Is it just me, or has ChatGPT gotten really lazy lately?
It used to write code carefully and explain the reasoning, but these days it feels like it just throws something together and ends with "This should work" lol
Did the model change... or am I expecting too much? I also wonder if I just need to write better prompts.
4 answers
That's why lately I just tack on "as if explaining to a beginner." Even just this makes a pretty noticeable difference.
Agreed, that's really how it feels these days lol. Just toss in some code and you're done.
I felt the same way, so I tried changing the prompt. When I explicitly said things like "explain step by step why it's written this way" and "include two alternatives and their trade-offs too," it made a clear difference. It doesn't seem like the model got lazy; it's more like the default output got shorter. I guess the answer is to just bake the requirements in more specifically.
Well, that's a bit... Maybe it's not that the model got lazy, but that you gave it a half-assed prompt? And the fact that it used to give good results might just be because your expectations were lower back then. Do you even have logs of the same question to compare? If not, it's probably just your imagination.