Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
×
Programming AI

US Programming Jobs Plunge 27.5% in Two Years (msn.com) 94

Computer programming jobs in the US have declined by more than a quarter over the past two years, placing the profession among the 10 hardest-hit occupations of 420-plus jobs tracked by the Bureau of Labor Statistics and potentially signaling the first concrete evidence of artificial intelligence replacing workers.

The timing coincides with OpenAI's release of ChatGPT in late 2022. Anthropic researchers found people use AI to perform programming tasks more than those of any other job, though 57 percent of users employ AI to augment rather than automate work. "Without getting hysterical, the unemployment jump for programming really does look at least partly like an early, visible labor market effect of AI," said Mark Muro of the Brookings Institution.

While software developer positions have remained stable with only a 0.3 percent decline, programmers who perform more routine coding from specifications provided by others have seen their ranks diminish to levels not seen since 1980. Economists caution that high interest rates and post-pandemic tech industry contraction have also contributed to the decline in programming jobs, which typically pay $99,700 compared to $132,270 for developers.

US Programming Jobs Plunge 27.5% in Two Years

Comments Filter:
  • It's the ideal problem space, really.

    Programming is literally about exacting rules and deterministic outcomes. It's exactly the sort of thing I'd expect a computer to be good at.

    Kind of like I used to say about chess; I was only surprised that it had taken so long for computers to get good at it.

    • It's exactly the sort of thing I'd expect a computer to be good at.

      There is no rational basis for such an expectation. AI is good at taking human-made snippets of code and passing it off as its own, but not much else. It has no creative ability whatsoever, and can only use code it has found elsewhere. It is incapable of creating something new.

      • I don't disagree with that, but that describes much of human coding too.

      • by ranton ( 36917 )

        It's exactly the sort of thing I'd expect a computer to be good at.

        There is no rational basis for such an expectation. AI is good at taking human-made snippets of code and passing it off as its own, but not much else. It has no creative ability whatsoever, and can only use code it has found elsewhere. It is incapable of creating something new.

        One of the ways the BLS differentiates between programmers and software developers is that programmers don't do much of that creative work you mention. That's why only programmers show a decline, while software developer positions are still growing. You should also keep in mind that by BLS definitions, there are less than 10% as many programmers as there are software engineers in the marketplace.

      • Something new like changing a return type of a function given no other context? Or designing a new browser that doesn't use chromium with no other context? Two things that most people can't do while the second is something people refuse to do but im convinced an AI can do easily.....
        • by Shaitan ( 22585 )

          There is nothing difficult about designing a new browser. That's just a whole lot of wheels to reinvent without any need to justify it... browsers parse and render a lot of languages and formats and backwards supports for the same.

    • I used to think the same. But generation of artistic outputs is actually the ideal problem space because measuring the value of the resulting outputs is subjective. There is no actual way to measure that. "Good" is opinion.

      Whereas, in code, programming, networking, technology, you are into multi orthogonal domains of knowledge, where measurements can made against agreed upon standards and value can found in performance and utility.

      Chess is a single constrained domain knowledge.
      • where measurements can made against agreed upon standards and value can found in performance and utility.

        With all the crypto and AI and other VC funded nonsense out there couldn't we say that say Sturgeon's Law applies to code the same as it does for art?

        • I think so. I haven't seen any value in new commercial software in 15 years. Never saw any true value in blockchain. You can do everything blockchain does with 25 year old technology, and it's a ton cheaper and faster and comprehensible. AI? Candy coated surveillance. We don't need it. We don't want it.

          In defence of my previous statement however, I'm saying that the value of art is truly subjective, there is no agreed upon standards of value, excellence, whereas, at least in the case of generated code, you
          • If it runs is a factual statement not a standard of value, it existing is the just the first step before we can give it some sort of normative value.

            My C++ "Hello World" from high school in the 90's runs but does that give it value of any standard? I would say no, just the same that every crypto networks runs successfully but I would say it'd gives no value or even negative value. Anytime we assign normativity we are in some world of subjectivity, I think we are just talking about two different yet related

      • That's basically saying art is subjective so you can't tell if bad stuff is bad.

        I disagree. The point of art is to communicate something to other humans, and make you think. Like the banana and tape: you might dislike it, but it got people world-wide taking about at and arguing about artistic merit. That's what art can do. The AI slop? Not so much.

        • I agree that art is about communicating and encouraging thought. And one can determine the success of a work of art (i.e., whether it's "good" at being art) by how well it achieves that purpose.

          Still, I'd argue that much of art appreciation is indeed subjective. Suppose your daughter drew a picture of your family for Grade 3 art class that you adored and hung on the wall. For you, it's good -- but it means nowhere near as much to others aside from being cute.

          I am impressed by what AI art has achieved. Just

          • setting aside how it was created in the first place.

            My belief is that this is and always should be linked to art, not the only thing but it can't be separated either. Art should be gatekept a little bit, there should be some effort and skill put into it's creation, that process of learning and growing is what makes the artist. There shouldn't be zero friction to create art, the friction is where the art becomes something more than the image or the sound.

            If someone tells you they are a "musician" and they don't know an instrument or how to do production or

            • setting aside how it was created in the first place.

              My belief is that this is and always should be linked to art, not the only thing but it can't be separated either. Art should be gatekept a little bit, there should be some effort and skill put into it's creation, that process of learning and growing is what makes the artist. There shouldn't be zero friction to create art, the friction is where the art becomes something more than the image or the sound.

              Fair enough.

              If someone tells you they are a "musician" and they don't know an instrument or how to do production or operate a DAW but they just prompt in songs they like into an AI and it spits out something, would you think it's fair for them to use that term? We would expect them top have put in some of the work.

              I suppose the definition of a musician is as fuzzy as the definition of an artist. I agree that one should expect some kind of commitment to a craft before one can be deemed a practitioner of it. However, your hypothetical person might be making music without having earned the title of musician.

              Even in the case of the taped banana, if I tape a banana to the wall and call it art nobody would care and rightly so, but artist Maurizio Cattelan who was already well accomplished in several artistic disciplines can get attention, because he is an artist, he's done the work as it were, even as simple as it is so we have a base level of respect for the work even if we don't like it.

              Yes, there's an element of reputation that comes with recognizing a work of art. No question. Consider an analogous example from music: John Cage. He sought to re-invent what we even call music. He's the

              • However, your hypothetical person might be making music without having earned the title of musician.

                Those two are kind of linked though right? Musicians make music, if I managed to create a piece of music on an instrument or my voice or through a production system then by all rights I am a musician.

                I suppose my personal distinction is that if I just prompted a song, then I all i did was make a prompt, the software made the music and then the question is should we value that? I would say no, especially when we have just hundreds of thousands of hours of music that humans have created and are creating now.

          • who said that "beauty is in the eye of the beholder"... ?
            That's where we are at, in this thread.
            i.e. there is no agreed upon standard of what makes art.. or what makes art good or bad.
            • i.e. there is no agreed upon standard of what makes art.. or what makes art good or bad.

              Indeed, but that doesn't mean anything is art.

          • Appreciation of art certainly is. I like the banana and duct tape thing. I thought it was playfully sparking discussion, but of course that only works because make people think it's utter garbage. One could (indeed I am) arguing that it has artistic merit precisely because art is subjective.

            Not in a some think it's good done think it's bad way, but it's merit depends on different opinions on it existing.

            As for my hypothetical daughter (I have no kids but do have niblings which produce plenty of---in the wor

        • However, mentioning AI sure gets the conversations going more than a banana on the wall.
    • by gweihir ( 88907 )

      Programming is literally about exacting rules and deterministic outcomes. It's exactly the sort of thing I'd expect a computer to be good at.

      That is because you have no clue what you are talking about.

    • by Shaitan ( 22585 )

      Actually this is an extremely bad outcome when you understand what is happening in the industry.

      Developers used to be a small subset of the tech world. You also had security, helpdesk, and engineering/administrators, also architects, etc. With the exception of helpdesk at least 95% of the senior jobs in those areas are asking for 5-7 yrs of unrelated development skills like devops and ci/cd even where the jobs normally would require only light opportunistic scripting.

      There are some cases where these skills

  • by Anonymous Coward on Saturday March 22, 2025 @09:09AM (#65251869)
    There'll be growth in programmer jobs to fix up all the crap AI caused due to gullible managers and C-levels that bought into today's hype.
    • Re: (Score:2, Insightful)

      by ranton ( 36917 )

      There'll be growth in programmer jobs to fix up all the crap AI caused due to gullible managers and C-levels that bought into today's hype.

      No, there will be growth in software developer jobs to fix the crap AI is generating. By BLS definitions, programmers wouldn't be qualified or capable of that work. That is why software developer jobs are still growing while programmer jobs are dropping.

      • that's not how automation works. The companies will test the automation, find what works and what doesn't, keep what works and move on.

        Like how Tesla tried to build fully automated factories (mostly as a pump and dump but I digress) and they were told it wouldn't work, it didn't work, and they stopped doing it and automated what they could.

        We didn't get more jobs, we still lost jobs because over time Tesla and every other car company has managed to automate a few more things (there's some complex we
  • by cascadingstylesheet ( 140919 ) on Saturday March 22, 2025 @09:18AM (#65251885) Journal

    While software developer positions have remained stable with only a 0.3 percent decline, programmers who perform more routine coding from specifications provided by others have seen their ranks diminish to levels not seen since 1980.

    And there is the money shot, so to speak.

    Code monkeys, who basically just mechanically implement code, can be replaced by engines. Just as horses could be replaced by ICE engines.

    If anyone is surprised by that, they really shouldn't be.

    • by gweihir ( 88907 ) on Saturday March 22, 2025 @10:18AM (#65251987)

      Indeed. Those that are affected are not really programmers. They are more like data typists.

      • You're going to lose what you have soon. They'll come for it. Your home. Your car. They'll take it.

        And when they do somebody they haven't gotten around to yet will write the exact same post about you.
        • You're going to lose what you have soon. They'll come for it. Your home. Your car. They'll take it. And when they do somebody they haven't gotten around to yet will write the exact same post about you.

          Code monkey hands typed this post.

        • by gweihir ( 88907 )

          I understand what you are trying to say, but you are entirely missing your target in my case. The fact of the matter is that we have tons of marginal and bad coders, and that is a massive problem. It has to stop. The other part is that any real, competent coders are not threatened at all.

    • For now (Score:5, Insightful)

      by abulafia ( 7826 ) on Saturday March 22, 2025 @10:52AM (#65252051)
      More senior roles are safe for now, especially in firms where code actually matters. (By "matters", I mean consequences of robot hallucinations that result in a business losing money, compared to saying something weird in your Facebook spam or generating obviously fake homework or non-con nudes with too many fingers.)

      But the thing is, these tools do not become worse over time, and this is early days. I expect raw LLM coding capability to top out somewhere, at which point more gains will come from tilting the field more towards the robots - as libraries, tools and protocols slowly shift to being robot-generated, I'd expect humans are going to do worse with them - robots just do not code like we do and in particular do not make mistakes like we do.

      I don't expect a lot of problems as the current crop of graybeards goes away - this is my cohort. The folks coming up behind us are capable and smart, they'll be fine. But I wonder where the next generation of troubleshooters of last resort are going to come from, if everyone is learning from/coding with robots. There is a lot of low-level code out there that balances real-world tradeoffs that aren't obvious - sometimes there's no right answer. LLMs can't plagiarize about these things, because aside from the code, in many cases nobody has written their reasoning down beyond "tweaked the bad thing" in a commit message.

      (As one example - go try to find any documentation about the reasoning behind fsync batching in ext4. You can see what the code does, you can see the history of changes, but I don't think anyone but the authors (assuming they remember) know exactly what considerations made it the way it is. And this is extremely salient code - I'd be willing to bet there are hundreds of people trying to squeeze a bit more juice out of their Mysql install as I type this.)

      There's always some unreasonable person who wants to do things the hard way, so we'll have domain experts in the age of robots. And I'll likely be dead before it is a real problem, so what do I care? But folks who assume "it'll be fine, we'll figure it out" should look to historic examples of abrupt knowledge loss in industries - usually war shocks, sometimes disasters or economic attacks short of war. You can't just stop and restart these things on demand like a pump; kill an economic ecosystem and you might get something else kinda vaguely shaped like it later, but the dead one doesn't come back.

      • The thing is, even if it does get cheaper to make software, the cost of software won't go down, I'm sure of it. Everything will become cloud subscriptions and customers will end up paying more, while a good chunk of the delopment process will be automated. I see being a "Non-AI/cloud" company eventually becoming an actual value proposition.
      • OTOH, out of curiosity I spend some time asking claude.ai about a project I've been working on for a few months. The project requires implementing some moderately tricky GlobalPlatform specifications, which interact in subtle and non-obvious ways. It's clear that the authors of the specs didn't really think about how a system that needed to support multiple specs would operate, how it would reconcile the points of disagreement. They also didn't think much about the practical issues involved, especially pe

  • don't stop buying our tech stocks we are not having a recession. that is not why we are laying people off. no not at all. normally laying people off would mean we are in a recession but we are replacing people with AI. that is great you should buy more of our stock because we are more efficient. think about the future and all the amazing stuff we will do. dont look at what we have actual done recently the future is so bright.
  • I think this is more to do with economic factors than AI taking over. You can't just replace programming jobs with AI, because you still need someone who understands programming to command the AI what to do. Otherwise you'd end up with managers writing "make my software prodcut better than all the competitors". I accept that maybe some of the most menial jobs could be replaced by a highly skilled programmers who can just crank out all the boiler plate code through an AI, which might result in a dip in junio
    • I believe there's a confluence of factors. A perfect storm, if you will.

      During COVID companies hired devs in troves. After the pandemic, they readjusted and started to lay off thousands of them.

      That, associated with the AI craze, is now making companies get even more stricter about it. They're definitely pushing for a near future with the need of even less devs.

    • by ranton ( 36917 )

      You can't just replace programming jobs with AI, because you still need someone who understands programming to command the AI what to do. Otherwise you'd end up with managers writing "make my software product better than all the competitors".

      The BLS defines the people who will command the AI what to do as software engineers, not programmers. Just like those software engineers tell the programmers what to do today.

      Programmer jobs dropped 30% from 2020 to 2022, before ChatGPT kicked off the latest AI craze. The software developer profession added nearly 10 "developers" for every "programmer" lost. The move away from hiring code-monkeys to hiring more skilled software developers / engineers has been a long term trend that has nearly nothing to do

    • Otherwise you'd end up with managers writing "make my software prodcut better than all the competitors".

      Managers are not that smart. They'd ask the AI, "print me more money".

    • by djb ( 19374 )
      It doesn’t need to, if AI can double my productivity as a developer, companies only need half as many developers. Once supply overtakes demand you can say goodbye to those crazy salaries we’ve been getting for the last couple of decades.
  • by jrnvk ( 4197967 )

    Most of those jobs never existed in the first place, and the ones that do are always looking for cheaper labor (outsourcing or H-1B).

    Until we have a political party that ends that gravy train, we will see more of the same.

  • Answer: not many

    The linked data in the article lists 2023 estimates of 120 thousand programmers and 1.6 million developers.

    So "Programming jobs" make up a small minority of jobs involving coding.

  • You could not twist yourself up in knots any more to avoid saying the words GitHub Copilot. Keep living in denial slashdot
  • This seems more likely to be a job reclassification issue. Programmers and software developers are essentially the same job, and the narrow difference the BLS could easily lead surveyors and researchers to shift slightly in their definitions from year to year. Programmer + Developer employment grew 2.55% YOY in the latest BLS data. The job outlook for developers is still 18% growth over the next decade.

    Even a year ago there were 9% as many programmers as there were software developers, so while this 25% dro

    • The "engineers" I read so much about on /. are gonna be ok?

      • by ranton ( 36917 ) on Saturday March 22, 2025 @10:12AM (#65251973)

        The "engineers" I read so much about on /. are gonna be ok?

        Yes, they will be. Software development will continue to grow as a profession until AI can already do 95%+ of ALL white-collar work. Only a tiny fraction of other white-collar work requires as much creative problem solving as software development. There will be 50%+ unemployment in the workforce before the software development field starts shedding jobs.

        • Um, did you even bother reading the summary? Let alone the article. The profession is clearly shrinking.

          You can try and "no true Scottsman" your way out of this I guess. That's one way to cope...
          • by ranton ( 36917 )

            Um, did you even bother reading the summary? Let alone the article. The profession is clearly shrinking.

            The summary, and article, are very confusing for someone who doesn't understand how the US Bureau of Labor Statistics classifies these jobs. The combination of Software Developer and Programmer jobs (two different occupations according to the BLS) have grown nearly 3% over the past two years. Only the jobs they classify as "programmer" have seen a decrease. But BLS reports there are over 10x as many software developer jobs, and jobs in that occupation are growing rapidly. Almost 10x as many software develop

    • Exactly this; there's no story here except that hiring of *junior* programmers is down.

  • by reinke ( 230131 ) on Saturday March 22, 2025 @10:20AM (#65251993)

    Yes, AI is impacting IT jobs. We are seeing it first hand. In our case, it is with graphic design. Situations where we would have hired approximately 1 years worth of coop student work, we are instead assigning to an existing IT professional that is using tools such as Blender (with blenderkit), Gencraft, and NightCafe, With quite acceptable results.

    The issue is - where will this leave us in 20 years? Our highly experienced individuals are reporting they are able to use AI as a search engine, BUT, AI cannot solve problems. Need to manage a cache of information? Do you use redis? In memory heap? In memory array? Or is it even worth it, and you can just reach in to the authoritative SQL store? AI cannot make those decisions for you. And if cut off at the knees the job opportunities for junior developers, we may risk eventually running into a shortage of highly skilled resources that are adept at architecture and design of new software.

    My take? AI is a great tool for the inclined. It is a horrible crutch for the lazy. And at scale, the jury is out on whether or not it will be beneficial to advancement of technologies.

  • I have no doubt that AI is affecting coding jobs, but the timing imagined here is off. More, weâ(TM)re seeing the tail end of the devaluation of individual contributors weâ(TM)ve watched for the last few decades. Itâ(TM)s more of the MBA fever dream that imagines a worldwide labor market with like for like candidates and 7-layer deep outsourcing that, even after all seven layers make a profit, somehow remains cheaper than just keeping talent and doing things in-house. AI will continue to make

    • I have no doubt that AI is affecting coding jobs, but the timing imagined here is off.

      Or this is mostly an effect of the rebound from irrational exuberance hiring at the start of the pandemic.

  • But, but I was told by Slashdot.org experts that AI is just a scam and useless.... Experts what happened?
  • Reappear overseas.
  • Golly Boys. Have seen me a lot of developers in the past 25 years, I have seen about 3 written specs to a level that the project could be completed by a spec as laid out in version 1.01B as delivered when the first round of programers were hired. I have been parts of many projects where the specification did not get closed until the application V11 3rd quarter update was accepted, and someone said we need to close out the current invoicing series.
  • A few points:

    * The (linked) article notes that the Bureau of Labor Stats makes a distinction between "computer programmer" and "software developer". Why? It wasn't fully explained in the article. In any case, the earlier is declining in employment and the latter is increasing. What was also not addressed; is the correlation between the two scalar invariant? That is, is the decline in employment of programmers equivalent to the increase in developers?

    What was not mentioned in the article: a comparison of var

    • Here is the BLS explanation, at least the quick one:

      https://www.bls.gov/ooh/comput... [bls.gov]

      Computer programmers write, modify, and test code and scripts that allow computer software and applications to function properly.

      Software developers design computer applications or programs. Software quality assurance analysts and testers identify problems with applications or programs and report defects.

      • Thanks -- think that was mentioned in the article. What was not: why make this distinction, regarding data collection? Intuitively, they're describing the same class of jobs, right?
        • My (unfounded) opinion is that these categories were devised right out the mid 90's to the 'aughts and are in need of some modernization.

          Also I believe the BLS gets all these stats via survey so a lot of this I think is people self-reporting their job's and BLS has to put them into buckets so they can only be as accurate as to what people provide.

          My (again unfounded) when I think of someone who isn't a software dev but a "programmer" in todays world isn't so much a code monkey but like an old school guy who

      • Judging only by those two definitions, it sounds as though what's now called a software dev is hte same thing as was called a systems analyst back when I was learning about computers. The only difference is that there's an awful lot more of them now. Job title inflation?
  • The headline is really misleading. It's all about confusing gummint (specifically the Bureau of Labor Statistics) job classifications.

    Programmers: code monkeys who convert specifications into applications. Those jobs are getting crushed by Copilot and ChatGPT.

    Software developers: people like me who ask customers "what problem are you trying to solve?", write the specifications, and when I get lucky, write the code. Software developers are apparently doing fine. There are a lot more of them than programmers.

  • Infinite monkeys with keyboards have written all of the software. They've written themselves out of jobs.

  • It's really about the change from the term "programmer" to "software developer" in job titles.

  • AI, while it may in the next decade, isn't cutting jobs. If anything, Telecommuting is. and when I mean telecommuting, I mean halfway across the world.

    The Reason H1B's exists, and the reason you have politicians on both sides of the isle fighting tooth and nail to keep them even though it displaces high paid American workers out of jobs, is because even though it's a job held by a foreign employee, it still counts as an American Job on paper complete with it's tax revenue. Simply put, if H1B's disappeared

  • I have a suspicion that the bulk of so-called programmers are just building websites, a task that can easily be automated and also offshored.

  • Most of us who read the headline, immediately equated "programmers" with "software developers." But reading the article makes it clear that "programmer" jobs have declined, but "software developer" jobs have not.

    The source cited lists the number of "programmer" jobs as being a total of 120,000 https://www.bls.gov/oes/curren... [bls.gov], while there are 1.7 million "software developer" jobs. https://www.bls.gov/ooh/comput... [bls.gov]

    So while a dip in programmer jobs of 27% sounds like a lot, if you combine "programmers" with

  • Between open source and cloud services many pre-canned solutions cover more use cases than they did a decade ago. Proprietary solutions spill into the public domain and most business activities are not supremely unique, so even niche domains eventually have either open source, managed or closed source solutions available at a cheaper price than hiring a developer. Thus software engineering is mutating into data engineering and systems engineering. You still have to know how to code and how a computer works,
  • This has little to do with AI and and everything to do with Section 174(c)(3) in the tax law. https://www.corumgroup.com/ins... [corumgroup.com] First, this caused companies to hire less developers, in particular startups that couldnâ(TM)t take the tax bill. Second it caused women companies to reclassify people as other roles since calling them software developers was no longer a write off. The tax law is 90% of the jobs lost, AI is 10% at best.
  • hence less money available for other ventures that would have hired programmers.

    That's the nature of bubbles, it attracts more money than it merits, leaving less for other activities.

    When AI cannot deliver the astronomical returns the current valuation suggests, the music ends and the bubble bursts, just like in housing bubble in 2008 and the dotcom bubble before it.

  • Triumph! Acclaim! for the new (dear) Leader.
  • At first glance ("programmers" down 25%), I was like "that can't be true". Its so rare to distinguish between programmers and developers in the industry, in a bit surprised the BLS has good statistics on the difference.

A list is only as strong as its weakest link. -- Don Knuth

Working...