ChatGPT Has Secretly Been Running a New Model

OpenAI has quietly rolled out a new GPT-4.0 model within ChatGPT, sparking discussions and speculation across the AI community. While some users have noticed improvements, others remain critical of the changes and the lack of release notes. Get all the details about this update, the reactions from the AI world, and what it might mean for the future of AI development.

The AI Daily Brief helps you understand the most important news and discussions in AI.

[Ben_D.]

Once they improve recursively, we sit back and watch. It will be awesome.

[hotlineoperator]

The good thing about online services is that the service administrator can constantly update it.

[OriginalRaveParty]

Release notes would be very useful, but I do understand that it's quite a nebulous thing to explain.

[miketrago]

Love how you brought up Lovecraft, I love and agree with everything that man ever said

[aigriffin]

Why haven’t they told us thanks so much!❤

[superfliping]

Two categories selection in my queries it ask me what query i like better. Has done this two or three times in the last week

[Shaunmcdonogh-shaunsurfing]

I noticed some really poor results with it today. Normally I’m quite positive about it.

[travisporco]

This idea that you shouldn't use an API doesn't make sense. If you select the specific model over an API you get that model.

[surfercouple]

Didn’t Sam Altman say in the early days he’d roll things out slowly so the public could adapt/adjust? Could this be what we are seeing with these incremental rollouts? GPT4o blew me away yesterday with its accuracy understanding my prompts and also accuracy in image generation. It was really great stuff.

[leeprice]

Thoughts on Tao and the bittensor network??

[Churchofexponentialgrowth]

Oh it's absolutely terrible I even said to it when it activated did you get stupider?

[sullyguy]

I’d like to be optimistic about AI scientists but I’ve also heard that predictions about AI benefits to research are overblown. So it would be nice to see a list of real discoveries start getting cranked out that don’t involve “here is something that barely qualifies as a paper” discoveries. Not like “here’s a chemical humans hadn’t gotten around to discovering but isn’t apparently useful for anything yet” type discoveries. I’d like to know the AI scientist is making some real advances before getting excited about it.

[KevinKreger]

I'm tired of OpenAI. I can't use it. Its a moving and unreliable service.