EP 07 – The Talent Gap: Who Becomes a 100x Engineer When AI Does Everything?

Andrej Karpathy said it: regular engineers became 10x engineers, and 10x engineers became 100x engineers. But Harry raises the harder follow-up question — what happens to the people who were supposed to become 10x someday, if the training ground disappears?

This episode is about the talent question at the center of the AI transition, and it’s more interesting than the standard “AI is taking jobs” conversation.

Harry opens with the Groundhog Day frame: the movie isn’t about a time loop, it’s about recursive learning. Bill Murray’s character spent what researchers calculate was 20-30 years trapped in the loop — and came out speaking French, playing classical piano, and ice sculpting at expert level. The labs are doing the same thing: their code is reviewing and writing code. The recursive loop is compressing decades of iteration into weeks.

That’s great for the labs. What does it do to talent development?

Anthony and Harry work through it: the exceptional people — the Mozarts, not the Salieris — will get found faster now. Jonas Vingegaard was discovered because Strava made his watts-per-kilogram data public and a team scout noticed. Formula One finds Max Verstappen through video game leaderboards. AI creates more surfaces for discovering exceptional people who would otherwise stay invisible.

But there’s a real cost: if you don’t need 45 average engineers around 5 exceptional ones, those 45 people don’t get the reps that sometimes produce the next exceptional one. The training pipeline thins. And the companies that move fast may be inadvertently cutting off the mechanism by which great people get made.

What to do about it: identify and give exceptional people room to run. Don’t expect 1x talent to become 10x — the ceiling doesn’t change. But do create the conditions where exceptional people who haven’t been found yet can find their way to you.

Read the CO/AI newsletter at getcoai.com
Follow Anthony: @djabatt | Follow Harry: @hdemott