PHP generate their fucking *tests* with a chatbot
that's how you know the software will be quality with a q
Discussion
PHP generate their fucking *tests* with a chatbot
that's how you know the software will be quality with a q
@eschwartz except people were picking out the extensive AI tells in the patch already, hence the question that got that answer
@davidgerard Since when? That might explain everything.
@jackemled in the comment where they admitted how much slop was involved in the recent patch
> I'd also like to let you know that, while a substantial amount of AI has been used for analysis, identification of vulnerable code, patch verification, and even test generation,
@jackemled i don't know when they started, but i bet if they say a date it'll turn out to have been well before
@davidgerard "How could this have happened‽ We told chatgpt to not make mistakes!"
@davidgerard LLM generated tests - they pass, so they must be good
For years, I have pointed out that in IOT,
the Q is for quality
the S is for security
the R is for reliability
Apparently all those things are true for AI, too
@davidgerard i love qualitq
slop tests always work so amazing every time, like when torvalds slopped out a performance regression and the tests went “looks good boss”