Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

OpenAI is now everything it promised not to be: closed-source and for-profit

vice.com

Post reply on HN
Print view View original

571–580 of 595 posts

Re: OpenAI is now everything it promised not to be: closed-source and for-profit

#571

Post by startupsfail »

mckravchyk wrote: post #556
startupsfail wrote: post #313

Earlier quoted context omitted.

Possible. Coding as we know it might get obsolete. And it is a trillion dollar industry.

I was initially impressed and blown away when it could output code and fix mistakes. But the first time I tried to use it for actual work, I fed it some simple stuff and I even had the pseudo code as comments already - all it had to do is to implement it. It made tons of mistakes and trying to correct it felt like way more effort than just implementing it myself. Then that piece of code got much more complex, and I t…

Considering that OpenAI started instruction following alignment a month ago, with 1k workers, to do engineering tasks, coding might be solved now.

Re: OpenAI is now everything it promised not to be: closed-source and for-profit

#572

Post by HarHarVeryFunny »

goatlover wrote: post #541
HarHarVeryFunny wrote: post #384

Earlier quoted context omitted.

No, it's not just an autocomplete engine. The underlying neural network architecture is a transformer. It certainly can do "autocomplete" (or riffs on autocomplete), but it can also do a lot more. It doesn't take much thought to realize that being REALLY good at autocomplete means that you need to learn how to do a lot of other things as well. At the end of the day the "predict next word" training goal of LLMs is the…

Wouldn't the ultimate intelligence test involve manipulating the real world? That seems orders of magnitude harder than autocompletion. For a theory of everything, you would probably have to perform some experiments that don't currently exist.

> Wouldn't the ultimate intelligence test involve manipulating the real world?

Perhaps, although intelligence and knowledge are two separate things, so one can display intelligence over a given set of knowledge without knowing other things. Of course intelligence isn't a scalar quantity - to be super-intelligent you want to display intelligence across the widest variety/type of experience and knowledge sets - not just "book smart or street smart", but both and more.

Certainly for parity with humans you need to be able to interact with the real world, but I'm not sure it's much different or a whole lot more more complex. Instead of "predict next word" the model/robot would be doing "predict next action", followed by "predict action response". Embeddings are a very powerful type of general-purpose representation - you can embed words, but also perceptions/etc, so I don't think we're very far at all from having similar transformer-based models able to act & perceive - I'd be somewhat surprised if people aren't already experimenting with this.

Re: OpenAI is now everything it promised not to be: closed-source and for-profit

#573

Post by goatlover »

HarHarVeryFunny wrote: post #572
goatlover wrote: post #541

Earlier quoted context omitted.

Wouldn't the ultimate intelligence test involve manipulating the real world? That seems orders of magnitude harder than autocompletion. For a theory of everything, you would probably have to perform some experiments that don't currently exist.

> Wouldn't the ultimate intelligence test involve manipulating the real world? Perhaps, although intelligence and knowledge are two separate things, so one can display intelligence over a given set of knowledge without knowing other things. Of course intelligence isn't a scalar quantity - to be super-intelligent you want to display intelligence across the widest variety/type of experience and knowledge sets - not jus…

The biggest challenge might be the lack of training data when it comes to robotics and procedural tasks that aren't captured by language.

Re: OpenAI is now everything it promised not to be: closed-source and for-profit

#574

Post by mckravchyk »

startupsfail wrote: post #571
mckravchyk wrote: post #556

Earlier quoted context omitted.

I was initially impressed and blown away when it could output code and fix mistakes. But the first time I tried to use it for actual work, I fed it some simple stuff and I even had the pseudo code as comments already - all it had to do is to implement it. It made tons of mistakes and trying to correct it felt like way more effort than just implementing it myself. Then that piece of code got much more complex, and I t…

Considering that OpenAI started instruction following alignment a month ago, with 1k workers, to do engineering tasks, coding might be solved now.

I have just decided to give it another try on a very straightforward thing. I have asked it to get me a Rust function that uses a headless browser to get the HTML of a fully loaded webpage.

ChatGPT:

let screenshot = tab.capture_screenshot(ScreenshotFormat::PNG, None, true).await?;

let html = String::from_utf8(screenshot).unwrap();

>[...] Once the page is fully loaded, the function captures a screenshot of the page using tab.capture_screenshot(), converts the screenshot to a string using String::from_utf8(), and then returns the string as the HTML content of the page.

Of course, it admits to the mistake (sort of, it still does not get):

> You are correct, taking a screenshot and converting it to text may not always be the best approach to extract the HTML content from a webpage.

It's hilarious.

Re: OpenAI is now everything it promised not to be: closed-source and for-profit

#575

Post by DiggyJohnson »

matt_heimer wrote: post #518
DiggyJohnson wrote: post #510

Earlier quoted context omitted.

What's evil about Docker (the company)?

https://news.ycombinator.com/item?id=28369570 https://news.ycombinator.com/item?id=27013865

That is very far away from my personal definition of the word "evil".

Re: OpenAI is now everything it promised not to be: closed-source and for-profit

#576

Post by mitthrowaway2 »

pksebben wrote: post #561
mitthrowaway2 wrote: post #463

Earlier quoted context omitted.

That would make sense for climate change, but the context of this thread is a discussion about AI. Why would that be evidence of an existential threat from AI?

what I think they mean is; there are bigger tigers, and they're already in the house. no sense wasting time stressing out about the cub at the zoo.

I do at least carry hope that climate change will have survivors in all but the worst-case scenarios. And I'm not sure which tiger will strike first.

Re: OpenAI is now everything it promised not to be: closed-source and for-profit

#577

Post by HarHarVeryFunny »

goatlover wrote: post #573
HarHarVeryFunny wrote: post #572

Earlier quoted context omitted.

> Wouldn't the ultimate intelligence test involve manipulating the real world? Perhaps, although intelligence and knowledge are two separate things, so one can display intelligence over a given set of knowledge without knowing other things. Of course intelligence isn't a scalar quantity - to be super-intelligent you want to display intelligence across the widest variety/type of experience and knowledge sets - not jus…

The biggest challenge might be the lack of training data when it comes to robotics and procedural tasks that aren't captured by language.

Yes, and there's always the Sim2Real problem too. Ultimately these systems need to have online learning capability and actually interact with the world.

Re: OpenAI is now everything it promised not to be: closed-source and for-profit

#578

Post by fragsworth »

pixl97 wrote: post #566
fragsworth wrote: post #436

Earlier quoted context omitted.

Sometimes yes, and sometimes yes. I can be convinced, I just want to see the arguments.

The best argument I can make is to say, do not come at the issue with black/white thinking. I try to look at it more of 'probability if/of'. Myself, I think the probability of a human level capable intelligence/intellect/reason AI is year 100% in the next decade or so, maybe 2 decades if things slow down. I'm talking about taking information from a wide range sensors and being able to use it short term thinking, and…

I don't see why the path of AI technology would jump from 1) "horribly incompetent at most things" to 2) "capable of destroying Humanity in a runaway loop before we can react"

I suspect we will have plenty of intermediary time between those two steps where bad corporations will try to abuse the power of mediocre-to-powerful AI technology, and they will overstep, ultimately forcing regulators to pay attention. At some point before it becomes too powerful to stop, it will be regulated.

Re: OpenAI is now everything it promised not to be: closed-source and for-profit

#579

Post by credit_guy »

Whenever I see such righteous anger, my first instinct would be to ask: Would you not do that same if you were in the same position?

These guys came up with potentially the most important technological advancement since the start of the Industrial Revolution. Many years ago they didn't dare think they'd ever get here. But now they got here. And they can become billionaires.

That person out there who would say no to becoming a billionaire, please pick the first stone and throw it.

Re: OpenAI is now everything it promised not to be: closed-source and for-profit

#580

Post by mlsu »

credit_guy wrote: post #579

Whenever I see such righteous anger, my first instinct would be to ask: Would you not do that same if you were in the same position? These guys came up with potentially the most important technological advancement since the start of the Industrial Revolution. Many years ago they didn't dare think they'd ever get here. But now they got here. And they can become billionaires. That person out there who would say no to b…

I'd call myself the Big Money Company, and then make a billion dollars.

What I wouldn't do is call myself the Teeny Tiny Non Money No Money Here Company, and then make a billion dollars, and then while making my second billion dollars, print on a gilded pamphlet that I wasn't actually making a billion dollars, I was a just a teeny smol bean with only pennies who was impoverished.

Post reply on HN