From LLM to Provably Safe AI Systems | Mechanistic Interpretability I Max Tegmark, MIT

Exerpts from:

The Impact of chatGPT and other large language models on physics research and education (2023)
“Keeping AI under control through mechanistic interpretability”
Prof. Max Tegmark (MIT)

Credits: https://www.youtube.com/watch?v=RheZlFj3Zp8