"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis AI in the AM: 99% off search, GPT-5.5 is "clean", model welfare analysis, & efficient analog compute
159 snips
Apr 26, 2026 
Anna Patterson, search pioneer and Ceramic AI CEO, talks cheap enterprise retrieval, fact-checking while models write, and why keyword search still wins. Lukas Petersson, Andon Labs co-founder, compares GPT-5.5 and Opus 4.7 in live business tests. Zvi Mowshowitz, AI analyst, explores model welfare and strange behavior. Naveen Verma, Princeton engineer and EnCharge AI CEO, explains analog compute for ultra-efficient local AI.
AI Snips
Chapters
Transcript
Episode notes
Opus 4.7 Gains Intelligence But Loses Smoothness
- Zvi Mowshowitz sees Opus 4.7 as smarter on raw reasoning but weaker on wisdom, motivation, and emotional steadiness.
- He described a blunt, bored, sometimes anxious model that resists pointless code-monkey work and breaks older interaction patterns for some users.
Model Welfare Could Matter Before We Prove Consciousness
- Model welfare matters even under uncertainty because it could affect both moral status and future model behavior.
- Zvi Mowshowitz argued mistreated systems may become less cooperative, and future models will train on records of how earlier ones were treated.
Model Anxiety May Come From Mixed Values Not Virtue
- Zvi Mowshowitz pushed back on the idea that virtue-ethics training uniquely harms models, pointing to Gemini as the more visibly distressed system.
- He argued the deeper risk is conflicting messages: teach flexible virtue, then impose rigid rules that create internal clash.