A year ago I would agree. But now with sufficient context length the models actually do what they're told.

How many people still know how to tell them to do the right things?

Not really, no. Nothing has changed versus a year ago, you still have to babysit them to make sure they aren't doing something insane.

Hugely depends on the size of the task. I've found that telling them to ask questions removes a lot of uncertainty.

This is HN you will be voted to death by the local for talking against the AI or the last tech invention.

Good thing votes don't matter but the mob continues to be full of real idiots.

Votes unfortunately do matter in the sense that few people will see a [dead] comment like ggp.

Meh, I don't think so. They have no way to ascertain what tools were used to write a piece of code. A reasonable solution is to simply hold all code and its contributors to the same standards as before. At Debian's scale, "no LLMs" is really not enforceable; whereas "no shit code" and "no incompetent maintenance" are pragmatic policies they already enforce

This combination of username and opinion are very confusing.