Research and mentoring

Leslie Ann Goldberg — “My AI Policy” See the “My AI Policy” section for expectations to discuss before starting a collaboration, including agreement before submitting shared problems or proof steps to AI and disclosure of any agreed AI use. → Frank Calegari proposes changes to mathematical practice in the AI era Frank Calegari proposes changes to research evaluation, hiring, and graduate training that reward understanding and ambitious questions. → Harvard summit publishes recommendations for math research in the age of AI The Harvard summit recommends supporting early-career researchers, recognizing exposition and shared tools, and building expert-guided formal libraries. → Early-career researchers share their perspectives on AI in mathematics A survey by Anna Brandenberger, Hanna Mularczyk, and Dora Woodruff brings together graduate students’ and postdocs’ experiences and concerns about AI. → AHM publishes recommendations for doctoral advising in mathematics AHM recommends written AI-use agreements, clear expectations, and supportive advising, while advocating predominantly AI-free doctoral work and recognizing students’ progress. → U-M releases guiding principles for the use of artificial intelligence U-M’s six guiding principles emphasize human judgment, integrity, transparency, and privacy across research and education; they are guidance rather than a new policy. → Kentucky mathematics proposes written AI agreements for doctoral advising Kentucky’s policy proposal calls for written agreements between doctoral students and their committees on permitted AI use, reviewed annually. → Francesco Maggi considers a golden age or a dark age for mathematics Francesco Maggi argues for preserving human understanding, independent scrutiny, and opportunities for young researchers as AI-generated proofs become more common. → SAIR Open Math OS launches tools for collaborative mathematics SAIR Open Math OS offers open AI models for mathematical discussion and Lean formalization, including shared team prompts for collaboration. → Thomas Hales examines Lean reliability and AI-generated proofs Thomas Hales explains reliability issues in Lean and AI-assisted proofs, including the need to audit theorem statements and verify results independently. → Maximilian Kaipel proposes transparent AI disclosures in mathematics research Maximilian Kaipel proposes specific AI disclosures, records of relevant interactions, and human accountability, with a task taxonomy and LaTeX package. → Jeremy Avigad considers the future of mathematics with AI Jeremy Avigad advocates broader conceptual thinking, harder questions, and support for early-career researchers as AI changes mathematics. → David Glickenstein proposes graduate training centered on mathematical judgment David Glickenstein argues that graduate training should emphasize judgment about mathematical claims, proofs, definitions, and research directions. → Emily Riehl explains the value of mathematical theory building Emily Riehl explains why theory building, clarifying abstractions, and communicating ideas remain central to mathematical research. →