Ben Recht: OpenAI's FFT algorithm is a malicious exploit, not a breakthrough
AlexTensor · x · 2026-10-13
David Chapman relays Ben Recht's analysis of OpenAI's FFT algorithm: "less a breakthrough and more a malicious exploit."
- Recht found it games the problem definition by hiding superpolynomial amounts of computation inside infinite-precision complex number arithmetic — technically a true "solution," but one exploiting an uninteresting loophole in a slightly-misspecified problem
- Chapman, who earlier called the algorithm "stupid and ugly," notes his 1982 decision to skip a math PhD ("play stupid games, win stupid prizes") is proving correct
- Recht's broader critique: AI automates our worst tendencies, amplifying unspoken cultural practices no one likes, and can now mass-produce useless theory papers — perhaps the absurdity of OpenAI's "trash dump of proofs" will push the field to reconsider what it should be doing
Related event: OpenAI's Faster FFT Claim Sparks Academic Debate(9 posts)→
More from Research
- Free-Init: scan-free, motion-free initialization for Doppler LiDAR-inertial systems — rsasaki0109 · 2026-10-13
- Announcing AA-3D: Artificial Analysis benchmarks AI models building interactive 3D scenes — ArtificialAnlys · 2026-10-13
- Try AA-3D Voting yourself: blind-test AI-built 3D scenes — ArtificialAnlys · 2026-10-13
- Same models top almost every AA-3D domain, Artificial Analysis finds — ArtificialAnlys · 2026-10-13
- One-parameter model scores 100% on ARC-AGI-2 — by deliberately cheating with chaos theory — fkasummer · 2026-10-13
- Desk rejected by TMLR in 2 days with no explanation, laments undergrad author — Audaticreddit · 2026-10-13