AI has access to a vastly larger working memory than the human brain
Points and comments are a snapshot, not live.
AI's mathematical edge may come from vastly larger working memory, not superior reasoning.
The author argues that AI's success in mathematics stems from an enormous symbolic workspace (context window) that overcomes human working memory limits. Human working memory can hold only a few items; AI preserves hundreds of constraints, equations, and intermediate steps explicitly. Studies show working memory predicts math performance beyond IQ. AI's advantage is largest on tasks with many interacting constraints, long calculations, and extensive case analysis. The author cautions that AI's performance reflects augmented memory, not necessarily deeper insight, and predicts humans with equally powerful external tools could narrow the gap.
What commenters are saying
Commenters split on implications. Some argue super-intelligence from expanded memory means AI will produce proofs beyond human comprehension, forcing trust in systems. Others counter that humans can decompose problems, use abstractions, and study proofs over time; shared understanding remains the goal of mathematics. A skeptic notes that incomprehensible proofs lack commercial value, while another points out that context windows scale quadratically, limiting LLM working memory. Several cite examples from computer-assisted proofs (e.g., four-color theorem) where acceptance required trust in code.