Evidence that AI is destroying jobs for young people

The 2022 drop for SWE is easy for me to explain, and it's not on these analysts' list of factors (though I'm not an economic quant, I don't know how you could really control for it): In 2017, a tax bill was passed that cut a particular tax incentive in 2022 in an effort to be counted as "revenue neutral" despite being otherwise a massive tax cut overall. The incentive in question was a writeoff for "Research and development". This means that in 2022, it got effectively much more expensive to hire anyone who falls under that category, including developers not directly necessary for the day-to-day function of a business (hell, one might argue they would have counted anyway) and scientists of most kinds. That this hit big firms, which have a higher relative amount of R&D efforts going at a given time, first makes a lot of sense.

For customer service, my explanation is that companies literally do not care about customer service. Automated phone trees, outsourced call centers whose reps have no real power to help a customer, and poorly-made websites have been frustrating people for decades, but businesses never seem to try to compete on doing better at it. It's a cheap win with investors who want to hear about AI initiatives to lay off yet even more of this department, because it doesn't matter if the quality of service declines, there are no market or regulatory forces that are punishing this well enough to ever expect firms to stop breaking it, let alone fix it

I do consulting, I'm constantly scouting clients. Right around November 2022 something very stark happened. I went from fighting off prospects with a stick, to crickets, almost over night. I deal mostly with startups and mid-size companies, nobody with insider knowledge or cutting edge interests. I can tell you that GPT was not heavily on anyone I dealt with's radar as an opportunity to reduce costs.

Some sort of cultural zeitgeist occurred, but in terms of symptoms I saw with my own eyes, I think ZIRP ending (projects getting axed) and layoffs starting (projects getting filled within ~24 hours) were huge drivers. I have no proof.

I had the same thoughts, there are clearly indicators that the weakness in the labor market started happening before LLMs and AI took over popular discourse.

All the more reason to believe that while correlated, LLMs are certainly not the largest contributor, or even the cause of the job market weakness for young people. The more likely and simple explanation is that there are cracks forming in the economy not just in the US but globally; youth employment is struggling virtually everywhere. Can only speculate on the reasons, but delayed effects from questionable monetary and fiscal policy choices, increasing wealth gaps, tariffs, geopolitics, etc. have certainly not helped.

>The paper notes they did an adjustment for the end of ZIRP. I dont know enough econometrics to understand whether this adjustment was sufficient

Looking at the paper [0], they attempted to do it by regressing the number of jobs y_{c,q,t} at company c, time t, and "AI exposure quintile" q, with separate parameters jointly controlling for company/quintile (a), company/time (b) and quintile/time (g). This is in Equation 4.1, page 15, which I have simplified here:

log(y_{c,q,t}) ~ a_{c,q} + b_{c,t} + g_{q,t}

Any time-dependent effects (e.g. end of ZIRP/Section 174) that would equally affect all jobs at the company irrespective of how much AI exposure they have should be absorbed into b.

They normalized g with respect to October 2022 and quintile 1 (least AI exposure), and plotted the results for each age group and quintile (Figure 9, page 20). There is a pronounced decline that only starts in mid-2024 for quintiles 3, 4, and 5 in the youngest age group. The plots shown in the article are misleading, and are likely primarily a reflection of ZIRP, as you say. The real meat of the paper is Figure 9.

A potential flaw of this method is that ZIRP/Section 174 may have disproportionately affected junior positions with high AI exposure, e.g. software engineers. This would not be accounted for in b and would thus be reflected in g. It would be interesting to repeat this analysis excluding software engineers and other employees subject to Section 174.

[0] https://digitaleconomy.stanford.edu/wp-content/uploads/2025/...

Yeah my company started stepping up outsourcing in 2023. We also started some AI projects. The AI projects haven't made much progress but the outsourcing is at an extremely advanced stage.

I personally sat in meetings in 2022 where we adjusted staffing projections in anticipation of AI efficiency. Sure some of it was "overhiring," but the reality was that those staffing goals were pre-ai. Once they were updated, that's when the layoffs started because management didn't want anyone who didn't have an AI or big data background.

in addition to this detail I might add I can't remember the last time I had a customer service call that took place with someone stateside. It's easy to point to AI when offshoring for favorable interest rates is really the reason.

I remember years of a no-backfill policy at Devon on a promises of automation. Since 2017 at least. The “desirable job market” for young people has been challenging well before LLM became popular. Want a dead end entry level job in food service earning $20/hour? No problem.

Wow, that's hilarious. So essentially hiring could be identical across all age groups, but due to a glitch in the analysis (young people don't stay young, who knew?), it appears that younger people are losing jobs more than the rest.

I'm not sure I understand. Your model shows that different group buckets (eg 20-24yo vs 25-29yo) peak at different years (in your figure, 2022 vs 2024) despite being driven by the same dynamics. Is that expected? I (naively?) expected the same groups to rise, fall and have peaks at the same times.

Also AI hype is sucking all capital out of traditional hiring

It's only a matter of time before AI becomes a commodity. The open source models are just a generation behind the proprietary ones and they're almost good enough for most users. Even if we're somehow limited to using proprietary models, since all of these "AIs" are based on natural language so you can practically hot swap the LLMs. Any company banking on selling any service other than AI hosting is going to be incredibly disappointed.

The only thing that could stop this commodification is some sort of vendor lock-in, but that looks to be technically challenging.

And the data for the two decades before that should be tossed out because of the global housing crisis, the sovereign debt crisis, and all the reverberations from that.

And the data for the two decades before that obviously needs to be tossed out because of the one-off nature of the dot-com boom, the dot-com crash, 9/11, and so on.

And before that... point is, you don't get clean data in economics. There's always something big going on, there are no double-blind trials to run, etc. It's called the dismal science for a reason. But that doesn't make it useless.

> 5 years from now, the AI hype will be over, and employers will be short on grads*

alterative view: the AI hype is real, AI takes over, and no one has any jobs anyway

also a thought: in 5 years the boomers will be retiring in droves as will the first series of GenX and the market, in most fields, should be opening up anyway

what makes you so confident that AI is just a hype?

This time is different. A fact right now is that software engineers now can orchestrate LLMs and agents to write software. The role of software engineers who do this is quality control, compliance, software architecture and some out of the box thinking for when LLMs do not cut it. What makes you think advances in AI wont take care of these tasks that LLMs do not do well currently? My point is once these tasks are taken care off a CS graduate won't be doing tasks that they learnt to do in their degrees. What people need to learn is how to think of customers needs in abstract ways and communicate this to AI and judge the output in a similar way someone judges a painting.

Thanks! Macroexpanded:

AI adoption linked to 13% decline in jobs for young U.S. workers: study - https://news.ycombinator.com/item?id=45052423 - Aug 2025 (629 comments)