“Math 2.0” will need to value mathematical progress more holistically
Points and comments are a snapshot, not live.
AI solving math problems autonomously hurts field growth; Math 2.0 must value exposition and community.
Terence Tao contrasts traditional "Math 1.0" with the emerging "Math 2.0" era. In Math 1.0, solving a conjecture sparked follow-up activity: talks, workshops, collaborations, and textbook integration that enriched the field. AI tools, when used responsibly, could support this digestion, but current practice sees problems solved autonomously by prompters with no interest in the broader field, leading to fewer community activities and withheld open directions. Tao argues that raw problem solving is being harvested unsustainably, "contaminating" fields. He calls for de-centering problem solving, elevating exposition, community building, and opening new directions, and for re-evaluating education, publication, and career criteria to reflect this new era.
What commenters are saying
Commenters largely agree with Tao but debate feasibility and AI's role. One camp argues AI will eventually explain proofs well, eliminating the gap; another insists human understanding remains essential. Some worry about junior researchers losing 'baby steps' to pursue. Others propose alternative rewards: valuing re-proofs, factoring proofs, or practical applications. A recurring theme: academia's hiring and rewards won't shift easily, and AI's impact may be inevitable, threatening mathematical community. A few note parallels to other fields, with math perhaps uniquely able to resist.