Miles Brundage analyzes sources of Anthropic's short timeline confidence
Miles_Brundage · x · 2026-08-24
Miles Brundage speculates that Anthropic's internal confidence in short AGI timelines stems from:
- Concrete knowledge: Explicit details on upcoming training runs and compute ramps.
- Direct experience: Visceral confidence or fear gained from multiple scaling efforts.
- Social proof: Consensus within their community.
He notes that this evidence might not be fundamentally different from external extrapolation, but it carries more weight due to its concreteness and immediacy.
More from AGI Musings
- Controversial View: Those Calling AI Models 'Bad' Are Flawed — repligate · 2026-08-24
- Stanford's Erik Brynjolfsson: AI Job Apocalypse Unlikely, Entry-Level Roles Squeezed — lmoroney · 2026-08-24
- Poll of 25 countries: Americans most anti-AI; Korea, India, Nigeria most pro-AI — ivan_bezdomny · 2026-08-24
- ChatGPT Is the Most-Read 'Author' in History, ~200x Any Human Writer — TuhinChakr · 2026-08-24
- 'What actually matters is what's true': AI circle debates whether CEOs honestly state catastrophe risk — sjgadler · 2026-08-24
- Why watch human chess or sports? Because we are unpredictable — PeterDiamandis · 2026-08-24