New evidence strongly suggest AI is killing jobs for young programmers

It's a brutal time to be a recent computer science graduate.

Last fall, I started hearing that demand for entry-level programmers was in free fall thanks to competition from AI. I decided to do some reporting to figure out if this was true.

One of my first calls was to Nicholas Bergson-Shilcock, CEO of the Recurse Center, which provides space for programmers to upgrade their skills and then connects them with potential employers.

“Starting in late 2022, it's been the worst market for hiring engineers since we started” in 2011, Bergson-Shilcock told me. He attributed the initial decline mainly to macroeconomic factors: the Fed raised interest rates in 2022 to ward off inflation, which led to a wave of Big Tech layoffs and less venture capital funding for startups.

When I talked to him in November, two years after that initial wave of layoffs, demand for programmers still hadn’t recovered. He thought this was “in part attributable to AI” because “companies, particularly early-stage companies that might have hired three to five people, now they're hiring one or two people and expecting them to get more done.”

As interesting as his prespective was, I didn’t feel comfortable writing about it without some hard data to back it up. And nobody could point me to clear evidence linking the weak market for programmers to AI adoption.

In May, I checked in with economists to see if there was new evidence of AI-driven job losses. But once again I didn’t feel like I’d gained enough clarity to merit a write-up.

Finally this week, I emailed Nicholas Bergson-Shilcock for an update.

“We've seen a big resurgence in 2025,” he responded. “The tech recruiting market is very much ‘back.’”

A couple of recent studies seemed to back this up. One from Stanford economist Bharat Chandar found that programmers had seen significant job growth over the last year. Another from the Economic Innovation Group found that across the economy, occupations that were more vulnerable to automation from generative AI did not seem to be suffering from higher unemployment rates or other negative outcomes.

So I decided to finally write an article about this topic. By Tuesday afternoon, I was putting the finishing touches on a piece arguing that the best available evidence suggested AI was probably not driving recent job losses in software development or other white-collar occupations—though I acknowledged there was quite a lot of uncertainty about this.

As I was getting ready to publish that article, I noticed that Chandar had just published another paper—this time with two colleagues. I started reading their study, and when I got to this chart, my heart sank. Because I immediately knew I’d have to re-write my article from scratch.

The dotted vertical line is set to November 2022, the month that ChatGPT was released. And the lines are normalized so that they’re all equal to 1 in that month.

This chart doesn’t settle the debate over whether AI is undercutting demand for early-career programmers, but it gives us a much clearer picture of what’s happening. And that picture seems entirely consistent with the “blame AI” thesis.