No, AI Has Not Wiped Out the Programmers

The claim is everywhere by now. Open any of the more breathless corners of the tech press and you will find some version of it: AI has eaten the programmers, the entry-level job is gone, the career is over.

The aggregate evidence does not support the strong version of that claim. A weaker version, about who is being hurt and where, holds up better. Both versions get conflated in the headlines, which is how the myth keeps its momentum.

What the aggregate numbers say

Software-engineer job postings on Indeed are up 11 percent year on year, faster than the rate of growth in postings overall. The US Bureau of Labor Statistics projects employment for software developers to grow 15 percent by 2034, considerably faster than the average for all occupations. CNN summarised the picture in April with a headline that has not aged badly: the demise of software-engineering jobs has been greatly exaggerated.

That does not mean nothing has changed. The Pragmatic Engineer's mid-year survey found that 65 percent of developers expect the job to be redefined this year, with less routine coding and more time spent overseeing AI-generated work, integrating systems, and making architectural calls. That is a real shift in the shape of the job. It is not the same as the job ceasing to exist.

Where the myth has a point

The strong version fails at the aggregate, but the entry-level data is genuinely concerning. A Stanford analysis found that employment among software developers aged 22 to 25 fell by close to 20 percent between 2022 and 2025, a period in which AI coding tools became standard inside large engineering organisations. Entry-level postings dropped in the US in 2024 and 2025 even as senior postings held up. We covered a related Stanford CS placement claim earlier this month, and the same pattern shows up here.

The mechanism is intuitive. Junior developers do exactly the kind of well-specified, contained work that current models do credibly: writing a function from a clear spec, fixing a small bug, scaffolding a new file. When a senior engineer can produce that work themselves with an AI assistant, the case for hiring a person to do it weakens. The new floor for being hireable as a developer is higher than it was three years ago.

Why the conflation matters

Treating these two stories as the same has practical costs. It pushes computer-science students away from the field at exactly the moment when the upper tiers of the profession are still hiring. It encourages employers to over-trust headlines and under-invest in junior training, then complain a few years later that they cannot find anyone with the senior judgement they used to grow in-house. And it gives cover to layoffs that have more to do with cost-cutting than capability, dressed up as AI substitution. We discussed the framing problem in the context of recent Meta and Microsoft cuts.

The honest summary is duller than the viral version. Aggregate demand for software developers is rising. Entry-level demand is shrinking. The job content is shifting toward design, integration, and oversight. People who learn to use AI tools well are doing better than people who do not. None of that is a job-apocalypse story. It is, however, a serious story about how the profession's training ladder is being remade in real time, and it deserves better coverage than the headlines have given it.