News for hackers. Stuff that matters.

A comment on OpenAI Employee: GPT-4 hasn't gotten worse since March. Parent comment

Re:OpenAI Employee: GPT-4 hasn't gotten worse since March by throwuwu on Thursday June 01, 2023 @04:03PM (#36156601)
The refusals are new. I’ve gotten a few myself recently. I’ve never seen them before. They’re not moderation related and they always include some text about the task being too complex.
[ Parent | on HN ]
Re:OpenAI Employee: GPT-4 hasn't gotten worse since March by ribosometronome on Thursday June 01, 2023 @05:24PM (#36157540)
Are there any 'complex tasks' you've had refused you can share?
[ Parent | on HN ]
Re:OpenAI Employee: GPT-4 hasn't gotten worse since March by insanitybit on Thursday June 01, 2023 @05:24PM (#36157544)
That sounds like it could just be a cap applied to free users. That's hardly the same as a new model.
[ Parent | on HN ]
Re:OpenAI Employee: GPT-4 hasn't gotten worse since March by throwuwu on Friday June 02, 2023 @06:51AM (#36163513)
I subscribe and I use GPT-4
[ Parent | on HN ]
Re:OpenAI Employee: GPT-4 hasn't gotten worse since March by sundarurfriend on Thursday June 01, 2023 @09:26PM (#36159727)
Disappointing that GPT seems to be getting more Bard-ish, rather than the other way around. Bard would refuse the request for the simplest of things ('what is kanban' got a "As an LLM I cannot..." response on its first day), which was one of the main things that gave a bad first impression of it. Now, perhaps in an inevitable game of corporate cover-your-ass, OpAI also seems to be going the same way.
[ Parent | on HN ]