Karpathy: the intelligence explosion began decades ago — both AI skeptics and boosters are wrong
binarybits · x · 2026-09-21
Timothy B. Lee of Understanding AI recaps the AGI debate around Berkeley's The Curve conference and argues both camps are wrong. Boosters included AI 2027 co-author Daniel Kokotajlo and MIRI's Nate Soares; skeptics like Gary Marcus cite MIT NANDA's finding that 95% of corporate AI pilots show little return and a projected $800B revenue shortfall by 2030, advocating narrow tools like AlphaFold instead. Lee sides with Andrej Karpathy's take that "we've been in an intelligence explosion for decades" — real progress, but gradual rather than imminent takeoff — and points to Narayanan and Kapoor's critical writing for balance.
More from AGI Musings
- She turned down $65M for her AI, gives away robots to save an endangered language — danielleboyerr · 2026-09-21
- Colin Breck: I don't want to read what you didn't write — AI-generated docs are unreadable — vykthur · 2026-09-21
- India Needs Independent AI Measurement Capacity, IIT Madras Scholars Argue — ravi_iitm · 2026-09-21
- Ezra Klein: Ban Labs From Letting AI Train AI Before It's Too Late — RazRazcle · 2026-09-21
- Instead of Pacing AI, We Should Boost Society's Absorptive Capacity — mrdrozdov · 2026-09-21
- Active Learning and Causal AI Challenge the Claim That Only Humans Can Run the Theory-Experiment Loop — maier_ak · 2026-09-21