Stanford Scholar Asks: Why Do Human Brains Learn Math More Efficiently Than AI?
SuryaGanguli · x · 2026-07-28
Stanford professor Surya Ganguli shared a guide on how to read mathematics, using it to reflect on the massive differences in learning mechanisms between human brains and machines.
He noted that humans often spend hours per page when learning math, absorbing knowledge through multiple readings, asking counterfactual questions, discussing with others, and writing out intermediate steps. In contrast, current LLMs are primarily post-trained via SFT (Supervised Fine-Tuning) and RL (Reinforcement Learning).
Fascinatingly, this slow "hours per page" approach allows human brains to reach the frontiers of mathematical capability with orders of magnitude less sample complexity than machines. This observation highlights the significant advantage of human cognition regarding data efficiency compared to current AI models.
More from AGI Musings
- AI could make custom software far more common and change software economics — jiayq · 2026-07-29
- A critic says cloud infra teams are doing more for practical AI cybersecurity than the safety community — basedjensen · 2026-07-29
- The AI Governance Dilemma: Why Human Steering Can't Outpace System Evolution — anderssandberg · 2026-07-28
- AI debate should shift from where intelligence comes from to how to use it — yunta_tsai · 2026-07-28
- With over $1 trillion spent on AI, the public vision still looks vague — ben_j_todd · 2026-07-28
- Young adults use chatbots to rehearse social interactions as EU AI Act enters its main phase on Aug. 2, 2026 — emmanuelvivier · 2026-07-28