GPT-4o • Hacker News Board

Re: GPT-4o

#201

Post by bamboozled » Mon, May 13, 2024, 6:09 PM UTC

brainer wrote: post #151

OpenAI's Mission and the New Voice Mode of GPT-4 • Sam Altman, the CEO of OpenAI, emphasizes two key points from their recent announcement. Firstly, he highlights their commitment to providing free access to powerful AI tools, such as ChatGPT, without advertisements or restrictions. This aligns with their initial vision of creating AI for the benefit of the world, allowing others to build amazing things using their t…

access to user information,

Sam, please stop ok, those things you saw on tv when you were a kid? They were dystopian movies, we don’t want that for real, ok?

Re: GPT-4o

#202

Post by willsmith72 » Mon, May 13, 2024, 6:09 PM UTC

> We plan to launch support for GPT-4o's new audio and video capabilities to a small group of trusted partners in the API in the coming weeks.

So no word on an audio api for regular joes? that's the number 1 thing i'm looking for

Re: GPT-4o

#203

Post by jamilton » Mon, May 13, 2024, 6:09 PM UTC

yumraj wrote: post #72

In the first video the AI seems excessively chatty.

Yeah, I would hope that custom instructions would help somewhat with that, but it is a point of annoyance for me too.

Re: GPT-4o

#204

Post by frabcus » Mon, May 13, 2024, 6:10 PM UTC

OutOfHere wrote: post #18

I don't see 4o or anything new at https://platform.openai.com/docs/models Overall I am highly skeptical of newer models as they risk worsening the completion quality to make them cheaper for OpenAI to run.

It's there now! And still 128k context window

Re: GPT-4o

#205

Post by joshstrange » Mon, May 13, 2024, 6:10 PM UTC

MBCook wrote: post #142

Too bad they consume 25x the electricity Google does. https://www.brusselstimes.com/world-all-news/1042696/chatgpt...

> Too bad they consume 25x the electricity Google does.

From the article:

"However, ChatGPT consumes a lot of energy in the process, up to 25 times more than a Google search."

And the article doesn't back that claim up nor do they break out how much energy ChatGPT (A Message? Whole conversation? What?) or a Google search uses. Honestly the whole article seems very alarmist while being light on details and making sweeping generalizations.

Re: GPT-4o

#206

Post by willsmith72 » Mon, May 13, 2024, 6:11 PM UTC

DataDaemon wrote: post #129

Now, say goodbye to call centers.

and say hello to your grandma getting scammed

Re: GPT-4o

#207

Post by UncleOxidant » Mon, May 13, 2024, 6:12 PM UTC

Looking at the demo video, the AIs are a bit too chatty. The human has to often interrupt them.

A nice feature would be to be able to select a Meyer's Briggs personality type for your AI chatbot.

Re: GPT-4o

#208

Post by michalf6 » Mon, May 13, 2024, 6:13 PM UTC

I cannot find the mac app anywhere. Is there a link?

Re: GPT-4o

#209

Post by Painsawman123 » Mon, May 13, 2024, 6:13 PM UTC

My main takeaway is that Generative AI has hit a wall... New paradigms, architectures and breakthroughs are necessary for the field to progress but this begs the question, If everyone knows the current paradigms have hit a wall, Why is so much money being spent on LLMs ,diffusion models etc,which are bound to become obsolete within a few(?) years?

Re: GPT-4o

#210

Post by dragonwriter » Mon, May 13, 2024, 6:13 PM UTC

msoad wrote: post #58

They are admitting[1] that the new model is the gpt2-chatbot that we have seen before[2]. As many highlighted there, the model is not an improvement like GPT3->GPT4. I tested a bunch of programming stuff and it was not that much better. It's interesting that OpenAI is highlighting the Elo score instead of showing results for many many benchmarks that all models are stuck at 50-70% success. [1] https://twitter.com/Lia…

> As many highlighted there, the model is not an improvement like GPT3->GPT4.

The improvements they seem to be hyping are in multimodality and speed (also price – half that of GPT-4 Turbo – though that’s their choice and could be promotional, but I expect it’s at least in part, like speed, a consequence of greater efficiency), not so much producing better output for the same pure-text inputs.