Newsletter Archives
-
AI slop
I spotted a post from Daniel Stenberg complaining about bug reports that turn out to be nothing but what he deems “AI slop” — the bug reporter got vulnerability details by using various AI tools and did not independently verify or confirm that the bug was real.
Bug reporting is a fine art. A good bug report needs to explain what operating system is installed and what steps or actions triggered the event in question. Most of us do not explain bugs very well at all.
Conversely, it’s sometimes hard to distinguish between a bug and an intended, dribbled-out change. All we know is that something changed and we often can’t quite explain what happened. If it’s hard for humans to determine whether something is a bug, is it any wonder that AI doesn’t find bugs? With many vendors using AI to do more of their coding, does this bode well for software?