AI modelsNovember 09, 2026
22
šŸŒ¶ļø

Claude Fable 5.2: Almost AGI, Still Jagged

An analysis characterizes Anthropic's Fable 5.2 as "Jagged AGI": the model reaches expert level on select tasks such as game design but lacks self-directed judgment and taste.

#Anthropic#AGI#LLM evaluation#AI capabilities
Claude Fable 5.2: Fast AGI, aber zackig
Share Article
šŸ”„ What happened Ethan Mollick tested Anthropic's new Claude Fable 5.2 pre-launch. It hacked a USB keyboard into a game console, drafted a business plan, and reconstructed a legendary college course – all from single prompts. šŸ’” Why it matters Fable 5.2 hits expert level in many domains but fails at running a business independently. Mollick calls it "Jagged AGI": nearly general, yet with unpredictable gaps – a shift from fixing errors to exercising real judgment. ⚔ Our take Celebrating a model as "almost AGI" because it builds games from a prompt should also ask why it pitches a business plan with only a 20–40% success rate. The jaggedness isn't the problem – the confidence is.
The title, summary and analysis of this item were produced automatically by an AI system and have not been editorially reviewed. They may contain errors, bias or omissions — when in doubt, read the linked original source.