Terence Tao: AI Excels at Problem Solving, but Evidence for Building New Theories is Lacking
GaryMarcus · x · 2026-08-04
Leading mathematician Terence Tao gave an insightful lecture on the applications and limitations of AI in mathematics.
- Proof indigestion: As AI generates more proofs, academia faces the risk of "proof indigestion," as not all proofs are inherently useful.
- Non-uniformity of math: Math is not a monolithic task. AI might excel at specific types of calculations or problem-solving, but it cannot cover all mathematical work.
- Lack of new theories: Using DeepMind's Astra as an example, while it is strong at solving certain problems, there is no evidence yet that AI can build entirely new mathematical theories.
More from AGI Musings
- AI won’t solve cancer in one shot, but it could make research far cheaper — paraschopra · 2026-08-04
- AGI should be open-source infrastructure, not a closed private weapon — NoiseTraditional2699 · 2026-08-04
- Google reportedly pre-builds compute for AI self-improvement, with RSI eyed for 2027–2028 — kimmonismus · 2026-08-04
- Musk says lifespan doubling may be possible, but argues the real fight is over scale — r0ck3t23 · 2026-08-04
- What safeguards exist if AI can quietly rewrite shared memories? — PierceLilholt · 2026-08-04
- Andrew Chen says AI may make PR matter again as models read TechCrunch — andrewchen · 2026-08-04