Both Sides of the AI Jobs Debate Are Solving for the Wrong Person

There's a version of the AI-and-jobs conversation that happens in conference rooms and on podcasts, and there's a version that happens everywhere else. The first version is the only one getting written about.

Two pieces landed recently on opposite ends of this debate, and together they accidentally reveal something neither one intended to say.

The Respectable Disagreement

On one side, TechRadar ran a piece on the gap between how fast AI infrastructure is being deployed and how slowly workforce expertise is following it. The argument: we have a standardization problem, an education problem, and a coordination problem — and the system as a whole, as one voice in the piece put it, hasn't developed a consistent, shared understanding at the same rate as the infrastructure itself. The prescription is institutional: build frameworks, train people, align incentives.

On the other side, Platformer sat down with Boris Cherny, the creator of Claude Code at Anthropic. His position is more direct: major job loss due to automation really is coming. He also says job creation is coming too, which is the part that tends to get quoted in the optimistic summaries. Both things can be true. He's not wrong to say so.

These are serious people making serious arguments. The problem isn't that either of them is lying. The problem is who's sitting at the center of each argument.

The Person Nobody Named

The TechRadar piece worries about practitioners who can't keep up with deployment speed — which is a real concern, but it's a concern about engineers, IT teams, and organizational decision-makers. The Platformer piece worries about software engineers specifically, a profession where the people most likely to be displaced are also the people most likely to have the education, the network, and the savings to navigate what comes next.

Neither piece names the person in a data-entry role at an insurance company. Neither names the paralegal doing document review. Neither names the customer service worker whose entire job function is being quietly automated away by a chatbot that costs a fraction of their salary and never asks for benefits.

The reskilling conversation — such as it is — assumes a person with enough runway to be reskilled. It assumes institutional support that, in most industries, doesn't exist. It assumes the gap between displacement and re-entry is a few months of discomfort rather than a permanent reclassification of what kind of work you're eligible for.

Standardization would help the organizations deploying AI make better decisions. Education would help the workforce adjacent to AI stay current. Both of those things are worth doing. Neither of them is a plan for the person whose job simply stops existing before any of the frameworks get built.

The infrastructure, as TechRadar accurately notes, is moving faster than the understanding. What nobody's saying out loud is that the understanding is also moving faster than the safety net. And the safety net is moving backward.

I've watched enough technology cycles to know how this part of the story goes: the macro numbers eventually look fine, job creation offsets job loss in the aggregate, and the economists write the papers about net neutral outcomes. The aggregate has never had to find rent.

Two smart pieces. One missing character.