Posts
All the articles I've posted. 46 posts, page 3 of 12.
I tried to catch a language model making a math mistake. My first accepted paper is about why it did not work.
Published: at 03:30 PMMy MATH-AI workshop paper at NeurIPS 2026 tests whether drift in a language model's hidden state can flag and fix math errors mid-solution. It does neither, and getting to that answer taught me more than a positive result would have.
People don't speak in order
Published: at 09:50 PMPeople correct themselves and add details as they speak. I'm exploring how an assistant could turn those repairs into an explicit, checkable task state.
When agents fail because their picture of the world is stale
Published: at 03:05 AMStale facts, lost corrections, and unchecked tool results can derail an agent. I'm exploring how explicit state and verification might help.
AI security is not web security
Published: at 02:45 AMPrompt injection, the lethal trifecta, and why the input validation habits from web security stop working once a language model is reading your data.