Summary
Dan Schwarz, CEO and co-founder of Future Search, joins Nathan and Pash on the Cognitive Revolution podcast to make a bold claim: AI superforecasters have finally arrived. Schwarz, who previously built Google’s internal prediction market and served as CTO at Metaculus, argues that complex multi-agent architectures are now matching — and in some cases beating — top human experts on real-world forecasting benchmarks, a thesis recently validated in a notable Scott Alexander post titled “The AI Superforecasters Are Here.”
The conversation traces Future Search’s evolution from its 2023 founding, when early systems built on Claude 2 drew little enthusiasm, to its current pipelines using frontier models including Claude, Fable, and Mythos. Schwarz pushes back against purely theoretical compute-scaling arguments for AGI timelines, instead emphasizing verifiable capability milestones and the ways government regulation and commercial incentives are actively reshaping the race. He also describes Fable as exhibiting AGI-like properties in its ability to manage its own compute allocation and orchestrate other tools.
Host Nathan also previews an upcoming trip to China for the World AI Conference (WAIC) in Shanghai and visits to AI safety research groups at Tsinghua University and several Chinese AI companies — framing it as a firsthand attempt to understand how Chinese researchers conceptualize AGI and the felt reality of AI development on the ground.
📺 Source: Cognitive Revolution “How AI Changes Everything” · Published July 07, 2026
🏷️ Format: Podcast
