LLMs are very useful tools for software development, but focusing on employment does not appear to really dig into if it will automate or augment labor (to use their words). Behaviors are changing not just because of outcomes but because of hype and expectations and b2b sales. You'd expect the initial corporate behaviors to look much the same whether or not LLMs turn into fully-fire-and-forget employee-replacement tools.
Some nits I'd pick along those lines:
>For instance, according to the most recent AI Index Report, AI systems could solve just 4.4% of coding problems on SWE-Bench, a widely used benchmark for software engineering, in 2023, but performance increased to 71.7% in 2024 (Maslej et al., 2025).
Something like this should have the context of SWE-Bench not existing before November, 2023.
Pre-2023 systems were flying blind with regard to what they were going to be tested with. Post-2023 systems have been created in a world where this test exists. Hard to generalize from before/after performance.
> The patterns we observe in the data appear most acutely starting in late 2022, around the time of rapid proliferation of generative AI tools.
This is quite early for "replacement" of software development jobs as by their own prior statement/citation the tools even a year later, when SWE-Bench was introduced, were only hitting that 4.4% task success rate.
It's timing lines up more neatly with the post-COVID-bubble tech industry slowdown. Or with the start of hype about AI productivity vs actual replaced employee productivity.
To me it seems that LLMs are a tool that only increase productivity for given headcount in dimensions that were neglected in the past.
For example, everyone now writes emails with perfect grammar in a fraction of a time. So now the expectation for emails is that they will have perfect grammar.
Or one can build an interactive dashboard to visualize their spreadsheet and make it pleasing. Again the expectation just changed. The bar is higher.
So far I have not seen productivity increase in dimensions with direct sight to revenue. (Of course there is the niche of customer service, translation services etc that already were in the process of being automated)
I've got a few buddies over at Microsoft, they've all said something along the lines of "I really hate using copilot. They at least let us use pre-approved models in VSCode, we get most that come out. But all AI metrics are tracked and there are layoffs every quarter. I have kids now man. Strange times. I know you would have quit months ago" and they're right.
Now that bs work has next to no cost, I see a lot more bs work being done, and often on pointless bureaucratic activities involving generating questionnaires and answering them. It's as if the activities add up to a big net zero.
I think mRNA vaccines and green energy are equally transformative economic opportunities. In the US though we are becoming a one trick pony. Instead of investing in all 3, we will prioritize AI because Silicon Valley sucked up to Trump in the recent election.
All to say we could have quite a bit more resilience as an economy, but we decided to sacrifice our leadership in these areas.
The 10% reduction in hiring for young workers is entirely because industry (software, manufacturing) at least in the united states (and probably the world) is contracting and in recession, while the services and government sector has been the main sector growing since a long time now - completely due to economic and geopolitical reasons, nothing to do with AI.
I had a team of developers and essentially told them all 'either learn to code with Claude' or you're out. What I found is the more junior developers started 'vibe coding' resulting in a net decrease in performance, where the more senior ones used it to accelerate their speed cautiously and selectively.
My conclusion was senior engineers were better because they were used to managing developers and taking on more managerial tasks building 'LLM Soft Skills' and also frankly fixing mistakes, the junior developers were pressured for speed and had their managers to correct them.
Within 12 months, despite extensive attempts, only the mid level team members remained.
The problem with your reasoning is that its discriminatory in the parameters you set.
When you set impossible constraints that neglect requirements for sustainability, and tell people to do the impossible, you pigeonhole and sieve only the people you are actually looking for (the ones that can meet that sieve).
The lying, and deceit that you do, naturally occur after-the-fact (which blind people don't notice, often making them evil). The danger of most deceit and lying occur where you say something truthful omitting something important that they know, but then later contradict yourself in the outcome. These are called lies of omission. Deceivers and Vipers take full effect of these actions pretending its not them, its the circumstance, but a circumstance they control.
Of course the middle team would be the only ones left, after all you tortured your senior team having them baby a LLM driving them to burnout, and the junior team lacked the knowledge that makes the difference between Junior and Mid, to be productive. You set a filter that only your midlevel team could meet, and any conclusions you make will equally conform to your initial decisions in the criteria you set which is you don't want to hire young people, or old people. You want them just right, and there are laws against age discrimination. People have tried to get around these laws for years, and have never had more luck in evading these than now; with the advent of blackbox AI which can obscure these type of decisions through hiding them in the weights. Short term, there will of course be more profit, long-term the lawsuits will get you, and your good character which you thought you had will not be good.
Additionally, you have your perfect team left that is wholly dependent on LLM, who won't be able to solve the rare but inevitable problems the senior level people would (before they occur).
Is 2 years or so big enough sample size for any conclusions? You’re also seeing massive money movements; last quarter larger than consumer spending (!!!) This money isn’t going to junior head counts(labor), it’s going to compute(capital) Also what’s everyone’s balance sheet really saying? Will this money movement to capital rather than labor ACTUALLY pay off ? I think that would take a 10year hind sight to prove no?
>early-career workers (ages 22-25) in the most AI-exposed occupations have experienced a 13 percent relative decline in employment even after controlling for firm-level shocks.
As a professor of software engineering, I'm feeling an existential crisis coming on. Are we preparing students in vain? Last term was the first time I had senior project students who didn't have a job lined up in the Fall of their final year. Maybe it's time to retire.
I'm curious what percentage of the "tech industry" is in fact the classic dumb money, and what they have their developers doing is basically fantasy nonsense for the owner(s). I know of several very good sized tech companies, purchased by Saudi Princes, for nothing other than their bragging rights and their staff are an odd mix of their friends, the original tech team, and then an odd series of fantasy vanity projects. Remember 15 years ago when mounting cameras on a person's face for "face captures" during gameplay was a thing? Several of those companies are now Saudi playpens, with over worked developers making demos on deadlines that are just for bragging rights and not products at all.
The first figure alone is insane, for me it helps explain why some friends feel like the market has fallen out from under them (early career) while others aren't having such a tough time. It seems like so far, for folks 30+ AI hasn't really changed things radically (yet).
The series in figure 6, though, I think suggests that we may just be seeing a time-delayed effect and eventually everyone's going to be impacted.
> The first figure alone is insane, for me it helps explain why some friends feel like the market has fallen out from under them (early career) while others aren't having such a tough time. It seems like so far, for folks 30+ AI hasn't really changed things radically (yet).
Not just early career, also independent devs who are any from early-career to late-career.
I did a thing for a client a few weeks ago (embedded, with prototype board + code for industrial sensor containing multiple sensor types, using wifi to both configure and monitor the device, and to retrieve sensor values).
They balked at a two-week bill; their argument was that this should have not been more than a 2-day bill (literally, 2 days for everything from soldering up the prototype board to writing the code).
Their frame-of-reference was that their in-house dev (or similar, not sure now) could do that in two days with an arduino or ESP32 devkit.
My suspicion is that because their in-house dev took only two days to get claude code to write a ping program for a esp32 dev board (no soldering, sensors, etc. Purely WiFi comms and nothing else), their expectations were that this should have been the same.
(I eventually accepted the payment for only 2 days worth of dev, of which at least half was expenses for me - driving, meetings with them, purchases, etc.)
It is good that people are paying attention to this problem, but unfortunately not in time to correct the problem before consequences get bad.
Hysteresis is a bitch.
This study showed that early career workers of which they only focused on the 22-25 range had a ~20% drop in employment between 2022 and now. If you include the 26-30 range which includes most early-career that's roughly ~30% less jobs, from the Payment Processors perspective.
The study doesn't seek to cover other impactors such higher costs on the labor pool, and interference in employment matching, which are also of great concern.
30% after shock normalization is well beyond statistical significance. This is happening, people said it would happen, and no one acted to stop it because they listened to evil people seeking short-term profit; blind to all else.
Sad and dark times are ahead. There are things that can be reasonably predicted ahead-of-time, but the moment you give preferential treatment to liars is the moment you lock in losses. Sure the data proving the prediction will come, but not in time to take corrective action; such is the structured cascading failures involving hysteresis.
So what they actually found is that there is a decline in employment of young people's decline in AI-exposed industries. There are thousands of confounding variables that could have caused this, and they tested only a few. This is by no means a conclusive evidence that AI is taking away entry level software jobs.
18 comments
[ 0.18 ms ] story [ 41.9 ms ] threadSome nits I'd pick along those lines:
>For instance, according to the most recent AI Index Report, AI systems could solve just 4.4% of coding problems on SWE-Bench, a widely used benchmark for software engineering, in 2023, but performance increased to 71.7% in 2024 (Maslej et al., 2025).
Something like this should have the context of SWE-Bench not existing before November, 2023.
Pre-2023 systems were flying blind with regard to what they were going to be tested with. Post-2023 systems have been created in a world where this test exists. Hard to generalize from before/after performance.
> The patterns we observe in the data appear most acutely starting in late 2022, around the time of rapid proliferation of generative AI tools.
This is quite early for "replacement" of software development jobs as by their own prior statement/citation the tools even a year later, when SWE-Bench was introduced, were only hitting that 4.4% task success rate.
It's timing lines up more neatly with the post-COVID-bubble tech industry slowdown. Or with the start of hype about AI productivity vs actual replaced employee productivity.
For example, everyone now writes emails with perfect grammar in a fraction of a time. So now the expectation for emails is that they will have perfect grammar.
Or one can build an interactive dashboard to visualize their spreadsheet and make it pleasing. Again the expectation just changed. The bar is higher.
So far I have not seen productivity increase in dimensions with direct sight to revenue. (Of course there is the niche of customer service, translation services etc that already were in the process of being automated)
Think like a forestry investor, not a cash crop next season.
All to say we could have quite a bit more resilience as an economy, but we decided to sacrifice our leadership in these areas.
My conclusion was senior engineers were better because they were used to managing developers and taking on more managerial tasks building 'LLM Soft Skills' and also frankly fixing mistakes, the junior developers were pressured for speed and had their managers to correct them.
Within 12 months, despite extensive attempts, only the mid level team members remained.
When you set impossible constraints that neglect requirements for sustainability, and tell people to do the impossible, you pigeonhole and sieve only the people you are actually looking for (the ones that can meet that sieve).
The lying, and deceit that you do, naturally occur after-the-fact (which blind people don't notice, often making them evil). The danger of most deceit and lying occur where you say something truthful omitting something important that they know, but then later contradict yourself in the outcome. These are called lies of omission. Deceivers and Vipers take full effect of these actions pretending its not them, its the circumstance, but a circumstance they control.
Of course the middle team would be the only ones left, after all you tortured your senior team having them baby a LLM driving them to burnout, and the junior team lacked the knowledge that makes the difference between Junior and Mid, to be productive. You set a filter that only your midlevel team could meet, and any conclusions you make will equally conform to your initial decisions in the criteria you set which is you don't want to hire young people, or old people. You want them just right, and there are laws against age discrimination. People have tried to get around these laws for years, and have never had more luck in evading these than now; with the advent of blackbox AI which can obscure these type of decisions through hiding them in the weights. Short term, there will of course be more profit, long-term the lawsuits will get you, and your good character which you thought you had will not be good.
Additionally, you have your perfect team left that is wholly dependent on LLM, who won't be able to solve the rare but inevitable problems the senior level people would (before they occur).
As a professor of software engineering, I'm feeling an existential crisis coming on. Are we preparing students in vain? Last term was the first time I had senior project students who didn't have a job lined up in the Fall of their final year. Maybe it's time to retire.
The series in figure 6, though, I think suggests that we may just be seeing a time-delayed effect and eventually everyone's going to be impacted.
Not just early career, also independent devs who are any from early-career to late-career.
I did a thing for a client a few weeks ago (embedded, with prototype board + code for industrial sensor containing multiple sensor types, using wifi to both configure and monitor the device, and to retrieve sensor values).
They balked at a two-week bill; their argument was that this should have not been more than a 2-day bill (literally, 2 days for everything from soldering up the prototype board to writing the code).
Their frame-of-reference was that their in-house dev (or similar, not sure now) could do that in two days with an arduino or ESP32 devkit.
My suspicion is that because their in-house dev took only two days to get claude code to write a ping program for a esp32 dev board (no soldering, sensors, etc. Purely WiFi comms and nothing else), their expectations were that this should have been the same.
(I eventually accepted the payment for only 2 days worth of dev, of which at least half was expenses for me - driving, meetings with them, purchases, etc.)
Hysteresis is a bitch.
This study showed that early career workers of which they only focused on the 22-25 range had a ~20% drop in employment between 2022 and now. If you include the 26-30 range which includes most early-career that's roughly ~30% less jobs, from the Payment Processors perspective.
The study doesn't seek to cover other impactors such higher costs on the labor pool, and interference in employment matching, which are also of great concern.
30% after shock normalization is well beyond statistical significance. This is happening, people said it would happen, and no one acted to stop it because they listened to evil people seeking short-term profit; blind to all else.
Sad and dark times are ahead. There are things that can be reasonably predicted ahead-of-time, but the moment you give preferential treatment to liars is the moment you lock in losses. Sure the data proving the prediction will come, but not in time to take corrective action; such is the structured cascading failures involving hysteresis.