OpenAI's Sam Altman on iPhones, Music, Training Data, and Apple's Controversial iPad Ad (youtube.com) 37
OpenAI CEO Sam Altman gave an hour-long interview to the "All-In" podcast (hosted by Chamath Palihapitiya, Jason Calacanis, David Sacks and David Friedberg). And speaking on technology's advance, Altman said "Phones are unbelievably good.... I personally think the iPhone is like the greatest piece of technology humanity has ever made. It's really a wonderful product."
Q: What comes after it?
Altman: I don't know. I mean, that was what I was saying. It's so good, that to get beyond it, I think the bar is quite high.
Q: You've been working with Jony Ive on something, right?
Altman: We've been discussing ideas, but I don't — like, if I knew...
Altman said later he thought voice interaction "feels like a different way to use a computer."
But the conversation turned to Apple in another way. It happened in a larger conversation where Altman said OpenAI has "currently made the decision not to do music, and partly because exactly these questions of where you draw the lines..."
Altman: Even the world in which — if we went and, let's say we paid 10,000 musicians to create a bunch of music, just to make a great training set, where the music model could learn everything about song structure and what makes a good, catchy beat and everything else, and only trained on that — let's say we could still make a great music model, which maybe we could. I was posing that as a thought experiment to musicians, and they were like, "Well, I can't object to that on any principle basis at that point — and yet there's still something I don't like about it." Now, that's not a reason not to do it, um, necessarily, but it is — did you see that ad that Apple put out... of like squishing all of human creativity down into one really iPad...?
There's something about — I'm obviously hugely positive on AI — but there is something that I think is beautiful about human creativity and human artistic expression. And, you know, for an AI that just does better science, like, "Great. Bring that on." But an AI that is going to do this deeply beautiful human creative expression? I think we should figure out — it's going to happen. It's going to be a tool that will lead us to greater creative heights. But I think we should figure out how to do it in a way that preserves the spirit of what we all care about here.
What about creators whose copyrighted materials are used for training data? Altman had a ready answer — but also some predictions for the future. "On fair use, I think we have a very reasonable position under the current law. But I think AI is so different that for things like art, we'll need to think about them in different ways..." Altman:I think the conversation has been historically very caught up on training data, but it will increasingly become more about what happens at inference time, as training data becomes less valuable and what the system does accessing information in context, in real-time... what happens at inference time will become more debated, and what the new economic model is there.
Altman gave the example of an AI which was never trained on any Taylor Swift songs — but could still respond to a prompt requesting a song in her style. Altman: And then the question is, should that model, even if it were never trained on any Taylor Swift song whatsoever, be allowed to do that? And if so, how should Taylor get paid? So I think there's an opt-in, opt-out in that case, first of all — and then there's an economic model.
Altman also wondered if there's lessons in the history and economics of music sampling...
Q: What comes after it?
Altman: I don't know. I mean, that was what I was saying. It's so good, that to get beyond it, I think the bar is quite high.
Q: You've been working with Jony Ive on something, right?
Altman: We've been discussing ideas, but I don't — like, if I knew...
Altman said later he thought voice interaction "feels like a different way to use a computer."
But the conversation turned to Apple in another way. It happened in a larger conversation where Altman said OpenAI has "currently made the decision not to do music, and partly because exactly these questions of where you draw the lines..."
Altman: Even the world in which — if we went and, let's say we paid 10,000 musicians to create a bunch of music, just to make a great training set, where the music model could learn everything about song structure and what makes a good, catchy beat and everything else, and only trained on that — let's say we could still make a great music model, which maybe we could. I was posing that as a thought experiment to musicians, and they were like, "Well, I can't object to that on any principle basis at that point — and yet there's still something I don't like about it." Now, that's not a reason not to do it, um, necessarily, but it is — did you see that ad that Apple put out... of like squishing all of human creativity down into one really iPad...?
There's something about — I'm obviously hugely positive on AI — but there is something that I think is beautiful about human creativity and human artistic expression. And, you know, for an AI that just does better science, like, "Great. Bring that on." But an AI that is going to do this deeply beautiful human creative expression? I think we should figure out — it's going to happen. It's going to be a tool that will lead us to greater creative heights. But I think we should figure out how to do it in a way that preserves the spirit of what we all care about here.
What about creators whose copyrighted materials are used for training data? Altman had a ready answer — but also some predictions for the future. "On fair use, I think we have a very reasonable position under the current law. But I think AI is so different that for things like art, we'll need to think about them in different ways..." Altman:I think the conversation has been historically very caught up on training data, but it will increasingly become more about what happens at inference time, as training data becomes less valuable and what the system does accessing information in context, in real-time... what happens at inference time will become more debated, and what the new economic model is there.
Altman gave the example of an AI which was never trained on any Taylor Swift songs — but could still respond to a prompt requesting a song in her style. Altman: And then the question is, should that model, even if it were never trained on any Taylor Swift song whatsoever, be allowed to do that? And if so, how should Taylor get paid? So I think there's an opt-in, opt-out in that case, first of all — and then there's an economic model.
Altman also wondered if there's lessons in the history and economics of music sampling...
Seems like LESS debate over AI using context (Score:3, Interesting)
the conversation has been historically very caught up on training data, but it will increasingly become more about what happens at inference time, as training data becomes less valuable and what the system does accessing information in context, in real-time... what happens at inference time will become more debated,
The "conversation" has been about training data, because so much of what is used for training ends up in output, and so much of the training data has come from people who were never asked if stuff they had up could be used for training (especially art).
The "inference time" argument around AI's using context, seems vastly less debate prone to me - because that is so easily controlled, at least in terms of something like a local model making use of local app data. I would presumably have control over what apps allowed AI to see data they held, or even on what devices something would run so I could control context visible to the AI that way. It just seems so much less controversial since the inference is happening from my own data, based on my own request...
Honesty let's adjust the quote (Score:2)
The iPhone and cell phones have allowed businesses to sell subscription and pay as you go services to everyone on the planet and not just a magazine subscription or phone line to a person's home address.
AI is this decade's 'vision project' for managers of all levels, all companies and all parts to get up the corporate ladder or more to a higher paying job at the next company.
Managers need a vision to sell to others, otherwise, they are churning projects and cost cutting pennies.
Re: (Score:2)
I think you've got him confused with Sam Bankman-Fried [wikipedia.org].
Somewhat similar name. Both creepy techbros that belong in prison IMHO.
Re: (Score:1)
So it's OK to steal from people as long as you don't show it to the police?
Yes we stole this stuff and it's definitely in there, but if we don't show it to anyone it's all good.
Re: (Score:2)
They would greatly care if they play an entirely different movie that just happens to focus on the same themes.
And for those 1-in-100 other disks, which the developer didn't release but you broke into their office to steal, if you think the developers are liable for that and not you, try again.
Re: (Score:1)
The other factor is that the "inference" these models can do is really pathetic, while the training data is at least really large and has a lot of useful stuff in it.
Re: (Score:2)
What happens at interface time? Why, the data gets sent to the mothership and the AI company now has a copy of your company's source code or whatever someone asked it about. That will indeed become more relevant than the initial training data -- but don't worry, the companies will pinky swear your data is safe.
Altman (Score:2)
>But the conversation turned to Apple in another way. It happened in a larger conversation where Altman said OpenAI has "currently made the decision not to do music, and partly because exactly these questions of where you draw the lines..."
Uh-huh. Altman's ENTIRE career, and the trajectory of post-Musk OpenAI, has been a cycle of: "guys we made something SOOO powerful you can't possibly use it......Okay we'll let you use it, but only if you pay us, and it's just sooo dangerous we can't possibly release t
Re: (Score:2)
OpenAI has "currently made the decision not to do music, ..."
Just like Apple did [wikipedia.org]
Re: (Score:2)
The general approach is much older and much simpler: https://en.wikipedia.org/wiki/... [wikipedia.org]
Re: (Score:2)
OpenAI has "currently made the decision not to do music, ..."
Just like Apple did [wikipedia.org]
That was about Apple agreeing not to start a Record Label, in order to secure the Rights to sell/stream The Beatles (and other Apple-signed Artists) on the iTunes (now Apple Music) Music Store.
Re: (Score:2)
That was about Apple agreeing not to start a Record Label,
Just semantics. iTunes/Apple Music are for all intents and purposes a record label. Just because they don't distribute content on big, black, round things with grooves is a minor nit to pick.
Wrong.
The iTunes Music Store is a Music Retailer, like Tower Records.
Apple Music is a Music Streaming Service, like MTV used to be with Music Videos.
Neither are anything like a Record Label.
Re: (Score:3)
Indeed. Very, very strong Scam-indicator. Reminds me, for example, of all those claims the "free energy" scammers make. And many others. The sad thing is how many people fall for a tactic this obvious. On the other hand, it is basically a modified form of the "Big Lie" (https://en.wikipedia.org/wiki/Big_lie) approach and adapted to a cyclic, incremental use. The "Big Lie" unfortunately works time and again on a lot of people.
Sam Altman is a walking dead man (Score:2)
Mark my words: one day, one of the legions of people who will have lost their jobs to AI will off him.
AI is not even his fault. It would have happened without him. But he made himself the face of AI and he'll pay the ultimate price for it.
Re: (Score:3)
He's not going to turn out to be important enough to even remember.
Re: (Score:2)
I think so too. Except for inane statements, he really contributes nothing.
Re: (Score:2)
You know, if someone wants to take some sort of revenge against him, there's a much simpler way - unique to him.
His sister Annie is a sex worker in Hawaii who hates his guts (Sam at one point offered to buy her a house; she refused because she didn't want to be controlled by him). If a man really wants to stick it to Altman, he could literally... pardon the crass pun.... stick it to his sister.
Sam Altman's OpenAI on your mom's Satisfyer (Score:2)
The possibilities are endless when you use your imagination.
LOL, stopped reading at this... (Score:4, Insightful)
Re:LOL, stopped reading at this... (Score:4, Insightful)
Indeed. I think we need to start at things like the water well, the knife, the building, the wheel, etc. All still in use millenia later and all hugely useful. Nobody will even remember the iPhone in 100 years.
Re: (Score:2)
Re: (Score:2)
I personally think the iPhone is like the greatest piece of technology humanity has ever made.
While I certainly agree that Altman's statement about the iPhone was over-the-top hyperbolic, I also knew that he would get attacked for saying anything Positive about Apple.
Huh? "iPhone is like the greatest piece of tech.." (Score:5, Insightful)
- the integrated circuit (without which there would be no iPhone)
- internal combustion engine / steam engine
- antibiotics
- the microscope
- sewer systems
- the electrical generator / turbine
- the sail
- bronze / metal alloys
- fire
Re: (Score:3, Interesting)
Refrigerator is the disruptive technology (Score:4, Interesting)
Refrigeration changed the way people live. Before refrigeration, there was a market for exporting ice from cold regions. Ice House Street [wikipedia.org] in Hong Kong is named because that's where they kept ice imported from North America was. People used to spend a large proportion of their time preserving food so it could be kept without spoiling. Refrigeration radically changed how people live. The iPhone is nothing compared to that.
Re: (Score:2)
Re: (Score:2)
I could understand if he said smartphone instead of iphone. After all, there was nothing that the original iphone could do that my original samsung galaxy S could not. Not that I'm suggesting I agree, the list from above shows why suggesting the iphone is the greatest piece of tech is rubbish.
Although it certainly wasn't the greatest tech invention of all time, Smartphones were universally less User-Friendly before the iPhone.
If that were not true, why did we immediately see this happen?
https://appleinsider.com/artic... [appleinsider.com]
https://bgr.com/general/origin... [bgr.com]
https://www.idownloadblog.com/... [idownloadblog.com]
https://arstechnica.com/tech-p... [arstechnica.com]
Re: (Score:2)
I think there's a dance to be choreographed from historical data at some point about which particular cell phones did what, but there's no denying the iPhone was the pickup moment, where humanity went from, "What's a cellphone?" to "How do you even live without a phone on you 24/7?" To an exec? That's "greatness." It's about market penetration, not about actual value. And there's a real argument to be made that the iPhone may actually be the worst piece of tech of all time, because it also marks the moment
Now we know (Score:2)
Now we know which silicon valley insiders actually started that stupid story about the apple crusher ad.
This is all just silicon valley VC money manipulating the media for their own gain and attention.
Why didn't they ask boxers or briefs? (Score:3)
Well that's encouraging. (Score:2)
So he's giving answers in interviews, today, that are basically still catching up with thought experiments I did well over a year ago [dreamwidth.org].
And this guy makes HOW much money??