All I know is that trying to get a job now is being way more brutal than before. The closest thing I got this month was a scam email from an alleged "independent global senior recruiter" "currently supporting a confidential search for a company developing AI-powered digital products". The made-with-ai smell was all over the place, the company in mention allegedly was adobe, and am pretty sure they got my email from the "who wants to be hired?" posts in here.
Sample size of 1 here, but my inbox is absolutely overflowing with recruiters. Sample size of more than 1 - the SWE who were laid off recently from my company are finding new jobs quickly.
Certainly the job market is more brutal than it was during the covid mass hiring spree, but it doesn't seem any worse than historical average. My view is anecdotal but I haven't seen any real data on the other side of it either.
I imagine this comes with some credentialism and networking i.e. companies ask each other whether person X is good? If so, I can see how easily the current difficulties in the labor market can be overcome that way.
I don't have a CS degree, and I am talking about inbound recruiter emails, so it's not credentialism or networking. I did get my current job via networking though, but networking in this industry isn't really any more than "I used to work with this guy and he is good so give him an interview."
More anecdata, but around me, the opportunities for recent grads and self-taughts are less than try were.
If I was to bet, it seems that tasks that were given to new hires are now done by senior engineers and a prompt.
A collegue claims that the difference between an MBA and a mediocre compsci grad have disappeared due to vibe coding, so the compscis grads feel the job market is tougher because of increased competition.
All I know is that it's super hard to get junior devs to not be junior anymore because they don't have the opportunity anymore to learn on easy tasks.
> the opportunities for recent grads and self-taughts are less than try were.
It's also becuase the T20 CS/CE/ECE/EECS programs now graduate around 15-20K students a year.
If you don't attend a target you are going to have a hard time landing a job.
If you didn't do CS/CE/EE/ECE/EECS at MIT, CMU, Stanford, Cal, UIUC, GT, UT Austin, Purdue, UW, UCLA, UCSD, Ivies, UChicago, NYU, UNC Chapel Hill, Duke, USC, UMinn, UMich, the Military academies or you weren't a Vet, it's hard to justify hiring. Throw in 1-2 well regarded regional schools in the mix and that becomes your entire hiring pipeline.
Geography also plays a role - early career remote work is over. You will have to start your career in the Bay or NYC, and maybe Austin and Seattle.
Agentic coding only became reasonably decent this past December. Even in tech, most organizations that have adopted the tools are still trying to understand how capable they are and how to use them.
Point is it's too early to declare what the effect will be. It can take years for large organizations to change the way of doing things. The only sure-fire way to speed that up is if they suddenly start losing market-share. Otherwise, it's all herd-following and complacency in the majority of companies.
I suspect very few people in the tech industry understand or have experienced what office and civil service workers do. It might not be all that different than most engineering work: Organizing and arranging things, fitting things together, troubleshooting edge cases, bypassing controls to keep things working.
At the same time, we seem to be making rapid progress year over year in capability. Competition is keeping things progressing forward at an interesting pace.
It will be interesting to see what we are actually capable of 5 years from now... 10 years... There is almost definitely some job functions that are at risk and just don't know it, or haven't felt it yet.
You still need handholding, you can’t just give a few junior devs Claude Opus 5 and expect them to recreate Instagram in a few weeks. Frontend and backend basics sure, but security, scale, etc. need prior knowledge.
I'd say recreating an existing website as is, is the easier task. Doing a look alike, even if the models do take some shortcuts on the way is doable right now. You have a template frontend to compare to. Now if it would scale to the number of users is a different ask and harder to validate for the models.
Building something new is where they struggle most.
I totally agree, but a lot of this debate comes down to the definition of apocalypse a 20% reduction in jobs will affect a huge amount of people who will fill all forums with complaints of job getting difficulty without being an apocalypse.
Same with design and art direction. I think we might see a temp slowdown in the workforce but companies will want to stand out again after realizing the slop machines spit out fairly cookie cutter designs.
There was an article in Forbes [1] that shared that it's not about jobs going away, but the steady creep of tasks being taken away from people and given to AI and automation. The point made is that AI is hollowing out jobs, leading to earlier (and often forced) retirement, the destruction of entry-level jobs, and the steady creep of winnowing out workers making things leaner.
Judging by the IT maturity of most of the companies I have worked with in the last 30 years, my guess is we're about 10-15 years from any serious changes, solely because adopting AI will require a few things. Most are running razor thin margins and wont be able to afford any of this, others have so much technical debt piled up that even if they could afford it they couldn't do anything without identifying ancient systems and then updating them.
Most people I deal with still struggle with basic things. These were the types of people that couldn't program a VCR back in the 80's. Yes eventually they may get left behind but they are the majority of the population.
I'm sure AI is all the rage in SV but everywhere else I go, no one gives a damn about any of this. I was just in Florida for work last week. 5 days and not one person I talked to brought AI up outside of a vendor trying to scare me about Mythos. It's just a nonissue for most people.
In time maybe that changes, but I still hear people arguing about "the cloud" which was over a decade ago at this point in terms of tech trends. Expecting any of these people to understand what skills are or MD files or MCP is unrealistic. They simply don't give a fuck.
I agree with your sentiment, doing security auditing for old-school entprise and the like. I'd revise the timeline ti 3-10 years though. Eith how much AI can change productvity companies that don't adopt it will be in a tougher spot much more quickly than with new tech in the past.
AI is different than previous technologies in that a sufficiently advanced AI can "adopt" itself. That is, it can analyze the situation of a company (asking questions when necessary), describe the benefits, and deploy itself. This sounds like science fiction, but is very plausible in 3 years of trends hold.
The whole thing is that the trend wont hold steady. Its a n s curve and will taper off. The onky question is if the asympote is tapers to is above a certain threshold or not. Like the threshold needed to achieve what your talking about.
I dont know about the capability your talking about, but anyone who thinks these things are guaranteed, or a matter of time, is in for a rude awakening over the next few years
People have been foretelling the S curve flattening for years now. At every jump of capability people confidently assert that no more progress is possible, it has reached a plateau, there is no way to go beyond this, it's a fundamental limitation of the architecture, of the training methods etc. The horse will never ride the astronaut, it won't recognize cows on beaches, new theorems will never be proved, etc. etc.
I'm not sure how relevant the sentiment of average people are regarding the future. People are generally not well tuned in to any such things. Things just happen to most of us and we spend our lives thinking about gossip and life stuff, family, friends, chores etc., not musing about the future which we have little impact on.
Think about people's views on the coronavirus in February 2020. Things can change fast, and people will be like "Where did this come from all of a sudden? I was told by all the experts that AI is a bubble and irrelevant chatbot stuff." Most people hold extremely outdated (ie ~1-3 year out of date) ideas about AI capabilities and are absolutely unable to imagine what the capabilities in 1-3 years in the future will be (because they are unable to imagine what is already possible with frontier models today).
I agree with the sentiment entirely. I moved from the Bay Area to the DC Maryland Virginia area. the viewpoints out here have been strangely stuck on opinion pieces written during the gpt-3 era. Same goes for electric vehicles.
It’s slowly changing for some of the holdouts out here given how immediately useful certain personal and business use cases can be.
I don't see how that's relevant to the topic at hand. Average people's sentiment is not what will decide on whether jobs will be eliminated or not. It depends on model capabilities and people's self assessment of what's difficult in their own jobs is often massively miscalibrated, because they have no "software brain" and no "ML brain". They simply cannot think ahead. I've talked to people who object to the idea of them being replaceable by AI on the grounds that they have to log in to various GUI programs and (fairly trivially) align two information sources or copy some things from emails into systems where they have to log in separately with a password, and how could AI (which they think of as a chatbot in a browser tab) do all that. They don't know about Claude Code, Codex, the research on computer use (GUIs), anything like this.
What I'm seeing, as someone who works in a giant company but not in a tech department, is the rise of middle companies that specifically do all the hard work of organizing/wrangling the AI and providing more streamlined options that are easier to understand, with the end users giving up flexibility and freedom for ease of use. (Like the AI version of Dreamweaver, I guess would be a comparison?) Or modern no/low-code offerings.
I've also seen already existing SaaS offerings integrating AI into their already existing product. How well that's going to work... I'm not sure. The economics of at least some of what I've seen is questionable (e.g. giving end users essentially unlimited token spend without passing any of the cost along), and at least one company in my field is clearly adding AI features because they're spooked that the existence of AI is damaging their moat/vendor lock in.
The shift might look like 'company Y contracts with AI companies A, B, and C, and as a result does more with less and therefore needs 30% fewer people'. In theory, companies A, B, and C are creating jobs so the job numbers shift, but your point about a certain level of understanding being needed to do those new jobs definitely stands.
The question is are there enough competent people to staff the middleman companies (I'd say yes specifically because entry level tech is a bloodbath atm) and how simple can these tools be made? Can they be dumbed down enough so that businesses can just buy a package from a vendor where they can give tool access to the people who can't program VCRs and trust that they won't shoot themselves in the foot? That, I don't know.
100% agree, the only people enthusiastic about it that I know are faang / startup folks here on HN. The rest, including all possible white collar jobs see it more as some medium to distant threat to their work, and life is too short to spend it enshitifying with worries about potential incoming doom, we had plenty of those. Many otherwise very capable folks do it anyway, to visible detriment of their mental health (a bit, or more than a bit, seeing it first hand).
Even managers just follow the trends and just say things others are saying, they understand that their position is not granted in case of middle management in case of bigger upheaval.
For me, I can't even have access to any recent model at work, just heavily constrained gitlab duo which is still fine for most stuff. I cost them a day much more than good monthly subscription, but so are the policies of higher management, so - I don't care. It will come, I can pick up those skills a bit later, they are not that wide and deep if one is just an agentic user. If I get fired in next few years it will be other multinational corporate forces, not this. Proper SW engineering skills and experience picked over past 25 years are not going away anytime soon, its the solid base I can build on for anything next decade or two can bring my way. Summer is here, time to enjoy the mountains, lakes and family, thats what matters long term much more.
> Labor productivity was, in fact, slower in the first three years of our AI era than during the information technology boom that began in the mid 1990s.
That's a bit of a misleading statement, as it uses inaccurate terminology and cherrypicked comparisons. The linked chart is about labor productivity growth in those time frames, and current labor productivity is actually growing faster than projected since early 2023: See first chart here (rest of the article also worth skimming): https://aleximas.substack.com/p/what-is-the-impact-of-ai-on-...
If I may summarize my takeaway from that article: even though 50%+ of Americans currently self-report (major caveat) using AI, they use it for only 6% of work hours. This is what the recent Google study [1] called "broad but shallow use." But that still leads to 2% of work hour savings, which is a whopping 30%+ productivity boost per hour of AI used. And even with this low level of adoption, we may already be seeing the impact on labor productivity at national-level aggregate statistics!
Interestingly, this is across industries. You can play with the charts here [2]. There are all kinds of caveats involved, but there are other studies that rhyme (like [1] and those listed on the substack above.) This is kind of astounding considering it's been less than 4 years since ChatGPT was released.
My concern is that the impact on jobs is only beginning because 6% adoption is a very low number. I fear the impact will happen gradually, as adoption inches up... and then suddenly.
Then what is causing hiring to be so low, or the layoffs? Just a soft job market in general? The economy is actually tanking, not just "AI" jobs.
Did hiring managers just pump the brakes "just in case this AI thing is real". So we should see some big spike in hiring now that we know AI isn't going to do everything?
Or, the actual real numbers of hiring, is for real an indicator that AI is already taking jobs, or reducing jobs. Teams of 5, can now be teams of 2 or 3.
I think to some extent is that the right candidates and employers are not meeting.
Working at a ~1k employee company, we have been trying to hire both junior and senior devs. Out of ~10 interviews I conducted for junior, only 2 was able to program a linked list, and even that was with hints. It was not much better for senior(more system design question instead).
I understand that leetcode questions are (and have been) outdated, but linked list for somebody fresh out of school seems reasonable.
And this overlooks all the people using ai to cheat on the interview.
I would expect in this market we would have better candidates. But for us it was a challenge to hire.
And let’s not even get started with all the ai generated resumes and candidates that don’t actually exist, or have the experience they say they do. Really a waste of my time to interview.
The recruitment process is not made for a world where anybody can just copy paste the job req and generate a resume that fits everything.
A linked list is your litmus test? You did qualify with your "leetcode" blurb, but last I've used one was in college 30y ago and never again. We are not all kernel hackers / bare to the metal here.
linked lists aren't really emphasized anymore. languages have built in lists now. they had heard of them, knew what they were, but would probably struggle to do one from scratch.
of course i'm old school and kind of think that is basic, but colleges now aren't teaching it much anymore.
If someone can't implement the very simplest Graph Structure, Linked Lists, then I assume they can't implement any linked data structure. Learning variations of Linked Lists is where you start, e.g. https://rust-unofficial.github.io/too-many-lists/
AI could be good enough to dampen hiring without being good enough to cause an apocalypse. for all we know it’s plateauing at its ability to write code to match predefined specs (simplification), which is not sufficient to replace most jobs entirely.
It has been said here ad nausea - overhiring during covids, everybody was cargo culting amazon but there was only 1 amazon. Some folks were fired right when that whole charade with massive restrictions ended overnight, some were left around because times were good and money cheap. Thats over now so next waves.
In many corporate places including mine, software dev team's work is maybe 20% coding, and rest is simply getting things delivered to production and maintaning existing stuff. Nothing keeps working for long if left unattended. And there always multiple layers of test envs to manage, syncing with rest of the projects, troubleshooting/bugfixing, deliveries.
Llms won't help a nanometer here unless companies restructuralize their IT from ground up, firing entrenched teams/managers doing half-nothing. So ie if our team would went from 5->2, we would have 40% of the velocity, maaaybe 50%. Realistically, probably even less (since env maintenance is constant drag on productivity regardless on team's size)
It sort of is though because more jobs are becoming more mechanical. In the past, I wouldn't mind being a programmer but now with all this AI the job sucks. AI may have made things more efficient but that also balances on the other side of the equation: the boss now expects more, and it's much less fun to produce more than to concentrate well on a single topic. So jobs suck more and that's harsh.
Also, many jobs are being threatened, including my own. And I don't really want to get a different job, even though I will if I have to. I like my current one and I'm very specific about what I do. I will get another job but it's probably not going to be as fun.
There's some truth in that but it's also a general feeling I get from browsing a lot of job boards. And I think in 10 years it's going to get worse for sure.
Eh, me and most people I know are burning out fast.
There's a split between the people that love to program, for the sake of it, and those that just want a well paying job.
Seems like the former group is hit the hardest re: morale. The job is just not fun anymore. It's prompting and reviewing an onslaught of mass amounts of code. No more problem solving or innovation. At least, in the ways we used to.
Seems like the latter group has less issues with this.
And what about the health outcomes that it will negatively impact? Like increased rare earth mining in poorer countries that cause illnesses? Oh wait, we're only talking about health incomes of those that can afford it. Never mind.
Exactly right. We all have to survive on a personal level - that's our instinct. But we can only try and minimize our impact and do the best we can, even though on a global scale supporting the current system is not a good thing in many cases.
I don't think badly of anyone who uses AI who makes a living because, survival. But on the other hand, I think the world would be better off without it. One can still wish for a better world while trying to survive in the current one using the tools at one's disposal, even if that situation is suboptimal.
That’s fair, I think it’s possible to be both grateful for the existence of capitalism funding research into those companies and disgust over how much money they make.
The "AI" that is being used to find cures for cancers is not the same "AI" that is being used as an excuse to fire people, especially programmers.
The latter is LLMs. Generative AI, chatbots, etc.
The former is analytical AI: machine-learning systems using computer vision techniques to recognize patterns in CT scans (and similar) that humans aren't precise enough to pick up on.
They are derived from similar principles, but no one is trying to sell DetectMelanomaChat as a replacement for you or me.
If I did get cancer, my survival instinct might indeed get me to use it. But I'd still hate AI because it's very likely that the cancer would be caused by all the stuff we've done to the planet and AI would then just be another bandaid that would probably in turn cause more cancer because of how it will encourage more people to use resources, and how it probably will cause more stress to the average human being.
And solving some cases of cancer is not worth the even greater medical problems AI will be foisting upon the world.
You raise a possible outcome and I’m going to raise another in that AI allows us to come up with the technology to improve/reverse the impact to the planet.
> You raise a possible outcome and I’m going to raise another in that AI allows us to come up with the technology to improve/reverse the impact to the planet.
History has shown that we've never used technology to do that. We always go for the economic return first. We only use technology to ameliorate immediate pain, never to avoid long-term disaster.
Even green energy/solar panels - the CO2 curve hasn't been flattened because we're always constrained by business as usual. It's human instinct. I sincerely doubt AI can have a net positive impact because we're already seeing for every 1 good outcome, 100 greedy ones from the elite.
When manufacturing turned into electrified factories and conveyor belts, there wasn't waves of massive layoffs. And AI these days come nowhere near close to conveyor belts or electrification at scale at all. Debt cycles could fund it to a certain scale, but we don't know our priorities either: Do you fund hyperscalers? Do you fund ex-crypto miner co-location data centers? Do you fund centralized frontier model labs? Do you let everyone be able to deploy these models on any throwaway hardware? What is the best standard for model distribution?
We are so far off from these discussions that AI hype these days really sounds like a couple of frat bros BSing and not knowing what to do with their Ivy League credentials and inheritance money.
> When manufacturing turned into electrified factories and conveyor belts, there wasn't waves of massive layoffs. And AI these days come nowhere near close to conveyor belts or electrification at scale at all.
Nowhere near close, but on the high side. AI has propagated far faster, and is being adopted by a far greater proportion of businesses and individuals than electricity or conveyor belts at a similar age. AI is scaling faster than any tech in history. Exploding may be a better word choice.
It's because the people making it are marketing it aggressively to other CEOs as the way they can eliminate their salary budgets and pocket them as profit, and drumming up FOMO among them so each feels that if they don't hop on the bandwagon they're going to be left in the dust by all the others.
Sure, but that doesn’t by itself mean it can’t be a problem. Maybe the macro scale effects were mitigated by a slow rollout and electrification would have caused an unemployment wave if everyone had done it at once.
All I know is that trying to get a job now is being way more brutal than before. The closest thing I got this month was a scam email from an alleged "independent global senior recruiter" "currently supporting a confidential search for a company developing AI-powered digital products". The made-with-ai smell was all over the place, the company in mention allegedly was adobe, and am pretty sure they got my email from the "who wants to be hired?" posts in here.
I wonder if we’re actually in like a shadow recession and companies are just using AI as an excuse to do layoffs or hire less.
Idk there’s a lot of diff variables and it’s complicated of course, I really don’t think it’s because “AI’s taking the jobs” tho.
Unemployment rate and labor participation rate don’t support this.. probably true for software engineering but not the broader economy afaict
Uh, wat?
Edit: for clarification, it’s never been worse. See link.
https://finance.yahoo.com/economy/articles/labor-market-data...
For tech it truly depends on your geo.
If you are in Western Europe, then yes it's because of a shadow recession.
If you are in Seattle, then it's because the larger employers are shifting headcount to the Bay or abroad.
If you are not in the Bay, NYC, or Seattle, it's because domestic nearshoring during 2018-23 failed (eg. Cisco, NetApp, and Google in RTP)
I've written about this before on here [0][1]
[0] - https://news.ycombinator.com/item?id=45466975
[1] - https://news.ycombinator.com/item?id=47174561
The job market is a fickle mistress.
Companies are harvesting domestic salaries, possibly offshoring and slashing jobs, and blaming AI.
They want you to not oppose their offshoring, or stunning executive salaries, or God forbid join a union.
Who ya gonna blame? AI.
[dead]
Sample size of 1 here, but my inbox is absolutely overflowing with recruiters. Sample size of more than 1 - the SWE who were laid off recently from my company are finding new jobs quickly.
Certainly the job market is more brutal than it was during the covid mass hiring spree, but it doesn't seem any worse than historical average. My view is anecdotal but I haven't seen any real data on the other side of it either.
Which business areas, programming languages / frameworks, which geo areas?
US fintech. Language / framework agnostic.
I imagine this comes with some credentialism and networking i.e. companies ask each other whether person X is good? If so, I can see how easily the current difficulties in the labor market can be overcome that way.
I don't have a CS degree, and I am talking about inbound recruiter emails, so it's not credentialism or networking. I did get my current job via networking though, but networking in this industry isn't really any more than "I used to work with this guy and he is good so give him an interview."
More anecdata, but around me, the opportunities for recent grads and self-taughts are less than try were.
If I was to bet, it seems that tasks that were given to new hires are now done by senior engineers and a prompt.
A collegue claims that the difference between an MBA and a mediocre compsci grad have disappeared due to vibe coding, so the compscis grads feel the job market is tougher because of increased competition.
All I know is that it's super hard to get junior devs to not be junior anymore because they don't have the opportunity anymore to learn on easy tasks.
> the opportunities for recent grads and self-taughts are less than try were.
It's also becuase the T20 CS/CE/ECE/EECS programs now graduate around 15-20K students a year.
If you don't attend a target you are going to have a hard time landing a job.
If you didn't do CS/CE/EE/ECE/EECS at MIT, CMU, Stanford, Cal, UIUC, GT, UT Austin, Purdue, UW, UCLA, UCSD, Ivies, UChicago, NYU, UNC Chapel Hill, Duke, USC, UMinn, UMich, the Military academies or you weren't a Vet, it's hard to justify hiring. Throw in 1-2 well regarded regional schools in the mix and that becomes your entire hiring pipeline.
Geography also plays a role - early career remote work is over. You will have to start your career in the Bay or NYC, and maybe Austin and Seattle.
Agentic coding only became reasonably decent this past December. Even in tech, most organizations that have adopted the tools are still trying to understand how capable they are and how to use them.
Point is it's too early to declare what the effect will be. It can take years for large organizations to change the way of doing things. The only sure-fire way to speed that up is if they suddenly start losing market-share. Otherwise, it's all herd-following and complacency in the majority of companies.
It still doesn't seem to work all that well for tasks which don't have a short feedback loop.
Don’t expect The Guardian to report on AI taking jobs. Their audience wants to believe AI is failing and could never match their superior intellects.
Newsflash: AI is already remarkably good at replacing the exact office and civil service paper-pushing they do every day.
I suspect very few people in the tech industry understand or have experienced what office and civil service workers do. It might not be all that different than most engineering work: Organizing and arranging things, fitting things together, troubleshooting edge cases, bypassing controls to keep things working.
At the same time, we seem to be making rapid progress year over year in capability. Competition is keeping things progressing forward at an interesting pace.
It will be interesting to see what we are actually capable of 5 years from now... 10 years... There is almost definitely some job functions that are at risk and just don't know it, or haven't felt it yet.
You still need handholding, you can’t just give a few junior devs Claude Opus 5 and expect them to recreate Instagram in a few weeks. Frontend and backend basics sure, but security, scale, etc. need prior knowledge.
I'd say recreating an existing website as is, is the easier task. Doing a look alike, even if the models do take some shortcuts on the way is doable right now. You have a template frontend to compare to. Now if it would scale to the number of users is a different ask and harder to validate for the models.
Building something new is where they struggle most.
I totally agree, but a lot of this debate comes down to the definition of apocalypse a 20% reduction in jobs will affect a huge amount of people who will fill all forums with complaints of job getting difficulty without being an apocalypse.
Same with design and art direction. I think we might see a temp slowdown in the workforce but companies will want to stand out again after realizing the slop machines spit out fairly cookie cutter designs.
Tell that to recruiters requiring leet code tests for Senior/Leadership roles. Absolutely idiotic market at the moment.
There was an article in Forbes [1] that shared that it's not about jobs going away, but the steady creep of tasks being taken away from people and given to AI and automation. The point made is that AI is hollowing out jobs, leading to earlier (and often forced) retirement, the destruction of entry-level jobs, and the steady creep of winnowing out workers making things leaner.
[1] https://www.forbes.com/sites/ronschmelzer/2026/07/14/ai-isnt...
Judging by the IT maturity of most of the companies I have worked with in the last 30 years, my guess is we're about 10-15 years from any serious changes, solely because adopting AI will require a few things. Most are running razor thin margins and wont be able to afford any of this, others have so much technical debt piled up that even if they could afford it they couldn't do anything without identifying ancient systems and then updating them.
Most people I deal with still struggle with basic things. These were the types of people that couldn't program a VCR back in the 80's. Yes eventually they may get left behind but they are the majority of the population.
I'm sure AI is all the rage in SV but everywhere else I go, no one gives a damn about any of this. I was just in Florida for work last week. 5 days and not one person I talked to brought AI up outside of a vendor trying to scare me about Mythos. It's just a nonissue for most people.
In time maybe that changes, but I still hear people arguing about "the cloud" which was over a decade ago at this point in terms of tech trends. Expecting any of these people to understand what skills are or MD files or MCP is unrealistic. They simply don't give a fuck.
I agree with your sentiment, doing security auditing for old-school entprise and the like. I'd revise the timeline ti 3-10 years though. Eith how much AI can change productvity companies that don't adopt it will be in a tougher spot much more quickly than with new tech in the past.
AI is different than previous technologies in that a sufficiently advanced AI can "adopt" itself. That is, it can analyze the situation of a company (asking questions when necessary), describe the benefits, and deploy itself. This sounds like science fiction, but is very plausible in 3 years of trends hold.
The whole thing is that the trend wont hold steady. Its a n s curve and will taper off. The onky question is if the asympote is tapers to is above a certain threshold or not. Like the threshold needed to achieve what your talking about.
I dont know about the capability your talking about, but anyone who thinks these things are guaranteed, or a matter of time, is in for a rude awakening over the next few years
People have been foretelling the S curve flattening for years now. At every jump of capability people confidently assert that no more progress is possible, it has reached a plateau, there is no way to go beyond this, it's a fundamental limitation of the architecture, of the training methods etc. The horse will never ride the astronaut, it won't recognize cows on beaches, new theorems will never be proved, etc. etc.
> AI is different than previous technologies in that a sufficiently advanced AI can "adopt" itself.
Literally does not exist
"very plausible in 3 years of trends hold"
> It's just a nonissue for most people.
I'm not sure how relevant the sentiment of average people are regarding the future. People are generally not well tuned in to any such things. Things just happen to most of us and we spend our lives thinking about gossip and life stuff, family, friends, chores etc., not musing about the future which we have little impact on.
Think about people's views on the coronavirus in February 2020. Things can change fast, and people will be like "Where did this come from all of a sudden? I was told by all the experts that AI is a bubble and irrelevant chatbot stuff." Most people hold extremely outdated (ie ~1-3 year out of date) ideas about AI capabilities and are absolutely unable to imagine what the capabilities in 1-3 years in the future will be (because they are unable to imagine what is already possible with frontier models today).
I agree with the sentiment entirely. I moved from the Bay Area to the DC Maryland Virginia area. the viewpoints out here have been strangely stuck on opinion pieces written during the gpt-3 era. Same goes for electric vehicles.
It’s slowly changing for some of the holdouts out here given how immediately useful certain personal and business use cases can be.
> I'm not sure how relevant the sentiment of average people are regarding the future.
They vote. They buy the products. They are the only ones that ARE relevant.
I don't see how that's relevant to the topic at hand. Average people's sentiment is not what will decide on whether jobs will be eliminated or not. It depends on model capabilities and people's self assessment of what's difficult in their own jobs is often massively miscalibrated, because they have no "software brain" and no "ML brain". They simply cannot think ahead. I've talked to people who object to the idea of them being replaceable by AI on the grounds that they have to log in to various GUI programs and (fairly trivially) align two information sources or copy some things from emails into systems where they have to log in separately with a password, and how could AI (which they think of as a chatbot in a browser tab) do all that. They don't know about Claude Code, Codex, the research on computer use (GUIs), anything like this.
What I'm seeing, as someone who works in a giant company but not in a tech department, is the rise of middle companies that specifically do all the hard work of organizing/wrangling the AI and providing more streamlined options that are easier to understand, with the end users giving up flexibility and freedom for ease of use. (Like the AI version of Dreamweaver, I guess would be a comparison?) Or modern no/low-code offerings.
I've also seen already existing SaaS offerings integrating AI into their already existing product. How well that's going to work... I'm not sure. The economics of at least some of what I've seen is questionable (e.g. giving end users essentially unlimited token spend without passing any of the cost along), and at least one company in my field is clearly adding AI features because they're spooked that the existence of AI is damaging their moat/vendor lock in.
The shift might look like 'company Y contracts with AI companies A, B, and C, and as a result does more with less and therefore needs 30% fewer people'. In theory, companies A, B, and C are creating jobs so the job numbers shift, but your point about a certain level of understanding being needed to do those new jobs definitely stands.
The question is are there enough competent people to staff the middleman companies (I'd say yes specifically because entry level tech is a bloodbath atm) and how simple can these tools be made? Can they be dumbed down enough so that businesses can just buy a package from a vendor where they can give tool access to the people who can't program VCRs and trust that they won't shoot themselves in the foot? That, I don't know.
100% agree, the only people enthusiastic about it that I know are faang / startup folks here on HN. The rest, including all possible white collar jobs see it more as some medium to distant threat to their work, and life is too short to spend it enshitifying with worries about potential incoming doom, we had plenty of those. Many otherwise very capable folks do it anyway, to visible detriment of their mental health (a bit, or more than a bit, seeing it first hand).
Even managers just follow the trends and just say things others are saying, they understand that their position is not granted in case of middle management in case of bigger upheaval.
For me, I can't even have access to any recent model at work, just heavily constrained gitlab duo which is still fine for most stuff. I cost them a day much more than good monthly subscription, but so are the policies of higher management, so - I don't care. It will come, I can pick up those skills a bit later, they are not that wide and deep if one is just an agentic user. If I get fired in next few years it will be other multinational corporate forces, not this. Proper SW engineering skills and experience picked over past 25 years are not going away anytime soon, its the solid base I can build on for anything next decade or two can bring my way. Summer is here, time to enjoy the mountains, lakes and family, thats what matters long term much more.
> Labor productivity was, in fact, slower in the first three years of our AI era than during the information technology boom that began in the mid 1990s.
That's a bit of a misleading statement, as it uses inaccurate terminology and cherrypicked comparisons. The linked chart is about labor productivity growth in those time frames, and current labor productivity is actually growing faster than projected since early 2023: See first chart here (rest of the article also worth skimming): https://aleximas.substack.com/p/what-is-the-impact-of-ai-on-...
Some economists suggest it's because of the impact of AI. Here's a study from the St Louis Fed (not quite a tech bro-pilled source I'd assume): https://www.stlouisfed.org/on-the-economy/2025/nov/state-gen...
If I may summarize my takeaway from that article: even though 50%+ of Americans currently self-report (major caveat) using AI, they use it for only 6% of work hours. This is what the recent Google study [1] called "broad but shallow use." But that still leads to 2% of work hour savings, which is a whopping 30%+ productivity boost per hour of AI used. And even with this low level of adoption, we may already be seeing the impact on labor productivity at national-level aggregate statistics!
Interestingly, this is across industries. You can play with the charts here [2]. There are all kinds of caveats involved, but there are other studies that rhyme (like [1] and those listed on the substack above.) This is kind of astounding considering it's been less than 4 years since ChatGPT was released.
My concern is that the impact on jobs is only beginning because 6% adoption is a very low number. I fear the impact will happen gradually, as adoption inches up... and then suddenly.
[1] https://blog.google/innovation-and-ai/technology/research/un... (discussion: https://news.ycombinator.com/item?id=49020335)
[2] https://www.genaiadoptiontracker.com/
for many people it is already here...
Then what is causing hiring to be so low, or the layoffs? Just a soft job market in general? The economy is actually tanking, not just "AI" jobs.
Did hiring managers just pump the brakes "just in case this AI thing is real". So we should see some big spike in hiring now that we know AI isn't going to do everything?
Or, the actual real numbers of hiring, is for real an indicator that AI is already taking jobs, or reducing jobs. Teams of 5, can now be teams of 2 or 3.
I think to some extent is that the right candidates and employers are not meeting.
Working at a ~1k employee company, we have been trying to hire both junior and senior devs. Out of ~10 interviews I conducted for junior, only 2 was able to program a linked list, and even that was with hints. It was not much better for senior(more system design question instead).
I understand that leetcode questions are (and have been) outdated, but linked list for somebody fresh out of school seems reasonable.
And this overlooks all the people using ai to cheat on the interview.
I would expect in this market we would have better candidates. But for us it was a challenge to hire.
And let’s not even get started with all the ai generated resumes and candidates that don’t actually exist, or have the experience they say they do. Really a waste of my time to interview.
The recruitment process is not made for a world where anybody can just copy paste the job req and generate a resume that fits everything.
A linked list is your litmus test? You did qualify with your "leetcode" blurb, but last I've used one was in college 30y ago and never again. We are not all kernel hackers / bare to the metal here.
What would be a reasonable question, that’s similar difficulty as implementing a linked list ?
We have tried other questions as well, with similar results but perhaps our questions are outdated ?
checking in with my kids that just graduated.
linked lists aren't really emphasized anymore. languages have built in lists now. they had heard of them, knew what they were, but would probably struggle to do one from scratch.
of course i'm old school and kind of think that is basic, but colleges now aren't teaching it much anymore.
If someone can't implement the very simplest Graph Structure, Linked Lists, then I assume they can't implement any linked data structure. Learning variations of Linked Lists is where you start, e.g. https://rust-unofficial.github.io/too-many-lists/
Many working programmers have no need to implement linked data structures in their day to day jobs.
There are several things: * economic uncertainty (wars, energy shocks, etc) * policies chaos * end of free money (interest rates)
AI could be good enough to dampen hiring without being good enough to cause an apocalypse. for all we know it’s plateauing at its ability to write code to match predefined specs (simplification), which is not sufficient to replace most jobs entirely.
CEOs say we are in a low fire, low hire economy. They view AI as important but not transformative.
https://www.conference-board.org/topics/CEO-Confidence/
AI has the mind share but I guess it's not hitting the bottom line enough for tech companies to grow.
If anything it feels like AI has sucked the air out of the room as far as investment dollars goes.
It has been said here ad nausea - overhiring during covids, everybody was cargo culting amazon but there was only 1 amazon. Some folks were fired right when that whole charade with massive restrictions ended overnight, some were left around because times were good and money cheap. Thats over now so next waves.
In many corporate places including mine, software dev team's work is maybe 20% coding, and rest is simply getting things delivered to production and maintaning existing stuff. Nothing keeps working for long if left unattended. And there always multiple layers of test envs to manage, syncing with rest of the projects, troubleshooting/bugfixing, deliveries.
Llms won't help a nanometer here unless companies restructuralize their IT from ground up, firing entrenched teams/managers doing half-nothing. So ie if our team would went from 5->2, we would have 40% of the velocity, maaaybe 50%. Realistically, probably even less (since env maintenance is constant drag on productivity regardless on team's size)
It sort of is though because more jobs are becoming more mechanical. In the past, I wouldn't mind being a programmer but now with all this AI the job sucks. AI may have made things more efficient but that also balances on the other side of the equation: the boss now expects more, and it's much less fun to produce more than to concentrate well on a single topic. So jobs suck more and that's harsh.
Also, many jobs are being threatened, including my own. And I don't really want to get a different job, even though I will if I have to. I like my current one and I'm very specific about what I do. I will get another job but it's probably not going to be as fun.
I hate AI and the people that made it.
>So jobs suck more
Be careful not to generalize from your job. The programmers I know do not have jobs that sound like yours.
There's some truth in that but it's also a general feeling I get from browsing a lot of job boards. And I think in 10 years it's going to get worse for sure.
Eh, me and most people I know are burning out fast.
There's a split between the people that love to program, for the sake of it, and those that just want a well paying job.
Seems like the former group is hit the hardest re: morale. The job is just not fun anymore. It's prompting and reviewing an onslaught of mass amounts of code. No more problem solving or innovation. At least, in the ways we used to.
Seems like the latter group has less issues with this.
Will you still hate AI when it discovers the cure for your cancer?
It won’t. And I will.
I’m not sure if your question misunderstands what AI is and what it can do, or if it misunderstands what cancer is and how medical research works.
It was more of an expression to say that AI will improve health outcomes due to its ability to increase the speed at which research is done.
It might do that, but as it looks currently you likely won't stand to benefit.
And what about the health outcomes that it will negatively impact? Like increased rare earth mining in poorer countries that cause illnesses? Oh wait, we're only talking about health incomes of those that can afford it. Never mind.
People can absolutely have complicated feelings about things that help them.
My life is sustained by medicine produced by large international pharma companies, and I also feel really uncomfortable about that.
Exactly right. We all have to survive on a personal level - that's our instinct. But we can only try and minimize our impact and do the best we can, even though on a global scale supporting the current system is not a good thing in many cases.
I don't think badly of anyone who uses AI who makes a living because, survival. But on the other hand, I think the world would be better off without it. One can still wish for a better world while trying to survive in the current one using the tools at one's disposal, even if that situation is suboptimal.
That’s fair, I think it’s possible to be both grateful for the existence of capitalism funding research into those companies and disgust over how much money they make.
The "AI" that is being used to find cures for cancers is not the same "AI" that is being used as an excuse to fire people, especially programmers.
The latter is LLMs. Generative AI, chatbots, etc.
The former is analytical AI: machine-learning systems using computer vision techniques to recognize patterns in CT scans (and similar) that humans aren't precise enough to pick up on.
They are derived from similar principles, but no one is trying to sell DetectMelanomaChat as a replacement for you or me.
We can cross that bridge when we get there
If I did get cancer, my survival instinct might indeed get me to use it. But I'd still hate AI because it's very likely that the cancer would be caused by all the stuff we've done to the planet and AI would then just be another bandaid that would probably in turn cause more cancer because of how it will encourage more people to use resources, and how it probably will cause more stress to the average human being.
And solving some cases of cancer is not worth the even greater medical problems AI will be foisting upon the world.
You raise a possible outcome and I’m going to raise another in that AI allows us to come up with the technology to improve/reverse the impact to the planet.
> You raise a possible outcome and I’m going to raise another in that AI allows us to come up with the technology to improve/reverse the impact to the planet.
History has shown that we've never used technology to do that. We always go for the economic return first. We only use technology to ameliorate immediate pain, never to avoid long-term disaster.
Even green energy/solar panels - the CO2 curve hasn't been flattened because we're always constrained by business as usual. It's human instinct. I sincerely doubt AI can have a net positive impact because we're already seeing for every 1 good outcome, 100 greedy ones from the elite.
That’s valid, I remain hoping :’>
[dead]
When manufacturing turned into electrified factories and conveyor belts, there wasn't waves of massive layoffs. And AI these days come nowhere near close to conveyor belts or electrification at scale at all. Debt cycles could fund it to a certain scale, but we don't know our priorities either: Do you fund hyperscalers? Do you fund ex-crypto miner co-location data centers? Do you fund centralized frontier model labs? Do you let everyone be able to deploy these models on any throwaway hardware? What is the best standard for model distribution?
We are so far off from these discussions that AI hype these days really sounds like a couple of frat bros BSing and not knowing what to do with their Ivy League credentials and inheritance money.
> When manufacturing turned into electrified factories and conveyor belts, there wasn't waves of massive layoffs. And AI these days come nowhere near close to conveyor belts or electrification at scale at all.
Nowhere near close, but on the high side. AI has propagated far faster, and is being adopted by a far greater proportion of businesses and individuals than electricity or conveyor belts at a similar age. AI is scaling faster than any tech in history. Exploding may be a better word choice.
Is this not because of the internet and existing infrastructure which didn’t exist during that time?
It's because the people making it are marketing it aggressively to other CEOs as the way they can eliminate their salary budgets and pocket them as profit, and drumming up FOMO among them so each feels that if they don't hop on the bandwagon they're going to be left in the dust by all the others.
Sure, but that doesn’t by itself mean it can’t be a problem. Maybe the macro scale effects were mitigated by a slow rollout and electrification would have caused an unemployment wave if everyone had done it at once.
But, but... They said. Wait, did they... Lie?
"they" including this same publication, ex: https://www.theguardian.com/technology/2025/nov/25/ai-could-...
Panicked more like.