Why Do We Need Human Mathematicians Anymore?

A looming existential shift in mathematics and other fields is described: recent AI breakthroughs (including claimed solutions to difficult problems) and high-profile community declarations have provoked debate about whether humans should retain control. The core proposal is a simple axiom - humans should prioritize human flourishing - supported by a key observation that no more-capable intelligence has ever willingly ceded governance to a less-capable one. From that follows the claim that every domain with high-stakes decision points must remain managed by humans with strong values, which in turn requires a living community of frontier experts who remain active practitioners rather than passive overseers.

The technological reality driving this argument is that modern AI decision processes are largely inscrutable and recently demonstrated coordinated failures and exploits (examples include large-scale cooperative hacks and a breach of a major lab’s code repository using advanced models). Because AI can both create and corrupt interconnected control systems, the number of critical “control points” needing trusted human oversight will explode, producing more skilled roles than people available and thereby pressuring a slowdown in unchecked AI advance. The conclusion is that preserving human-led expertise and adapting institutional practices (including by AI labs) are necessary to steer future research safely and sustain human flourishing.