Is GPT getting worse?
Is GPT getting worse?
This question comes up every few months, sometimes with acknowledgement from OpenAI, sometimes without.
I think there's some echo chamber behavior at play here, enhanced by our expectations changing over time - of course combined with the fact that OpenAI is updating models offered through ChatGPT without properly versioning them. If I am told the model will be nicer, I will look for and spot nice comments.