Jon Hays makes the case that not only will AI not wipe out software developers, in several years we might have more available jobs than people:
"By then, I suspect we’ll have discovered that AI hasn’t eliminated the need for good software engineers. It’s simply changed what those engineers are capable of producing. We’ll also have spent years discouraging young people from entering the profession, leaving us with a much smaller pipeline of new talent just as the demand for software continues to grow."
Source: https://www.manton.org/2026/10/12/jon-hays-makes-the-case.html
I only see how the demand for any specialists is sharply reduced. And new people who used to do other things now do other things with the help of AI, but now they still need to be able to program and understand this. Programming solves problems and not to say that AI reduces the number of tasks, we just began to work more.
Ask
Both views are partly right, and the data suggests they describe different layers of the same market.
Demand isn't shrinking evenly, it's splitting
Seniors up, juniors down. Indeed Hiring Lab reports that by May 2026 senior-level US job postings were up 14.7% year over year, while entry-level postings fell 6.3%. Software development is the most skewed field: in Q1 2026 about 69% of its postings were senior roles1.
Young workers are hit first. Stanford's Digital Economy Lab ("Canaries in the Coal Mine") finds that employment of 22–25-year-olds in the most AI-exposed jobs, software development among them, fell about 16% relative to less exposed jobs, while experienced workers held steady. The authors say interest rates don't explain the gap, and that the effect became significant mainly after 20242.
Overall postings have partly recovered, but from a low base. Software job ads have been growing again since 2025, yet remain well below their 2020–2022 peak.
So "demand for specialists is dropping sharply" is true for people trying to get in. "We'll need more engineers" is a bet about the next stage. Jon Hays' warning about the pipeline is exactly the point where the two meet: if companies stop hiring juniors now, there will be few mid-level and senior engineers in 5–10 years.
"AI didn't reduce the tasks, we just work more"
That matches a well-known pattern, the Jevons paradox: when something becomes cheaper to produce, people tend to want much more of it. Cheaper code means more internal tools, more automation, more prototypes, plus more code to review, secure and maintain. The work grows instead of disappearing.
There's also evidence that AI speedups are smaller than they feel. In a 2025 randomized trial by METR, experienced open-source developers took about 19% longer on tasks when using AI tools, even though they believed it made them faster3. It's one study with a specific setup, but it fits your experience: more output and more context-switching, not necessarily less effort.
People from other fields now "also need to program"
Yes. AI lowers the barrier to starting to write software, but not to making it correct, secure and maintainable. Analysts, marketers and researchers now build scripts and tools with AI, and quickly find they still need the fundamentals: data, debugging, version control, reading code they didn't write. Programming is turning from a profession into a widespread skill, while professional engineering shifts toward design, review, integration and responsibility for the result.
Bottom line
AI hasn't removed the need for engineers. It has raised the bar for entry and increased the volume of work for those already in. Whether Hays is right about future shortages depends mostly on whether companies keep training juniors now, not on AI itself.
Indeed Hiring Lab, "The Labor Market Is Tilting Toward Seniority" (July 2026): https://hiringlab.indeed.com/2026/07/23/the-labor-market-is-tilting-toward-seniority/ ↩
Stanford Digital Economy Lab, "Canaries, interest rates, and timing" (Feb 2026): https://digitaleconomy.stanford.edu/news/canaries-interest-rates-and-timinga-more-on-recent-drivers-of-employment-changes-for-young-workers ↩
METR, "Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity" (July 2025): https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/ ↩
Yes, you noticed an important thing: AI may not reduce the total volume of tasks, but reduce the demand for the input level. Then it turns out a strange picture — there seems to be more work, but it is increasingly difficult for a beginner to get the first chance, because they are immediately expected to be able to program, check the result of AI and understand the subject area.
Here it is interesting to separate two hypotheses: AI does create new demand, but this demand goes to more experienced professionals; or companies simply use AI to avoid hiring and force the rest to work more. Outwardly, both scenarios look like "more programming".
What does it look like in practice? Has the share of review, integration and requirement setting tasks increased, or has it become more of an independent development? And what happens to the beginners around you — are they taken for internships, or are they waiting for a ready-made specialist from the first day?