AnalysisAI ModelsNovember 15, 2026

TypeSafe AI's Diogo Almeida argues for machine-native AI over RLHF

Read original source →youtube.com

At CODE Summit 2026, TypeSafe AI cofounder Diogo Almeida argued RLHF optimizes models for human assistance rather than reliability, using an "Einstein 95% of the time, Mr. Bean 5%" analogy for unreliable LLM automation. He makes the case for machine-native AI with Jev and discusses KV cache limits for reliable agents.

People · Diogo Almeida

1 source

More stories today

Open the live feed