I know why the mathematicians weep
Rapid advances in AI that can autonomously prove long‑standing mathematical conjectures are prompting both awe and grief among mathematicians and technologists. Many fear not just for careers and funding but for the loss of human discovery, understanding, and the “journey” in fields where AI can now brute‑force results beyond any individual’s capacity. Others argue this is another technological shift that will raise new frontiers, insisting the real problems lie in economic structures, power concentration, and how society chooses to value human effort once machines outperform us at more intellectual work.
Pace and Nature of Recent AI Progress
- Many say they intellectually expected strong AI for decades but are still shocked by how fast capabilities, especially in math, arrived and accelerated.
- Some frame this as “singularity-style” constant acceleration; others argue physical and data limits will force a plateau and see current extrapolations as speculative.
Mathematicians’ Grief and Reactions from Other Fields
- Several commenters empathize with mathematicians mourning the loss of discovery, identity, and meaning, comparing it to losing a vocation or loved one.
- Others are unsympathetic, arguing everyone was warned for decades, or pointing out that artists, writers, clerical workers, and programmers were hit earlier.
- Some see mathematicians’ distress as exposing how tightly self-worth has been tied to productivity, prestige, and status.
What AI Is Actually Doing in Mathematics
- Many stress that current systems are “superhuman at deduction” over known literature, stitching together existing techniques at a scale no human can match.
- Debate over originality: some think models are confined to a “convex hull of ideas” from training data; others believe they already have generic reasoning that can scale to effectively unbounded novelty.
- Concerns that the recent proof “dump” is huge, messy, often unreadable, and only trusted because of machine checking, creating “knowledge without understanding.”
- Worries about career impacts: fewer grants, less need for human theorem-proving, and worsening precarity for non‑tenured mathematicians.
LLMs, Writing, and Style
- Several note that chat models have a distinct, often corporate/PR tone and struggle with long-form, human-feeling prose, despite excelling at formal math.
- Explanations offered: training data quality, optimization for safety and generic chat, lack of economic incentive for top-tier literary style.
Jobs, Capitalism, and Meaning
- Strong anxiety about broader job losses: software engineers, tech writers, translators, and others already feel displacement or hiring freezes.
- Some blame AI only marginally, citing higher interest rates and macroeconomics; others think companies are delaying hiring in anticipation of further automation.
- UBI is floated as a solution, but many doubt political feasibility; fear of a new permanent underclass and erosion of human leverage is common.
Value of Human Understanding and Social Knowledge
- Multiple comments emphasize that math and science are also about intuition, explanation, and shared human understanding, not just correct answers.
- Concepts like “deontic scorekeeping” are invoked: knowledge arises from humans taking and evaluating commitments over time, something current models cannot do.
- Some optimistic voices hope mathematicians will pivot to interpreting alien AI proofs, creating new abstractions, and working with tools rather than being replaced.