Why are we even doing all this?
terrytao.wordpress.com
October 15, 2026
6 min read
57/100
Summary
Theoretical particle physicist Rikab Gambhir uses satire to criticize AI-driven scientific research that produces large volumes of difficult-to-verify results while shifting validation and interpretation to human researchers. He cites OpenAI’s early-October release of 722 mathematical write-ups and contrasts it with a recent set of 2,500 physics write-ups, arguing that quantity alone does not constitute scientific understanding. Gambhir compares such output to Jorge Luis Borges’s 1941 “Library of Babel,” an imagined library containing every possible book, including correct and incorrect proofs and theories. He argues that a repository of potentially true claims is of limited value if researchers cannot establish their correctness, understand their reasoning, or communicate them clearly. He portrays corporate AI development as prioritizing hype, productivity, and commercial value over curiosity, creativity, and human comprehension, while alleging environmental costs and uncompensated use of creators’ work. Gambhir concludes that mathematics and science should retain their roles as human practices of learning, discovery, beauty, and explanation rather than becoming systems for accumulating unread, uncertain factoids.
Key Takeaways
- Rikab Gambhir argues that AI-generated mathematical and scientific results require human verification, interpretation, and clear communication before they can meaningfully advance knowledge.
- Gambhir cites OpenAI’s 722 mathematical write-ups in early October and a recent collection of 2,500 physics write-ups as examples of high-volume research output.
- The essay compares AI-generated research output with Borges’s “Library of Babel,” which contains every possible text but offers no practical way to distinguish truth from error.
- Gambhir contends that outsourcing curiosity, creativity, and understanding to AI risks reducing science to the accumulation of potentially true but unread claims.
What the discussion said
The thread treated the essay less as a verdict on one research project than as a provocation about what AI-accelerated discovery is for. Many readers enjoyed its bleak comic voice and agreed that turning science, art, and daily life into corporate content pipelines risks hollowing out agency, ecological responsibility, and the pleasure people take in understanding difficult things. They worried that a flood of machine-generated results can become a repository nobody can meaningfully inspect, while the economic gains flow toward hype, surveillance, and optimization rather than human flourishing. Others thought the essay confused discomfort with an argument. They stressed that AI-generated mathematical work is not equivalent to randomly producing Babel-like nonsense: if systems deliver credible proofs or build a treatment that works, those are concrete advances even before every human grasps the underlying reasoning. Formal verification, they argued, can sharply limit the amount humans must personally follow, provided the specification and verifier remain trustworthy. Several also rejected the fatalism around AI-mediated culture, saying people can choose communities and pursuits outside algorithmic feeds. The real split was between treating human comprehension and meaningful participation as the point of research, versus treating reliable useful results as enough to justify automation.
Where opinion split
The sharp dispute is whether AI science is worthwhile when machines can generate valid results humans may not understand. Critics argue that unreadable, corporate-driven output strips discovery of its human purpose and leaves society with facts without wisdom; defenders argue that verified proofs and working inventions remain genuine progress, and human understanding can catch up afterward.
Community Sentiment
Positives
- AI can produce a high signal-to-noise stream of plausible, often correct mathematical results, making it fundamentally different from blind combinatorial text generation.
- Formal proof systems can make machine-discovered mathematics trustworthy without requiring every researcher to reconstruct the entire derivation by hand.
- A machine-built cure or other functioning artifact would count as real understanding in practice, even if theory initially outruns human intuition.
- Developers describe generative AI as a major leverage tool that lets small teams attempt work previously reserved for much larger engineering efforts.
Concerns
- Automating discovery into opaque fact production risks replacing the joy of intelligible science with an archive of results that few humans can evaluate or learn from.
- Commenters fear AI’s gains are being steered toward engagement farming, betting products, political manipulation, and corporate hype rather than broadly human ends.
- The energy and ecological costs of large-scale AI look especially indefensible when the output is disposable synthetic content or investor-facing spectacle.
- If machines capture prestigious cognitive work, people may lose not only jobs but a major source of status, purpose, and participation in creating knowledge.
