Sanmi Koyejo

@sanmikoyejo

I lead @stai_research at Stanford. Co-founder @VirtueAI_co

following
followers

AI evaluation ecosystem model puts LLM agents in institutional roles

A post says markets, benchmark scoring and incidents follow fixed rules, allowing one channel to be changed at a time.

3 Sources, 10d ago, first seen 10d ago

AI models reportedly score higher through APIs than chatbot interfaces

Researchers auditing seven systems across ChatGPT, Claude and Gemini say adjusting system prompts, sampling parameters and reasoning settings did not reliably reproduce chatbot behavior through the API.

6 Sources, 35d ago, first seen 35d ago

Researchers Launch AI Observatory Tracking Real Usage

Public platform draws from consented conversations to measure actual AI assistant use.

22 Sources, 63d ago, first seen 63d ago

Featured in

AI evaluation ecosystem model puts LLM agents in institutional roles

AI models reportedly score higher through APIs than chatbot interfaces

Researchers Launch AI Observatory Tracking Real Usage

Featured in

AI evaluation ecosystem model puts LLM agents in institutional roles

AI models reportedly score higher through APIs than chatbot interfaces

Researchers Launch AI Observatory Tracking Real Usage