I’ve been a programmer for almost my whole life, I do it for a living and as a hobby. I’ve spent the last few days giving LLM-assisted coding an earnest shot.

I’m pretty AI-skeptical, but I’m being as unbiased as possible when I say: we have nothing to worry about (yet). Claude 3.7 Sonnet makes baffling choices on even the smallest bits of code.

Then when you say to it, “Why would you write it like that?”, it thanks you for correcting it and apologizes.