Hi everyone 👋
As a software development student, I read some version of "is coding dead?" almost every week. Big company, big layoff, and the same explanation: we're using AI now, so we need fewer people.
This week I watched an episode of BBC's New Normal with Katty Kay called What If We're Wrong About AI Layoffs? Katty Kay talks with labor economist Kathryn Anne Edwards, and it changed how I read those headlines. So I want to write down what stuck with me.
AI-washing
The episode starts with a term some economists use: AI-washing. A company says AI is the reason it let people go, but it would have done the layoffs anyway.
Why would anyone do that? Edwards's answer is simple. The stock market does not treat AI neutrally. If you say you are cutting jobs because you over-hired, that does nothing for your shareholders. If you say you are cutting jobs because you pivoted to AI, your company can suddenly look more valuable.
She also describes a kind of peer pressure during downturns. Three other companies cut 5% and said it was about AI. Why haven't you? Are you doing AI?
And she points out this is not new. Companies have always needed a fall guy. Her example: a store replaces clerks with iPads and blames the minimum wage, when it might have done it anyway.
We may never know
So how many of these layoffs are really about AI? Her honest answer: we don't know, and we may never know.
She makes the point with history. Imagine you had perfect US employment data going back to 1955, and someone asked you how many jobs computers destroyed. Some cases are clear. "Computer" used to be a job title, mostly held by women, and it is gone. We don't have typists or phone operators anymore. But beyond those obvious cases, putting a total number on it is close to impossible. The same goes for the internet, or Microsoft Office.
We know these technologies changed work. We just can't cleanly separate their effect from everything else that was happening at the same time.
Why this felt familiar
This part is mine, not hers.
If you have ever tried to find what caused a performance drop after a big release, you know this problem. Forty pull requests went out, one dependency got upgraded, and someone changed a config on the server. Then the latency goes up.
The first explanation in the meeting is usually the most convenient one. "It's probably the new library." Everyone nods, because it sounds plausible and nobody has to look at their own code. Sometimes it is true. Often it is not.
In software we at least have tools for this. We can roll back, compare logs, or run git bisect until we find the exact commit. The economy has no git bisect. Nobody can roll back a year and replay it without AI. So "it was AI" can stay a convenient explanation forever, because it is very hard to prove wrong.
The part about us
This is where the episode got personal for me. Edwards mentions a chart of job postings on Indeed that she had looked at that morning. According to her, software development postings were doing much worse than overall job postings at the start of 2024. Since then they have recovered, and over roughly the past year they have done better than the job market overall.
I haven't seen that chart myself, so I'm passing this on as her reading of it, not as my own number. But the point she makes with it is hard to ignore: the data and the story don't match.
She says that as an economist she cares about numbers, but she knows narrative is everything. Every May, she says, major outlets publish a piece asking whether computer science graduates still have a career waiting for them. That changes how students think, and it changes the advice their parents and mentors give them, even when the data itself is much less clear.
Her advice, and what I take from it
Katty Kay asks what she would tell an 18-year-old choosing a major. Edwards says she doesn't really have advice, but she wants to clear up a myth. The idea that people used to get one good job and keep it forever isn't really true. For almost everyone, the first 15 years of a career involve a lot of movement.
Her suggestion: pursue the thing you would want to do for the longest. The job market when you graduate matters, but it decides your first job, not your last one. She says she doesn't know anyone who could have predicted their career at 20.
For me, this changes the question. I have spent a lot of time asking which skills are "AI-proof." Maybe that is the wrong question, especially if some of those layoffs were never really about AI in the first place. A better one might be: what kind of work do I want to keep getting better at for the next fifteen years?
What I'm not sure about
The argument cuts both ways. If we can't prove a layoff was caused by AI, we also can't prove it wasn't. "It might be AI-washing" is not the same as "AI isn't changing anything." Edwards herself says AI changes the nature of work. Her point is that we can't measure how much.
This is also one interview with one economist, and a short one. The job-postings comparison comes from a chart she described in conversation, not from data I have checked. So I'm reading this as a good reason to be skeptical of headlines, not as proof that everything is fine.
À bientôt 🙂
Source
What If We're Wrong About AI Layoffs?, New Normal with Katty Kay, BBC: interview with labor economist Kathryn Anne Edwards.
