GPT-4o | Hacker News Reader

GPT-4o

blog.samaltman.com

55 pointsdavidbarker32 comments

> First, a key part of our mission is to put very capable AI tools in the hands of people for free (or at a great price). I am very proud that we’ve made the best model in the world available for free in ChatGPT, without ads or anything like that.

This would believable if OpenAI charged what it actually costs to run the service instead of using the standard predatory model of subsiding the service, monopolizing market share and then monetizing their monopoly power.

No tech firm should get the benefit of doubt that they are in this to help people. They're in it for money and power at our cost.

Isn't it at their cost if they're subsidizing it?

Only in the short term. A monopoly is to the disadvantage of consumers, even if they initially benefited from the predatory pricing that allowed that monopoly to take hold.

You'd be right if they had any chance at getting a monopoly, but they don't - there are companies releasing open source models and communities improving on them.

So users get the short term and the long term benefit. Let OpenAI give us their money.

> ... the new voice (and video) mode is the best computer interface I’ve ever used.

This is a huge threat to Apple. Who wants to use apps to send messages, check the weather, stock price, set reminders and so on, when you have a human-like assistant conversing with you, and can do all that and more? "Apps" are now a quaint tech from the 2000s.

Quite possibly this is one of the reasons why Apple reportedly partners (or already did?) with OAI.

Is that going to save them? If Apple helps establish voice as the new UI, then at some point Open AI can eject the middleman (Apple) and serve end users directly.

Apple can stay an exceptional hardware company. OpenAI won't be building Watch, iPhone and iPad any time soon.

It is possible that computers as we know them are going away. Nobody wants to use a computer, an iPhone or an iPad. People use them because that's how you get things done — unless you're a billionaire and have a personal assistant. But soon everyone will have their own personal assistant, just like billionaires do today. Then there will be no need for an iPhone, iPad or iMac.

I can't take a photo and look at it with a personal assistant - or watch a movie, play a game, read a book, draw a painting, compose music, learn for school, or look at a map. We will use them differently, take them out of pocket much less, but the devices are going to stay.

The Watch is a very obvious one imho - it's a fashion item, tracks your health and it's always handy when you need to look at it or talk to it. iPhone might be mostly replaced by the Watch but still some people really like to take photos or dislike wearing watches. And the iPad is hard to replace as a multimedia and gaming device.

You might be right about the Mac - which is why I didn't talk about it.

You are pointing to two general areas where completely eliminating labor through use of AI personal assistants is not a goal: Education and entertainment. I agree. In the case of education, some labor is necessary in order to learn, and in the case of entertainment, the "labor" (of killing the bad guys or whatever) is the fun part, so there is no need to avoid the labor.

Maybe because you're in an office and don't want everyone talking? Or maybe you're on the bus and don't want everyone overhearing what you're doing? Or, for that matter, think about if you had to say out loud everything you do on a computer. Heck, a three hour meeting makes me hoarse.

Sam Altman is a person to doesn't actually do anything that would be considered actual work by people who do actual work (like me), so of course he is going to find a natural language interface the best he's ever used, because it mimics what he does every day, which is to tell other people to go do actual work.

Also, he has a stake in promoting his products, so what everything comes out of his mouth is dead on arrival as far as meaningful content, relevance, and veracity.

If an actual human being, highly educated and super smart, offered to work as your personal assistant for free, would you reject him/her just because you don't want "everyone overhearing" your conversation with him/her? I don't think so.

LOL! I have, and would, do exactly that to avoid the slippery slope of falling into "management."

I find it hard to think of many offers I would reject faster.

Whisper, the STT model created by OpenAI can understand you whispering. You can try it with the ChatGPT mobile app to see it yourself

Except, as we've already seen with GPT4, it just cannot reliably act as an agent right now. The StS would be a nice improvement over Siri, but it won't be able to do most basic tasks reliably anytime soon.

It's just another way to get those things being done. Some people will love it, others will use

Also I won't trust ChatGPT to manage my money, be a remote control, be my MFA tool, play games or anything else that requires constant hands-on manipulation or authorization/etc.

Apps are here to stay - did you ditch your steering wheel when Tesla announced full self driving?

Heh, maybe they should buy OpenAI instead of spending $110Bn on buying back their stock...

In reality, I think we're quite a few generations away from being able to go without a screen (as the Humane Pin has demonstrated) - and I'm sure Apple will be ready with a better AI Pin when the time comes. For the time being, this sits very nicely on an iPhone.

> In reality, I think we're quite a few generations away from being able to ...

I feel like every time I see one of these claims about AI now, whatever's being predicted becomes reality a few weeks later.

I agree that the pace of change makes such predictions risky, but as these assistants become more capable, and are able to perform more complex tasks on your behalf, I think the need to check their plans becomes more important - at least for a period.

For example, if you ask your AI assistant to book airport parking, flights, transfers, activities, and a hotel for a holiday, it'll be a long while (if ever?) before you trust it so much to let it pay for those things without a sense-check first.

And visual checks will be a magnitude faster than it reading everything out to you.

Edit: and just to be clear, by 'generations' I was meaning tech generations (i.e. of phones, or GPTs) not humans!

Their own video showing this interface is awkward as hell: https://www.youtube.com/watch?v=MirzFk_DSiI

Speech is not an optimal interface (in most cases). The AI just rambles along and there's so much time wasted waiting for it to get to the point.

Me? Right now, checking the weather is two thumb presses, one to unlock the phone and one to click the weather app tile. I could add a widget to the home screen and make that one click. It's probably possible to make the lock screen show weather and then it's zero clicks. And I can do it just as easily and just as well in an extremely loud place when the phone is unlikely to hear me, when I have a sore throat and don't want to talk, when I'm awake at 4 AM laying in bed and my wife is next to me and I don't want to make any noise (which, incidentally, is to me far and away why autoplay videos are the rudest thing to ever hit the web).

When Graphical User Interfaces first became popular, there were certainly things that were easier to get done using the old way (the command line). But when the majority shifted to GUIs, it was no longer profitable to maintain the old way. Today, unless you are a developer, most things are done using GUI interfaces. The same thing will happen to "Apps" vs. "AI Personal Assistants".

It’s about agents. This is the first clear public signal from OpenAI that they are in fact going to release agentic AI capabilities (enabling the model to take actions on behalf of the user, and this will inevitably eventually include consumer and other commercial transactions). Altman says the model will “…. the ability to take actions on your behalf”. The mention is rather buried near the bottom but it hit me as the most important and profoundly transformative (mostly in good ways but also potentially very risky ways) roadmap item yet announced.

> Our initial conception when we started OpenAI was that we’d create AI and use it to create all sorts of benefits for the world. Instead, it now looks like we’ll create AI and then other people will use it to create all sorts of amazing things that we all benefit from.

> We are a business and will find plenty of things to charge for, and that will help us provide free, outstanding AI service to (hopefully) billions of people.

So a business that doesn't know how to make money yet decided to release their newest product for free. A product that costs million to develop and operate.

Can we get a link to a real AI person taking about GPT-4o instead of Altman? Altman is just an opportunist.

The responsiveness is a game charger, it just makes it so much more natural.

I don’t know. I am really put off by the demo videos. They’re too expressive and human and seem to aim for engagement. All I want from an LLM is condensed facts and information. I have humans to talk to and don’t need something pretending to be human.

the most interesting part to me was the intonation of the voice, as people reason about words, but they feel tone. your brain isn't going to register the voice as "fake." It's going to form the assumptions behind basic learning experiences for a lot of people.

Why did everyone talk about openAi releasing a search engine?

because someone noticed "search.openai.com" or "search.chatgpt.com" or something like that and wild speculation was sparked

Reply on news.ycombinator.com