The weird great man theory of AI
Why do people still matter when people aren't supposed to matter?
Yesterday, Google was rocked by some AI strategy news: longtime head Demis Hassabis will be changing his title (though perhaps not many of his responsibilities). And, as notably, longtime super-engineer Jeff Dean and some of his colleagues are leaving to run their own Google-adjacent AI project (which Google is partly funding). Google stock apparently sank on the news.
Why did this happen? Looking at some possible theories is a good indicator of where AI investment is right now.
Google is cooked and people are jumping ship
The most popular theory is that the AI exodus is an external indicator of the company’s own feelings about its AI prospects. As Google’s delayed release of a new flagship frontier AI model (the kind that can wow in demos), it’s lost the bleeding edge reputation that Anthropic and Claude have. Its core AI team changing roles is, supposedly, an indicator that Google is giving up. The leaders that are still around are forced to issue increasingly defensive sounding tweets.
Logan Kilpatrick@OfficialLoganKI remain steadfast bullish on Gemini :)2:27 AM · Aug 6, 2026 · 561K Views609 Replies · 108 Reposts · 3.9K LikesOf course, none of this would matter if investors didn’t think that frontier AI capabilities matter for a company’s future. If you didn’t think benchmark performance mattered (tests that measure AI capabilities on a number of tasks), it wouldn’t disturb you that Google is falling short, since they have massive distribution and product advantages. They even have a compute advantage! So investors must think this stuff matters, either for revenue or some possible race to AGI. Which leads us to another theory…
A few key people can change the AI narrative
The other possible theory around Dean-gate is the theory that a few key people can change an entire company’s trajectory in the AI race. This isn’t only the thinking of investors — Meta spent billions on a glorified Acquihire of Al head Alexandr Wang. Former Open AI researcher (and AI demigod) Ilya Sutskever raised billions, largely on his name and promise. I could go on.
The same way that CEO pay has exploded in the past few decades, big researchers have found big paydays due to the apparent theory that their abilities can disproportionately scale across AI products. When Google loses legendary engineers and kicks Nobel-prize-winning leaders upstairs, it’s showing complacency in these talent wars.
If these arguments don’t quite make sense to you, you aren’t alone: I don’t exactly get them either. If AI is going to be entirely dependent on Miracle Year style breakthroughs, is it at all investable? Can anyone know which gnomic genius will breakthrough?
More importantly: if there are incremental steps along the way to artificial general intelligence, shouldn’t people matter less, rather than more than ever? After all, the AI boom is partly premised on the idea that scalable computer intelligence can eventually improve itself and displace expensive software engineering resources.
To trade on the recent Google news, you have to believe that a few key people are more important than ever, at least in the near term. This requires embracing the extremist thinking that AI could dramatically transform the economy within a few years, and, at the same time, believing that individuals can still create billions of dollars worth of value on their own because of their short term impacts. There’s a contradiction there that only time will settle.