What should we tell our students?
Rapid advances in AI math capabilities are forcing students and academics to rethink whether pursuing a mathematics PhD is still a viable or ethical career choice. Commenters weigh optimism—that AI will be a powerful tool for human mathematicians and that teaching and problem selection will remain human domains—against fears of a “jobs apocalypse,” collapsing academic prospects, and math becoming a niche hobby or intellectual sport. Broader themes include the difficulty of planning a 40‑year career under such uncertainty, parallels to turmoil in software engineering, and disagreement over whether students should lean into AI tools or actively avoid helping systems that may displace them.
Impact of AI on Mathematics as a Career
- Many argue current and near-future AI is already competitive with most mathematicians, making math research a risky career choice.
- Several commenters recommend: study math for personal satisfaction, but secure income via another field or a math-adjacent area (CS, finance, formal methods, teaching).
- Others insist there will still be demand for human mathematicians, especially for choosing problems, interpreting results, and teaching, but concede academia was already oversupplied with PhDs and short on stable positions.
- Concern that optimistic career advice from tenured faculty underplays structural risk and focuses too much on “hopium”.
Nature of Mathematical Work Under AI
- Reports that AI systems can rapidly solve problems that were planned as multi‑year projects, sometimes with no intention by humans to publish the results.
- Worry that research becomes “homework with answers in the back” — solutions exist in the model and are retrieved rather than discovered, potentially slowing genuine knowledge accumulation and making it harder for younger researchers to find meaningful open problems.
- Debate over whether current AI is just recombining existing ideas within a “convex hull” vs. whether it will soon produce truly novel, “alien” concepts.
Parallels to Software Engineering and Other Fields
- Strong analogy drawn to software: “math is more than writing proofs” just as “software engineering is more than typing code”.
- Split views on AI coding tools: some say LLMs remain unreliable and hype-driven; others say they are already central to daily work and partly explain collapsing entry-level CS hiring.
- Broader anxiety that many white-collar jobs (CPAs, developers, mathematicians) are vulnerable as AI agents become capable of full workflows.
Education and Use of AI
- Big divide on whether students should use LLMs:
- Pro: powerful learning aid, like calculators or backfiles, essential to stay relevant.
- Con: encourages “answer-key learning”, erodes fundamentals, and demands unrealistic self-discipline; calls for more proctored exams.
- Some advocate boycotting frontier models (especially in pure math) to protect education and bargaining power; critics call this futile and self-sabotaging.
Wider Economic and Social Concerns
- Divergent predictions: from “no careers in 40 years” to “jobs will adapt as always”.
- Debates over UBI, ownership of AI/robots, and whether a world of automated labor is utopia, dystopia, or an “economic dark age”.
- Repeated theme: technical advice is inadequate without political and institutional change.