@bibliolater I want to see the AI’s side projects

@bibliolater @ai I used copilot to fix some bugs in a PR that I was testing. It worked. Although not on the first run, it weirdly created a mess on the first run that was so bad I even realized it was bad. But after I tried it again it worked.
I also tried bigger stuff but if you cannot check the code it feels more like a slot machine, even the smallest error sends you in the woods. I feel like it certainly adds to the process but you still need to know what you are coding.

@bibliolater @ai Good article, but please don't start flooding Open Source Projects with LLM generated PRs.