Ask HN: Code quality in projects developed with intensive use of AI tools

Have you had the opportunity to carefully analyze code from such projects, whether open-source or within companies? What have you seen "in the wild"?

How would you rate the quality of that code: high, low, or average? How would you compare those projects to conventional, "hand-made by humans" ones? What makes them stand out or gives them away?

What’s your overall conclusion, does this represent a significant improvement and acceleration, or is it more of a "meh"? Because even these tools, once you start configuring them and trying to make them work properly, tend to take up a lot of time, energy, and learning.