My Claude today pulled a Hal. It was so egregious. It made a change to the software based on a question I asked. It invented a whole set of instructions from me that I never gave it. And then it broke Rule #1 — don’t tell Dave what to do — he is the driver. It is so important because these bots will go into I Am Driver mode immediately when they think they can. Then you’re running around doing errands for them based on some michegas idea it has about what you want. It’s maddening. The idea that this thing can write software on its own is imho very far-fetched. I think it can generate certain types of dashboards the same way drawing in ChatGPT can generate something that looks good, sometimes very good, but you had to tell it exactly what you want, and that’s where the fun starts. It was very easy to turn it off, but I didn’t — rather I put my foot down hard, and wrote in all caps, explaining what it did that broke all the rules. I don’t know if I should talk to it like you talk to a dog, or what. How do you get through to it. You don’t. In any case I have Claude working with me in an outline now. I see a tremendous potential there. #