Updated Thoughts on AI Six months ago many of us wrote articles about AI. I certainly did: AI Zealotry, We're All Beginners Again, Multi-Agent Workflows, and Introducing Claude Chic. It feels like time for an update. We've all learned a lot since then. My biggest change in thinking is this: The surprising value of AI today isn't super-human depth. It's super-human breadth. AI isn't usually better than the best human specialist in one narrow domain. Good lawyers are better lawyers. Good accountants are better accountants. Good engineers are better engineers. But AI can be a B+ engineer and a B+ lawyer and a B+ accountant at the same time. That turns out to be extremely useful. A lot of real work lives between disciplines, and humans are pretty bad at crossing those boundaries quickly. The second theme of my work recently has been building feedback systems for agents. Most of my focus on complex projects today is around giving broad context and building systems to deliver good automated…
No comments yet. Log in to reply on the Fediverse. Comments will appear here.