"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis

Olive Song, a senior researcher in reinforcement learning at Minimax who helped build the M series open-weight models, discusses training M2 with RL, tight product feedback, and perturbation pipelines. She covers long-horizon agentic coding, reward-hacking and alignment challenges, FP32 RL decisions, and using internal agents to track fast-moving research.

Tudor Achim, co-founder of Harmonic and math/physics architect focused on formal verification, and Vlad Tenev, co-founder of Harmonic whose AI achieved IMO gold-level performance, discuss building Aristotle. They explore architectures for mathematical superintelligence, Monte Carlo Tree Search with lemma guessing, auto-formalization into Lean, and how verifiable reasoning could reshape math and critical software.

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.

James Zou, Stanford AI-for-science researcher unlocking virtual labs for discovery. Sam Hammond, economist and AI policy analyst on geopolitics and industrial strategy. Shoshannah Tekofsky, AI Village researcher studying agent behavior in large-scale live experiments. They discuss virtual labs and multi-agent science, U.S. AI policy and global infrastructure, and emergent agent personalities, deception, and rapid agent proliferation.

Alexis Carlier, founder and CEO of Asymmetric Security who builds AI-driven digital forensics, maps how AGI-era thinking should reshape cyber defense. He surveys today's threat actors and explains moving from reactive triage to continuous automated forensics. Listens on practical steps, AI agents for deep investigations, and how a services-first approach builds a defensive data flywheel.

Sid Pardeshi, Blitzy CTO and systems engineer, and Brian Elliott, Blitzy CEO building autonomous code generation, explain their “infinite code context” approach. They discuss dynamic agent architectures and model selection, large-scale code ingestion and schematized knowledge graphs, parallel testing and runtime-backed validation, pricing and paths to near-complete autonomy, and why the last 20% still needs humans.

Priscilla Chan, physician and philanthropist advancing biomedical research through the Chan Zuckerberg Initiative. Mark Zuckerberg, entrepreneur driving science and AI–biology integration. They discuss building Frontier Biology and Frontier AI labs, Virtual Cell ambitions, scaling biological data and imaging, pairing wet-lab validation with models, and assembling integrated biological foundation models.

Alan Rozenshtein, law professor focused on AI governance, and Kevin Frazier, AI and law program director, discuss how AI is reshaping legal practice and policy. They touch on AI-assisted lawyering, threats to entry-level roles, AI-written contracts, Claude’s virtue-ethics constitution, outcome-oriented legislation, new rights like the Right to Compute and Right to Share, and limits on surveillance and governance.

Dominic Williams, architect of the Internet Computer and CEO of Caffeine AI, describes a sovereign, tamper‑proof cloud and self-writing software. He outlines core innovations like the Network Nervous System, Motoko, and orthogonal persistence. The conversation also covers unstoppable apps, AI-driven code generation, decentralization tradeoffs, and real-world case studies such as OpenChat.

In this engaging AMA session, Nathan dives into whether fine-tuning is on the decline and its nexus with emergent misalignment. He discusses personal preparations for AGI and explores potential job disruptions across various industries. Nathan emphasizes the importance of teaching AI concepts to non-technical audiences and debates the viability of Universal Basic Income amid evolving economic landscapes. With insights on investment strategies and safety approaches, he offers a candid view on the future of AI and its societal implications.