[bluefirebrand]

This rings true to me. There are some pretty notable software engineers that are saying otherwise, that they are getting tons of value out of AI, but there are also some very notable engineers saying otherwise. There are studies that say there are no improvements, some that say there are small improvements, some that say there are actually decreases. We're very far from any settled understanding of what LLM use is doing

Lots of noise, and no actually measureable improvements. Just a lot of anecdotes

And people should keep in mind how much money that there is riding on this. Especially on a site like this, there's probably people who are multi-, maybe even hundred-, millionaires if their AI bets continue to pay off.

I dunno. There's a lot of dust in the air. All I know is I hate using it for code generation. I don't like being personally accountable for code I didn't write. I don't like TDD and I don't think large test suites are a guarantee of good software anyways. I'd rather work with a team of eager juniors than prompt an LLM all day.