Aaronson says AI turns mathematicians into curators

Hot take: this reads like a checkpoint on the road to AI-native research, not just another “AI is getting better” essay.

• –The core signal is mathematical, not rhetorical: the post treats AI as a credible contributor to open-problem solving, with humans moving into review and direction roles.
• –The argument is reinforced by multiple examples, not a single cherry-picked result: one-shot theorem discovery, formalized proofs, and a separate CS theory breakthrough.
• –The fiction and Vatican sections widen the lens, suggesting the disruption is cultural and institutional, not just technical.
• –For readers tracking frontier AI, the important takeaway is the shift in perceived agency: models are being discussed as research actors, not assistants.