Sara Hooker podcast: what comes after scaling — adaptive AI and non-verifiable tasks
ziv_ravid · x · 2026-09-17
A new episode of The Information Bottleneck features Sara Hooker, co-founder and CEO of Adaptation Lab, on what comes after scaling:
- Why today's models are mostly static, and what it takes to build systems that continuously adapt to new tasks, data, and users
- AutoScientist and self-improving AI
- Why non-verifiable tasks may be the next bottleneck
- Why interfaces may matter as much as models
- Open vs. closed models, distillation, and Chinese AI labs
- AI safety, cyber risk, and biorisk
- What comes after Transformers and tokenization
Available on the show's website, YouTube, and all podcast apps.
More from AGI Musings
- UW researchers push back on AI doom fears: the real risk is unaudited agent systems — lazowska · 2026-09-17
- Carnegie scholar: 'Pacing' frontier AI is sensible, but the US needs federal incident reporting now — mattsheehan88 · 2026-09-17
- ECCV seethes as GenCeption shows video models can swallow traditional CV research — CSProfKGD · 2026-09-17
- EA orgs accused of 'psyop' posing as anti-EA amid AI safety factional split — beffjezos · 2026-09-17
- Ari Schulman: the real question isn't whether EAs should rule us, but whether something momentous is happening with AI — deanwball · 2026-09-17
- Metzger: If Someone Gives You an AI Extinction Probability, Ask for the Math — nptacek · 2026-09-17