There's an image that keeps recurring in history, one I recognize from other contexts: someone announces a fire, sells the extinguisher, and it turns out they also sold the matches. Sam Altman and Elon Musk have spent years warning that artificial intelligence will wipe out jobs on a massive scale. At the same time, they present universal basic income as the natural response. They wrap it in language of humanitarian urgency and responsible preparation. Meanwhile their companies rack up record valuations while executing layoffs simultaneously.

That doesn't look like tech philanthropy. It's a pattern we've seen before, just with different tools this time.

This year's records show around one hundred forty-two thousand layoffs in tech. In the same period, Meta, Amazon, and Oracle are pouring roughly seven hundred billion dollars into AI infrastructure. This isn't growth that later demands adjustment. It's happening at the same time, with the same firms authorizing both decisions. That simultaneity matters more than either number on its own.

Those hit hardest aren't senior profiles with networks or savings to fall back on. Developers under twenty-six have lost about twenty percent of their job base since 2024. They entered the market on the promise that learning to code was insurance against the future. Now they're facing the fact that the same lever can replace them. This matters because it reveals who takes the first hit in any transformation: those with the least room to absorb it.

Job anxiety confirms this. The share of workers expressing real concern about AI went from twenty-eight percent in 2024 to forty percent in 2026. At Davos, the IMF admitted something unusual for that forum: most countries and companies aren't prepared for this transition. When that institution says so, it's worth listening.

This is where UBI enters as a proposal — and where the data complicates the optimistic narrative.

A meta-analysis from the American Enterprise Institute published this year reviewed the available basic income pilots. Small trials, with fewer than five hundred participants, show a positive employment effect of 0.8%. Trials with more than five hundred participants show a negative effect of -3.2%. That doesn't mean the mechanism never works. It shows that behavior changes at scale. And political promises are always pitched at scale.

The study most often cited as proof that UBI works — funded by Altman through OpenResearch — finished its data collection phase in 2023. Results have been rolling out since 2024. What was presented for years as the largest ongoing experiment is now a closed record. It can be analyzed with some distance now. And it doesn't deliver the conclusions its promoters would need to justify the mass rollout they're suggesting.

This connects to something I explored before about the different historical versions of basic income. There are at least five incompatible approaches that share the same instrument but pursue opposite ends. Thomas Paine sought to redistribute land rent. Milton Friedman's libertarian version aimed to dismantle the welfare state and replace it with a check. The version Altman and Musk propose doesn't transfer productive power: it hands out liquidity while consolidating control over infrastructure. It's not the same thing.

The crypto vehicle tied to the UBI-AI ecosystem doesn't help this story either.

Worldcoin, Altman's project that promised to distribute tokens in exchange for scanning irises, has lost nearly ninety-eight percent of its value from its peak. Nine out of ten tokens are concentrated in roughly a hundred wallets. This July brings a massive unlock that will let early investors and the founding team sell off their positions. The structure looks more like an extraction mechanism dressed in redistributive language than a tool for spreading global wealth.

I've seen in organizations how the language of transformation gets used to justify restructurings that benefit the same people as always. The vocabulary changes. Who wins and who loses, less so. That brings to mind the FTX case — not for the legal details but for the rhetorical architecture: effective altruism language, redistribution promises, and early backers that included Sam Bankman-Fried and Three Arrows Capital. When the wrapping of doing good surrounds extreme concentration, the question goes beyond personal intentions. It's about what incentives the design itself produces.

UBI as an idea isn't inherently bad. Some historical versions had solid redistributive logic. The problem lies in who controls it, how it's funded, and what power structures it leaves intact or reinforces.

When the same ecosystem carries out the layoffs, builds the infrastructure that justifies them, funds the experiments that measure them, and proposes the compensation that normalizes them, an uncomfortable question emerges. Are we looking at a response to the problem, or at the problem itself with better marketing?