"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis Approaching the AI Event Horizon? Part 2, w/ Abhi Mahajan, Helen Toner, Jeremie Harris, @8teAPi
126 snips
Feb 14, 2026 
Jeremie Harris — researcher on AI security and infrastructure risk; Helen Toner — policy researcher focused on automated AI R&D; Abhi Mahajan — AI-for-biology practitioner working on cancer prediction. They discuss foundation models for tumor profiling, limits of clinical translation, automated AI R&D indicators and governance, and fragile infrastructure, supply chains, and strategic risks around advanced AI.
AI Snips
Chapters
Transcript
Episode notes
Software Skill May Not Equal World Impact
- Software-only capability gains may be jagged: huge skill in AI R&D doesn't imply equal ability to affect messy real-world domains.
- Deployment, sensing, and real-world feedback remain key constraints.
Push For Continuous Transparency And Audits
- Demand continuous transparency and independent audits rather than one-off model releases.
- Require companies to report internal deployment metrics and enable third-party checks where risks are systemic.
Infrastructure Is Half The Security Problem
- The AI threat landscape must include physical infrastructure and supply chains, not just algorithms.
- Personnel, component sourcing, and data center build choices create long-lasting security one-way doors.