When Dan Abramov—the creator of Redux—recently sat down with an AI and \”vibed\” a mathematical proof of Conway’s conjecture, purists panicked. Math is supposed to be the last bastion of absolute, unyielding truth. \”Vibing\” is the opposite. It’s fuzzy, intuitive, and deeply informal.
The phrase \”vibed a proof\” is deliberately paradoxical. Proof implies rigor; vibe implies hanging out. But instead of ruining mathematics, this gap is exactly what makes it productive.
When you mix the absolute certainty of math with the fuzzy intuition of a \”vibe,\” you don’t get chaos. You get an assembly line of plausible nonsense.
The Hacker News comment section was a mix of awe and existential dread. One user joked, \”Someone please vibe-prove that ZFC is inconsistent.\” Another compared AI to the infinite monkey theorem, arguing that mathematicians now have more work than ever to unravel the AI’s output.
But here is the twist: everyone is asking the wrong question.
The debate is stuck on \”Is the AI correct?\” We obsess over hallucinations, fact-checking, and whether the model actually knows the math. Stop it. The real disruption isn’t downstream in verification; it’s upstream in generation.
AI isn’t replacing the thinker. It’s drowning the thinker in infinite variations, forcing them to become an editor of plausible garbage.
You’ve probably noticed this in your own work. You ask ChatGPT for a strategy, and it gives you five brilliant-sounding ideas. But are they actually brilliant, or just grammatically flawless? You’re no longer generating from scratch. You’re sifting. Your value isn’t in producing the answer anymore. It’s in judging which answers deserve to exist.
Your job is no longer to find the needle in the haystack. Your job is to realize the AI just handed you a haystack made entirely of needles, and you have to figure out which one actually pierces.
Fields Medalist Timothy Gowers has pointed out that this fundamentally changes the nature of mathematical work. The net output of knowledge increases, but only if the human acts as a ruthless editor. If you accept the vibe without the rigor, you fail. If you reject the vibe because it’s informal, you also fail.
The era of the solitary genius generating truth from the void is ending. We are entering the era of the curator. AI will generate a million wrong paths. Your job is to walk them all, quickly, and decide which one to pave.
The future of knowledge work isn’t about knowing the answer. It’s about having the taste to know which answers are worth keeping.
FAQ
Q: What is the key takeaway?
A: See the article.