Hacker News new | ask | show | jobs
by copperx 1073 days ago
I've had similar experiences in the last few days, and it never used to make those kinds of mistakes. It could hallucinate, but it never changed topics randomly or misunderstood prompts.

I haven't checked the API side, but I hope this reduction of capabilities is only happening on the ChatGPT Plus side of things.

It's upsetting. It's like the ChatGPT 4 we had a month ago never existed. Now it can't reason itself out of a paper bag.

1 comments

This behavior is happening in the default model (GPT-3.5). I edited the question as well.