# Why AI is getting good at math: depth-first search and verifiers

> Original title: AI Is Getting Good at Math. Here’s Why.

- Source: Latent Space
- Published: 2026-10-23T15:19:48.000Z
- Bad Influence: https://badinfluence.news/items/m38cduu0ps9hnqbnostpvrzig
- Original: https://latent.space/p/ai-is-getting-good-at-math/

## Summary

Ankit Maloo argues that AI models solve certain open math problems because they pursue many candidate proofs to completion and use verifiers to prune wrong ones. He contrasts this with human researchers, who prune early because each direction can take months to test. He predicts models will do best on problems with few conceptual leaps and manageable branching, and that human intuition still matters where the search tree is deep and wide.

## Why it matters

The essay explains why models solve some open math problems by comparing their depth-first search with human breadth-first pruning, a framework useful for judging which problems AI may crack next.