Stefano Ermon

The Parallel Model Wins at Inference

Today's chatbots write one word at a time. Stanford's Stefano Ermon is betting the next generation writes them all at once.

Oct 3, 2026 · 4 min read