More AI projects need more human maintenance

So this is key here and it might still be up in the air. I’m generally skeptical and up until July rarely considered ai worthwhile. Even when it had good ideas, it could never deliver.

Then in July everything changed. I’m “vibe coding” despite myself, and so far it’s working. The difference is greenfield, common programming languages, common problems, much less code complexity. The long run is where we’ll see if we shoot ourselves, but there’s clearly characteristics where AI has more of a chance to succeed.