No lab milestone or "RSI breakthrough" will suddenly eliminate jobs — economic impact unfolds gradually over decades.
- Reliability, not raw capability, is the real bottleneck holding back AI automation today.
- Historically, making work cheaper/faster (ATMs, radiology, coding) has grown employment, not destroyed it.
- Superintelligence claims misunderstand human intelligence, which is itself amplified by tools like AI ("co-superintelligence").
It is not a good idea to compress articles like this but there are many of these opinions to read and trying to get to the point quickly to uncover new viewpoints.
> A battle of two narratives > Build wealth before AI obviates our skills > Build skills, agency, taste, judgement
both narratives are portrayed as being odds with each other but, I can't come up with a single "build wealth" scenario that doesn't involve building skills, agency, taste and judgement.
what am I missing ?
I would doubt however that this would be an 'Equals' or 'Implies' scenario. Let go of seeing either of them as binary, and then not even as scalers.
But it's at my day job, and it's because I was able to write a prompt which automates having Copilot review uploaded scanned PDFs of invoices with checks (and the bank line obscured with a pen, so no PII) and then write a batch file which renames the files per a file-naming convention, removing the need to open them in batches of 50, find the Invoice ID, re-save using that filename, then quit and re-launch Adobe Acrobat (if left running, eventually I run into a bug where it stops saving files), then run a .bat file which renames based on Invoice ID as a filename.
Problem of course is I've been running into a limit of number of allowed files per 24 hr. period.
Even if it's not less work, it feels like less effort.
The position of (for instance) the check # on a physical check varies quite widely, as can date position and format, and a fair number of them are still written by hand.
On top of that, I'm scanning thousands of these each year, and the invoice underneath the check ranges from the gamut of: "pristine copy just re-printed 'cause the customer didn't include one" through "bad inkjet photocopy of a photograph taken w/ a potato phone and then printed" and includes variations such as "customer included half-a-dozen invoices to be paid w/ one check, and if arranging them and the check _just so_ all will fit in the document camera window, saving a trip to the sheet feeder scanner"
A co-worker actually worked on this for a different program, one where actually sending in paperwork in good condition was expected and customary, but his system had a reject/failure rate of ~10% --- that would be almost 1,000 invoices each year for the program I am handling.
> It’s funny. I was looking at my GH activity graph. It’s been pretty solid green, for years. I stay busy.
> But since I’ve been using an LLM, it’s been bright green.
> I always check in code manually. I don’t let the LLM do it.
[0] https://news.ycombinator.com/item?id=48843115
The problem is they are now paying me more, plus paying for the cost of using the AI, and the needless complexity also slows down the employees. So more costs there as well, any future debugging is going to cost far more and at the end of the day they are getting less quality on the core function but far more presentation data that is essentially meaningless.
I didn't do fewer hours in these weeks but had time to explore and innovate a little.
For example, all the work like translations that used to be done by humans, now it is a CMS AI feature.
Secondly teams setup.
It used to be we did everything ourselves for development, then cloud, SaaS products and serverless decreased the teams size required for delivery.
Now with AI, there is an even greater push for low code/no code tooling, with agents, leaving the actual programming left for MCP tools that might not yet be available for the project.
Thus you get a team of five doing what used to be about 15 a decade ago.
I can now write software quicker in most of the cases, but the rest of the organization moves as slowly as ever.
It pushes the bulk of the work to review. So teams with good practices can account for this.
For junior teams, the time saved is massive, because they aren't doing all the other practices required to prevent technical debt.
But I'm trying to figure out if this is a thing in my organization. So far the channels discussing AI have been about how they use AI, how to build agents, and complaining they've run out of tokens - but nobody says anything about how much work they finished, how it impacted the product(s), the user, and the company's bottom line.
I don't know if they just don't know, don't care, don't measure, or they don't actually want to admit they're just burning through tokens with nothing to show for it.
That said, right now I'm using AI to extract some code from our app to create a minimal reproduction for an issue we're having with a 3rd party library, it's a huge time saver there - that is, I wouldn't have bothered making a reproduction because of how time consuming it can be.
So to answer your question, it's not that "I have less to do", it's "I do things I wouldn't have done otherwise because they cost less effort/time". Which is also the promise of e.g. automation - the automated loom didn't just reduce how long it takes to make cloth, it made it so people make a bajillion times more cloth.
So if things continue as they are today, I think in the near future, being a software developer is going to be more analogous to the medical field, where in the medical field you have different levels of professional expertise.
Some will be like nurses, and some will be closer to a medic and a smaller set will be like doctors. Each with increasingly required knowledge and experience to fulfill a needed role.
Those who used to be actual software developers are going to be (or have to become) more in the doctor role with years of internship and practical experience to be the architects guiding the overall AI implementation of software development in organizations.
The medics are going to be people who are semi-technical, where they have some technical understanding but they don't dedicate themselves to it, like say product managers, where they jump in to help development along, but don't need to have many years of experience or very deep technical knowledge.
At the nurse level, it's probably going to be similar to what people would do in the past with no code tools, where somebody in marketing who knows very little to nothing about coding at all is just going to directly converse with AI systems, but they'll never be likely to get anything more advanced than the tools they could think up for themselves.
Of course, it's so hard to tell what the next big discovery or changes to the nature of world society might push things in one direction or another.
The medical field is also going to change though. Massively. Because people are going to realize you don’t need to pay someone $400k per year to hand out advice about moderate exercise and which antibiotic is appropriate for a sneeze-cough with yellow mucus.
Regulation isn’t going to prevent this. AI is already way too easily accessible to ever rein it in again. Not to mention that the US now has serious competition from a hostile country, so they can regulate their own AIs all they want without it making a difference in practice.
Who is going to realize that?
The same forces that prevent you from walking into a pharmacy and asking for antibiotics based on what you found on WebMD will prevent you from doing it with a ChatGPT printout in hand. Lawyers and doctors are the best-known examples of industries that are in control of who gets admitted to practice the profession.
Mostly for drugs that should probably be OTC anyway. Try doing that for a narcotic, or for an expensive biologic.
You could always lie to a doctor about your symptoms.
Someone she’s knows calls her up late and says their child has a minor fever, upset stomach or some other mild symptom. They ask “should I go to the ER or can I just wait till tomorrow to go to the pediatrician?”
My wife says it’s totally fine to wait until tomorrow. That is not an emergency. And frequently she’ll tell them they don’t even need to go to their pediatrician unless it gets worse or continues for several days.
9/10 what happens is that the person goes to the ER or their pediatrician first thing in the morning and calls my wife to tell her “you were right, it wasn’t an emergency”.
Despite that my wife is more highly trained than either their pediatrician or a non-pediatric ER doctor, and they know and trust her, they still wanted to see someone in person.
It doesn’t matter how accurate AI diagnosis gets, that’s not going to change anytime soon.
It's just hard to design robots that can handle patients that may be simultaneously fragile, mentally handicapped and aggressive in a way that doesn't hurt them and respects their rights. It can take multiple human nurses to do a seemingly simple thing like changing a diaper.
Wishful thinking by the managerial class. At best they can vibe code but they can’t verify that what was written is correct.
Heh. I wouldn't be so fast in calling this one, yet. It might go either way, or a combination of the two, who knows... Things are moving and progressing fast enough to at least be wary of, and keeping an eye on things. I wouldn't be surprised of any outcome, tbh.
On the one hand, PMs get to hone in a set of skills that include lots of juggling resources, bridging the gap between stake holders and people that execute, work with different layers of technical expertise, lots of back and forth and so on. Having to "call" a thing after talking to 5 people from 6 different PoVs is interestingly something that could make them quite good at using "AI". They're already used to working with "jagged expertise" so to say. And that's really close to what "agents" do today. You might get a session where claude/gpt/gemeni turns out a brilliant piece of technical artefact, or you might get an average piece of content that misses key important aspects that a "human" could see from a mile away.
On the other hand, LLMs do one thing quite well: they raise the floor of what "minimum effort" in a thing gets you. So things like language barriers, basic processes, procedures, etc. get elevated to a level where you can use these tools and be better off than not using these concepts at all. So a very talented engineer that previously had these issues, could now use LLMs to get better at auxiliary but "unimportant" aspects of integrated that technical experience into other places. In a "90% of the time people expect this to work like this" way, but with the floor being raised. One quick example of this would be a technically sound piece of software that now has bare-minimum ux/ui from this century, instead of "engineer GUI" aspect :)
Who knows where this all goes. Yesterday there was an article here about someone taking a bunch of languages, asking an agent to mix concepts from all of them and slop a "new uber language" design. Is it the ultimate language that many have tried before? 99% sure it's not. Or at least this one won't be. But there's a chance, with how many people can now prompt their way to a PoC quite quickly, that we eventually get something cool. No idea how that'd look like, but it's likely the "you'll know it when you see it" kind of thing.
Tests can't be exhaustive, while even a mediocre developer is likely to possess enough background knowledge to notice risks that a non-developer would not think to test.
People keep forgetting that we haven't reached AGI yet. These tools can still make serious and sometimes obvious mistakes. Not long ago, vibe-coded software could embed credentials directly. That particular blunder seems to have been addressed, but there is still no reliable way to tell an LLM to avoid every class of obvious blunder.
As for AI fixing things... Sure, at least for a while. But in my experience, AI can hit a wall and start going in circles. It doesn't happen often, but it can happen. And when it does, what recourse will you have for fixing the tangled mess that vibe coding tends to produce?
Is that true anymore when anyone can vibe code? Seems to me that quality, correctness, and performance will be huge differentiators in a see of vibe coded slop.
Your analogy isn’t necessarily wrong, but it might ignore the extreme importance of nurses. Many medical facilities are only staffed with permanent nurses, with doctors helicoptering in, from time to time, to take care of specific duties that may require certain licenses, or provide specific advice.
So lots of jobs for nurses.
Maybe for a very loose definition of medical facilities that includes assisted living facilities.
But for example in an ER, nurses come and go with very rapid turnover and it’s common to staff with temporary travel nurses.
> nurses tend to do most of the actual work
Techs, environmental services, phlebotomists, respiratory therapists, CNAs etc. probably do more of the “work” than nurses.
> highly experienced nurses actually taking up the mantle for many duties often done by doctors
Only if they are receive extra training and become a Nurse Practitioners or an CRNA, but in that case they are no longer functioning as a nurse. But even then they are general operating under the direct supervision of a physician.
> Nurses are in far higher demand, than doctors.
Only in absolute numbers. It’s far harder to hire a doctor than it is a nurse. I know an ex-NFL player who works as a physician recruiter.
It may take vastly more training but on average a full annual physical provides less benefit on average than a 30 second vaccination requiring minimal training. Value creation and skill are wildly different things in the medical profession.
Healthcare is a world of diminishing returns at ever increasing costs. But that doesn’t mean the low effort first steps are less valuable.
But nurses aren’t the driver of vaccination. Nurses are anti-vaxxers at a much higher rate than physicians, and if you removed physicians from the equation, vaccine rates would plummet.
>total medical care provided
At the end of the day a physician is responsible for every patient that comes through. Nurses are incredibly important, but all of the care they provide is at the direction of a physician, so it’s impossible to separate out “total medical care provided” into doctor vs nurse.
There's a much bigger group of people building games with Unity than the number of people building complicated engines like Unity. Same for [insert Javascript framework here].
> Some will be like nurses, and some will be closer to a medic and a smaller set will be like doctors. Each with increasingly required knowledge and experience to fulfill a needed role.
Nursing and being a physician aren't really the same thing at all, and they require different skill sets, it's not just "having more knowledge". Just because someone is an amazing surgeon doesn't mean they would also make a good nurse.
> Those who used to be actual software developers are going to be (or have to become) more in the doctor role with years of internship and practical experience to be the architects guiding the overall AI implementation of software development in organizations.
I think you just described a staff swe
> The medics are going to be people who are semi-technical, where they have some technical understanding but they don't dedicate themselves to it, like say product managers, where they jump in to help development along, but don't need to have many years of experience or very deep technical knowledge.
These people already exist. They are the business analysts who know SQL and maybe Python, R, or VBA. Marketing people who work on Wordpress landing pages. People doing systems integration, the IT department, sales engineers, and on, and on, and on.
> At the nurse level, it's probably going to be similar to what people would do in the past with no code tools, where somebody in marketing who knows very little to nothing about coding at all is just going to directly converse with AI systems, but they'll never be likely to get anything more advanced than the tools they could think up for themselves.
You said it, no code/low-code has existed forever.
So I think it would be more comparable to something like literacy. There was a time when that was a fairly uncommon and highly valued skill. Now the guy flipping burgers or pouring a cup of coffee is also almost certainly fully literate. And in fact many jobs have evolved in a way such that it became mandatory, but only because it was already ubiquitous. I expect to see the same thing with software. The industry of producing software that do fairly simple tasks will probably die, but in its place will be a vast array of heavily customized and oft iterated software for companies and people achieving their own stuff.
The mobile industry is a perfect example of where this will be massive shift. Right now there's a million mobile apps to execute extremely basic functionality on phones, but it's loaded with advertising, begging, and general annoyances. As are the app stores themselves. When you can make software that does that in a few minutes with a single prompt, and people realize this (as we're already practically at this point), then that will be the end of those apps. This is because the one thing LLMs have shown is that natural language interfaces are way less friction than using search, whether on the web or an app store. And so there will be a time when it will be lower friction to simply just quickly build your own app to do [whatever] than dealing with somebody trying to monetize an alarm clock.
I don’t understand how people can say this and then continue talking about software. So we’re saying machines can now casually do complex and cognitively demanding jobs like software development (or 90% of all white-collar jobs out there) and we’re NOT worried about the lynching mob going door to door and hanging IT people on lampposts? And I’m being serious, the impact this would have on societies would be unprecedented.
In such a future there would be a handful of lucky well paid artisans, a healthy community of hobbyists, and the overwhelming population of the planet who would be perfectly content to delegate their entire software diet to generative superplatforms that script themselves to perform any arbitrary software function. The concept of individual bespoke software programs will become an anachronism for an era where software was so difficult to produce that entire teams spent years painstakingly tweaking programs to spec.
Yes, but this might also be a counter-point to your position. In a world of rising baselines, having a "pot but without all the bells and whistles" might be a thing people need. Instead of having a pot that holds fluids at every conceivable angle with 20 temperature thresholds, 5 versions that hold liquid in a cloud for you, with monthly subscriptions, having the ability to get a "like a pot but make it left handed and only holds sand because this is what I need" could turn out to be something that people want/need/end up chasing.
In other words spec down, not up, and base + my particular kind of a pot.
Ah I see you're familiar with PowerPot by Atlassian(TM)
Very powerful potware xpity about how much soil it leaks.
I'd like to imagine that it's ultimately going to work out for humanity to be in a better condition overall if that happens.
However, if there reaches a level of near or full autonomy in all aspects of knowledge work, then there is a strong possibility that the field with the highest growth potential is going to be in "Private Security"; as those who have access to the most resources seek to defend their own positions in a societal return towards aristocracy and serfdom.
Let's hope not though.
I think many people will wish it was like this. But when AI becomes so capable that a nobody who doesn't understands computers a well as a "doctor" uses AI to create something BETTER or superior or just as good then employers will think... why am I paying that "doctor" more than that nobody?
See right now everything is fine, because AI is not that good yet. But if it gets better. Well. The future will be one where we trust AI more and more.
The sell was - Creating your apps was easier than before. You don't need to know any programming language. You don't need IT. You have a GUI and just drag and drop and visual your apps. Just fill the fields and voila your app is ready.
The criticism of the approach was that low codes apps might not be secure, unsuitable for large scale and critical apps, and lead to increase in unsupported apps by "shadow IT".
The same reasoning exists today for AI coding. Anyone can create apps, its not great for complex and mission critical apps, might not be secure etc. And lots of discussion like your post parsing the future too closely.
Low/no code apps have continue to grow. But many of the low/no code tool users are developers who use it to make their jobs easier.
While some might say - this time it is different. I believe we are currently we are the beginning of the cycle so everyone is excited to use their PowerApps shaped Claude Code/Codex to make their 100th budget tracking app but as time passes and edge cases are figured out the biggest users for AI are going to be software developers (and I believe they currently are the biggest users).
As for software engineering jobs just like before engineers will be expected to output more and faster. This has happened with every innovation from assembly compiler to IDE to low/no code ways of building.
The different levels of expertise exists even today and it will remain so.
with the latest agents, you can even be vague and they'll probably do an ok job if it's not something super complicated. some models are also pretty good at stopping to let you know about stuff you haven't thought about and ask for clarification.
so you can outsource thinking now, not just the doing part.
Is it different? because then:
> you can even be vague, and they'll probably do an ok job if it's not something super complicated
That's lot of qualifiers to say maybe, should be possible - how does someone who is not a software engineer know what is complex, vague or ok. There are lot of assumptions to say - This time it is different.
It is entirely possible it is different but I am yet to find a convincing argument.
AI can slop fork or clone existing software well, but a clone of an existing game is pointless, it's basically guaranteed to be derivative and worse than the original game, and games aren't so expensive that you can't just buy the original. AI can't know if new mechanics or angles to an existing genre will feel good to play, or if a new genre is fun, that requires a human to experience the game in its totality.
Games are also very resilient to sloppy AI coding, and if an indy game crashes nobody is getting paged.
Depends on what kind of game you are building. I can tell you that even Claude Opus absolutely hates desync bugs and has a rather hard tracing them. Maybe Fable or chatgpt-5.6 are better?
If you think it is different, just think of how many people write books professionally, or even publish online.
Once the noise settles down a bit and boardroom shakes off their delusions as you can see in rehiring in Ford and Zuck who was very bull on AI remark about "not being it". It will be just the same, but different.
Ford: https://www.bbc.com/news/articles/cgrkd41n2v9o IBM: https://qz.com/companies-rehiring-workers-ai-layoffs-automat...
However, this following quote has a simple reason that I don’t see anywhere in the article or framework:
“”” Why is there a huge gap between what people in various occupations could be using AI for and what they’re actually using it for? One reason could be that people are slow to adopt technology, and that’s certainly part of our framework. “””
I would like to add a reason: that the Silicon Valley companies who developed the LLMs are brigands: cognizant of their actions, they have stolen (and continue to steal) the world’s copyrighted material and are selling it back to the masses and the politicians as if they are the arbiters of information itself.
Specifically responding to the quoted question, I could be using Claude or ChatGPT or Grok or DeepSeek or any other to have come up with this comment, or to write emails, or to implement my Python for me, etc., but I use none of them for anything. Doing business with brigands is a choice, and a choice that I hope becomes less and less palatable so that the financial, political, social, and moral fever that is our zeitgeist finally breaks.
According to WRITER’s 2026 Enterprise Adoption Survey, 44% of Gen Z employees admit to sabotaging their company's AI strategy in at least one way compared to 29% of employees overall.
Sabotage behaviours include entering proprietary information into AI tools, using non-approved AI tools, refusing to use AI tools or outputs, ignoring guidelines or best practices, intentionally generating low-quality outputs, refusing to take AI training and tampering with performance metrics to make AI appear to underperform.
Very open definition of sabotage.
Everybody says they hate heroin but once you try it, you can't get enough of it.
Also new heroin users are only 30% likely to become addicted. https://jamanetwork.com/journals/jamapsychiatry/fullarticle/...
That seems a rather high baseline, maybe even more significant than younger generations putting another 50% on top of it as they are wont to do.
They say they are, but they have actually been using it so much that they can't even do they homework without AI at this point.
But I do believe (gut feeling) the novelty's worn off for a lot of people.
Jinxzi has Bernard as friend though
I sort of worry about things like AI figuring out scripts so well that even multi-tier support work is gone. And learning how to write fiction or create foods so in accordance to our tastes (sugar, fat, etc with food, exactly what each of us is interested in, with writing) that we even lose those truly human creative jobs. Might not ever wanna leave those bubbles.
So much of the human drive is exploration and why and what if. Assuming everyone in the world can have no money problems, what will AI not be able to figure out?
AI could be a horrible jailor. And better at cancelling than any perhaps sager Gen Z or millenial. Bears some caution to be wary of this and where that power sinkhole will go.
But then, I still think the previous AI winters were more a result of sense and caution than most of us know, and we cannot fathom our species' ways of reasoning/thought processes the way we did as a species thirty, fifty, eighty years ago. Erring on the side of caution is not a terrible thing.
Our technology has long since surpassed our capacity to understand its impact or to morally restrain its use - we've gone from chivalry to "oops, that was a school, oh well"
We will solve problems faster though, but we aren't showing signs that AI is making us better humans, and that is the problem.
If AI is subject to private ownership in a competitive market between competing suppliers, it will be like better cars, we’ll just drive faster.
Power consumption will be a limiting factor in those countries relying on intermittent, weather dependent power generation with no base load. Especially if users prefer Apple’s privacy first AI on edge devices.
Hopefully in western countries it can encourage more young women to bear three children before they turn 35. Young men have to pick up their game and create an environment to redirect their suicidal empathy into more productive pursuits.
“Where can you find another non-linear servo-mechanism weighing only 150 pounds and having great adaptability, that can be produced so cheaply by completely un-skilled labour?” - Albert Crossfield 1954
Animals don't "work". Not atleast for their own sake. If there is enough green pasture and water around, they don't even migrate to other places. So if work is meant to provide food and shelter and if machines can ensure that, humans don't need to "work".
Wealth is only a reserve capacity to help future generations so that they don't need to work for their basic needs. But if machines ensure that too, then wealth itself, as a reserve, is unnecessary.
How many man-hours go into various parts of the advertising distribution chain? Though a certain fraction of that energy goes to connecting people with goods and services they might find valuable, most of it goes into shifting numbers around for people that already don't personally have to worry about money.
We don't need to find endless ways for people to spin wheels, but as long as we're worried about "jobs," we will. We just need to find the social structures to provide people with basic needs and reserve "work" for things that are vital to society or truly inspired.
just?
Not that much. AI is already heavily automating advertising and has been for a long time. And a lot of activity that was previously exposed to untargeted passive advertising - like TV - has shifted to mediums that don't have any, like Netflix (for most subscribers).
And as you note, advertising isn't useless. It's how people find out about things they didn't know they wanted.
I don't think your assessment that this is mostly automated stands in the face of the number of man-hours both of these companies purchase each year.
Until the machines aren't owned by anyone (or owned by everyone, take your pick on the phrasing), the owners of the machine have no need to keep you alive.
This take is basically "Don't worry, people like Sam Altman are looking out for us"...
I can't wait to be kept in agistment by my overlords, fed on treacle and oats, ridden in circles once a fortnight, and shot when I break a leg.
[0] https://www.researchgate.net/figure/United-States-Farm-based...
These "needs" are sometimes enforced by the systems and government so that people don't stay away from the work and "economy" keeps churning. The housing prices could be a way to keep the people working for loan payments.
Instant foods, nursing homes for elderly, creches, roads, commuter trains - are all ways to have more workers and make them focused on work.
"Work" doesn't exist to keep people busy, it exists to keep them alive.
Even now the machine owners have enough power to change laws to their liking, bending governments and public opinion to their will.
I have no idea what that means.
> Why does the "world" have that power?
Okay, lets say "the environment". The universe does not care whether you live or die, so why are you so sure that someone will paternalistically care for you?
On what are you basing this viewpoint on?
> Machine owners don't have the power, for example.
That's because the world doesn't care about them either. So why would the world care about you?
I don't buy this. Tons of people in the West live on welfare and don't contribute shit to the economy, they are net takers. No one is plotting to kill them, in fact they're (arguably) getting more than they did 50 years ago and definitely more a 100 years ago. We live in democracies, not in a dystopian nightmare and I don't see why A.I is going to change that dynamic. People vote, people control the government, the government controls the armed forces, private citizens are not allowed to gather unregulated weapons... billionaire or no billionaire I don't see how you can beat that. And also, it's not like the billionaire class is some kind of a cohesive group that wants to work together to rule the world - they pretty much compete with and hate each other. I don't think Musk, Altman, Hassabis and Larry and Sergey are all going to agree to work together on "controling the machines" and killing all the other people.
And people had to fight for it. Through voting, yes, but also through strikes, sometimes riots, and some died for it. People all the world fought and died for the right to strike (basically, the right to refuse to work). For the right to organise as groups. For the right to only work 8 hours a day, for the right to a have a weekly day off. For the right to pay for pensions. For the right to have some sort of health insurance.
All over the world. Some still do. Many still don’t have half those rights.
Now, I don’t know about any billionaires specifically. But welfare didn’t fall from the sky. Or anyone’s good grace.
If in doubt, looking back at the late 1800s and early 1900s work and life conditions across the Western world is enlightening.
No one said anything about killing.
Of course, billionaires have other plans, and are the main obstacle in achieving any sort of social cohesion.
And what reason do they have for killing everyone else? Where is this abject nonsense even coming from? You can't just assert that the "owners of the machine" are all cartoonishly evil for some unknown reason.
All else being equal, at least Sam Altman et al. aren't constantly making up fantasies about exterminating people.
Who said they would kill everyone else? Since the rest of your strawmanned caricatured argument is a response to an argument never made, there's no point in addressing it.
Who said anything about killing?
Some may consider local LLMs to help with that as the power of LLMs would be more distributed. I think local LLMs would help marginally. Companies as an entity would still be better positioned to use these to their advantage vis a vis individuals.
So projecting into the future I still wonder if it won't become more challenging for individuals to make a living on average.
My dog was perfectly happy to snooze most of the day, play with toys occasionally, go for walks and try to hunt rodents. I on the other hand would be incredibly bored with that.
The drive some individuals have to control / exploit others isn't limited to the rich, they're just in a better position to exercise it
There are people who have absolutely no interest in what other people around them are doing; there are others who will move heaven and earth to help those around them whenever they can, even at the expense of their own interests; others still who can't help but compare themselves to everyone and make it everyone else's problem if they find themselves lacking; and yet others who at all costs want people to afford them "respect" which when you drill into it means stroking their ego
I'm sure we've all experienced petty individuals who've been granted just enough power that they can gain personal satisfaction from leveraging it to make themselves feel better at the expense of all other considerations
These are all just some of the many flavours of humanity
It's a status thing, right?
When enough green pastures are around animals usually reproduce till that is no longer the case and they need to move to other pastures. They continue doing this untill all green pastures are occupied. After this they start competing with one another to compete for the green pastures already occupied. Some animals will be so succesfull that they take larger green pastures, letting others starve. If by some miraclelous event (ai?) suddenly a lot of green pastures arise, animals will simply reproduce again till they occupied all green pastures again.
Looking at us? The agraric revolution, neither the Industrial revolution, nor this ai revolution have decreased how much we work on our 'jobs' (finding food indirectly). All it did was increasing the population so that all green pastures became occupied.
There will always be new, hard problems to work on. AI will not, and can not eliminate that.
But we already do have have some kind of measurement of most of these types of side factors, and they actually aren't at zero and are increasing rapidly. So the implication that they will not be human level until decades from now is just (hopeful?) speculation or fuzzy thinking.
To me this looks like a really academic and official sounding version of the same quasi-religious hopium that usually defends the sanctity of the human. He is essentially saying that there is just something so special about humans that it will never be reproduced in a machine. It's very similar to dualism (and in many people actually is religious dualism). No AI is going to have human creativity or judgement. Not anytime soon. Why? Well, we all just _know_ that's not possible. Okay, maybe in a couple of decades (but they don't necessarily believe that anyway). Why would that take decades? Well we all can just _tell_ it's no where close, right? Because AI of today just isn't special like humans.
Aside from that worldview issue, I think that people still are not taking seriously or internalizing the concept of exponential improvement.
Computing efficiency gains can actually level off. In fact, they have many, many times before. But they always tilt back up again when we invent the next approach to get beyond the current level. This is how it has been for 90 years.
There are multiple ways that we continue to see huge gains in AI software, architecture, and hardware. There are huge efficiency gains available still as we move towards more radical fully compute in memory and/or analog approaches and other options like models implemented in hardware.
My mother worked in an office in London as a shorthand typist in the 60s, along with thousands of other young women.
At some point computers began to enter offices, bosses typed their own letters and then emails, and this category of job simply evaporated. Of course there was still SOME secretarial work, and some workers retrained to do it, but most simply had to leave the sector.
Isn't that a common story with technology replacing workers?
What upset me a bit were phrases like “This is not a slogan. It’s a framework” which immediately devalued the work for me.
I have read so much Ai generated text recently, that I developed some AI-fatigue or AI-burnout, and I’m wondering if that might hit more fields - making more humans reject Ai work.
To be clear, I still like the text and I don’t know if it was written (partially) by Ai or not - but it’s this uncanny feeling I got reading it.
Why? Because it resembles the typical "You're not just implementing a text editor, you're reshaping the text editing landscape"?
I cant read this shit anymore.
"creating cross functional AI evaluation team to keep the company honest"
Fucking garbage.
I have an old book between me and the keyboard, and just read it while the AI is thinking, so I can read non AI words.. otherwise its like reading books by the same author over and over and over again.
I will go touch some grass now.