Can we read the language model's mind?

This month we saw some progress in machine interpretability, the ability to peer into the black box of language models. Here are a few readings to bring you up to speed. Read as much or as little as you like.

https://www.theatlantic.com/technology/2026/10/ai-neuron-blog-post/689058/
https://already-clean.goodfire.com/
https://files.twohands.co/chalmers2025.pdf

We're trying something different. Let's gather for drinks and snacks after work and just make friends and discuss the topics freely. There will be no coordinated discussion, only the assurance that most of us have done the reading, or anyway wish we had.