But there's something more. It's the benefits of a first-look perspective. With these tools that's lost.

Also, it seems like we've entered phase two of sorts in the sense of we solve LLMish #1 issues by running it by #2. So writing and reviewing. I wonder what #3 will be. Maybe confirming.

I feel like I've gotten much better at reading code now that I read so much of it(generated).

I don't mind it. As long as I'm not the first human filter, I'll review a colleagues AI generated code without kicking and screaming.