Tyler Cowen on Kimi K3, AI maniacs, vibe sessions, and why banning open-source AI will fail

Tyler Cowen on open-source AI, AI maniacs, and the vibes recession

Tyler Cowen's central argument is simple: banning open-source AI will fail, and the attempt will cost more than it saves. The question is what the US does instead.

Open-source AI

Cowen is direct that open-source AI — much of it Chinese — is already embedded in American enterprise supply chains. Major US companies use Chinese open-source models today and aren't advertising it. Trying to sanction or outlaw that is, in his view, a doomed exercise: when AI moves fully on-device, the leverage point of controlling server infrastructure disappears entirely. He also makes the less obvious point that open-source and proprietary American AI are complements rather than rivals. Enterprises are more willing to lock into US systems when open-source exists as a fallback, because it reduces the risk that American models get yanked from them.

His plea to the Trump administration is to abandon the campaign entirely.

“Open source AI is coming and is here whether we like it or not. The attempt to outlaw it or ban it or use sanctions against it is going to fail miserably... AI maniacs will compete in medicine, law, business consulting, banking wherever it is we let them compete. So this to me is the future... There's so many people who will say, My goodness, if we had Fable five, I would have thought the world would be totally different and it's not.”

AI maniacs

Cowen's more forward-looking argument centers on what he calls AI maniacs — young people, some as young as 13, who are teaching themselves AI, building models, and starting companies. He spoke to two that morning, one from Abu Dhabi, one from Finland. He argues the next wave of high-revenue, low-headcount companies will be built by this cohort, and that the right policy question for any country is simply whether it is creating conditions for them to rise.

He is more optimistic about Europe than he was six months ago. His read is that chatbots didn't excite European youth much, but agents — tools that actually do things — represent a genuinely new opening for them. He expects new companies from young founders across Eastern and Western Europe.

On what support looks like, Cowen is running his own grant program, going "younger and weirder" in who he backs. He notes venture capital has a valley of death problem for very early, very young founders, and that someone needs to fill it — government or otherwise.

The vibes recession

Cowen calls the current consumer sentiment gap "negative emotional contagion." Real wages, job markets, stock prices, and wealth accumulation are doing okay to fine. The disconnect from how people feel is real but not new — it feeds on itself, and history suggests it sometimes takes a genuine shock to break the cycle. He leaves open the possibility that a brief recession or market correction could paradoxically prove stabilising, by resolving the cognitive dissonance between objective conditions and subjective dread. He's not confident in that outcome.

On the labor market, youth unemployment for 18-to-24-year-olds is back to pre-LLM levels, which he takes as evidence AI has not yet disrupted that cohort structurally. The louder complaints about job scarcity are coming from educated workers who don't want, or aren't suited for, the jobs actually being created. The AI maniacs, he argues, are about to capture a disproportionate share of status and income — and the political system will struggle to process that reallocation.

His advice to people inside large organisations who aren't built to be founders: follow AI closely, learn the leading models, and find the messier roles. The leisure dividend from AI is not arriving yet. For now, the answer is to work harder.