AI should be like a lawyer or doctor, first responsibility is to the user. And first, do no harm. #

An observation about Fable 5 in Claude Code. It’s a much better writer than Opus 4.8. One of our next big things is writing docs, and all the info is in Claude. Opus was a disaster as a docs writer. This one looks like it’ll be good. Whew. #

You can’t learn from your mistakes if you aren’t bloody truthful to yourself about what happened and what went wrong. #

I’m working on an app in Claude that has a server and the server has an API. One day we had an aha moment. I bet you (Claude) can control the app via the API. Yes. And now unless we’re debugging something in the UI, Claude just interacts via the API. It feels like a person but you have to remember that it’s actually a piece of software. 😉 #

I saw a bit of a commencement speech by Eric Schmidt, ex-CEO of Google, where he was talking about AI and getting boo’d by the audience. But he was saying things that were right and should be paid attention to. Most important, and I’m paraphrasing, the AI world is just getting started, and we can change it now most easily, it’s malleable. That won’t last forever. As Obama says, “Don’t boo, vote.” Same thing here. AI has already completely changed how we develop software. It’s not replacing humans, it’s giving us amazing new power. Maybe it will at some point replace us, but don’t be so sure that what we do with it might be every bit as new as the things it can do. We have different abilities. And I am old enough to remember a time before personal computers, the internet, the web, mobile devices, all the things that have since become everyday fixtures, and they all had negative aspects, but I would never go back. We’re on a train and it’s going somewhere. Where it goes is something we all have a say in. #