Writing mathematics in the age of AI
Summary
Martin Hairer argues that mathematics written with AI assistance should follow clear principles of verification, transparency, attribution, and human authorship. Mathematicians should scrutinize AI-generated proofs much more carefully because AI can produce convincing but incorrect arguments, and any meaningful AI contribution to mathematical content should be explicitly disclosed, while routine uses such as proofreading need not be reported.
AI-generated ideas should also be checked against the existing literature to ensure proper attribution. Most importantly, Hairer advises mathematicians not to copy AI output directly into papers: they should understand, simplify, and rewrite the material in their own voice, preserving the big picture, precision, consistency, and intellectual contribution of the human author.
ARTICLE / ARCHIVE
Summary
Martin Hairer argues that mathematics written with AI assistance should follow clear principles of verification, transparency, attribution, and human authorship. Mathematicians should scrutinize AI-generated proofs much more carefully because AI can produce convincing but incorrect arguments, and any meaningful AI contribution to mathematical content should be explicitly disclosed, while routine uses such as proofreading need not be reported.
AI-generated ideas should also be checked against the existing literature to ensure proper attribution. Most importantly, Hairer advises mathematicians not to copy AI output directly into papers: they should understand, simplify, and rewrite the material in their own voice, preserving the big picture, precision, consistency, and intellectual contribution of the human author.
ARTICLE / ARCHIVE
┌────────────────────────────────┐
│ KONSTANTINOS MICHAILIDIS │
└────────────────────────────────┘
│ KONSTANTINOS MICHAILIDIS │
└────────────────────────────────┘