AI having poor quality is a bad take like over a year ago.

Meh. Just this week, I've had two Sonnet 4.8 agents generate, in parallel, a 2000 line wall of brittle bullshit, and a well architected solution with 20% of the amount of code, to the same problem, from the exact same initial context, and very similar prompts. Come on, they can do poor quality work too.

Depending on what you or another means by "quality", it may not have any at all.