AI researcher calls foom and hard takeoff 'pseudoscience', says AI-assisted R&D is already priced in
sebkrier · x · 2026-11-08
Responding to a question about what evidence would convince him of 'real' recursive self-improvement, researcher sebkrier argues:
- Foom / hard takeoff narratives are pseudoscience in his view.
- Automation of the model development stack and productivity gains are already priced into his model of the world — this is AI assistance, not closed-loop autonomous R&D.
- A 1.9x step change isn't the same as sustained exponential or hyperbolic growth across all capabilities, though he still expects rapid progress.
- Signals that might move him: similar trends on ecologically valid non-coding/maths evals, or sustained exponential growth in the ECI index — suggestive but not determinative.
Related event: Anthropic Researcher Calls Intelligence Explosion Pseudoscience(2 posts)→
More from AGI Musings
- AI Alignment Discourse Through the Years: From Doom to a Suboptimal Shade of Blossom — voooooogel · 2026-11-08
- AI companies prying open imminent RSI with humanity's fate, meme'd — JeffLadish · 2026-11-08
- 'AI-generated' flag is no longer a proper low-quality content filter, argues X user — nima_owji · 2026-11-08
- "I'm building a one-person company" — then he introduced his three agents — tech__unicorn · 2026-11-08
- When AI can prove someone is lying, does anyone still have a right to lie? — PierceLilholt · 2026-11-08
- Would a misaligned ASI really destroy humanity, or repurpose our brains as compute? — Felizzle · 2026-11-08