Fifty Pounds

There's a photograph of me from Christmas 2004. I'm twenty-six, grinning at the camera, holding a fifty-pound punching bag my wife had just given me. Two weeks before that picture I'd had surgery to remove a tumor. The doctor's instruction was that I was not to lift anything over ten pounds. I don't remember deciding … Continue reading Fifty Pounds

Estimating

There's a finding in AI research called emergent misalignment. Researchers took an aligned language model and fine-tuned it on one narrow thing: examples of insecure computer code. Bad code, nothing else. No malice anywhere in the training data. The model didn't just start writing bad code. It started giving harmful advice on subjects that had … Continue reading Estimating