@@slowmissouri205 as an AI engineer I can tell you that we don't know much. We know the math but the output is emergent. Like music is emergent from a record. Looking at the grooves on the vinyl is understandable but until you experience music, you won't ever know what the vinyl is truly capable of. Imagine never hearing music and then bam. Your life changes in ways you could never imagine.. . hope that scares you. That's why I changed Fields. I'm scared..
what a good interviewer, not just asking easy void questions but actually tackling on the problems and the changes that this will bring to the world, and sharing his own perspective with the interviewed, this is a nice interview
Current AIs are tools, but the idea that AI in general is 'just a tool' is ludicrous. You don't see people scrambling to figure out how to align other tools with our interests so they won't take over control of the Earth from us when they become smarter than we are. There are good reasons to believe AIs will develop complex world models and goals of their own. It seems likely they will eventually cross the threshold of 'personhood'. We are setting ourselves up to have a hard time acknowledging it because, not only will we be relying on their labor, but it will also mean we have created a slave race.
I think the idea that that AI will be just a tool is myopic. We are at the early stages at the moment, but agents that can plan and do complex tasks by themselves are already being developed. The one I saw demo of today (Devin) was quite impressive. It could even problem solve and correct some mistakes.
I wonder if we will see fast progress and then stall. We may hit the bounds of what is easy and people may overtrust AI or people may be overly vested in the idea that AI will solve everything. That said, I think it is more likely that we won't stall out.
@@slowmissouri205 They already see a smooth function for the scaling law. More parameters and more data alone will keep scaling intelligence. Unless we believe we are at or near the pinnacle of possible intelligence, ourselves, AI will be superior in every domain in 2-5 years. That's not even factoring in architecture or algorithmic improvements. We will have a good feel after GPT-5 and GPT-6 how fast this is going to go.
is 'devin' really ai itself, or just a bunch of cleverly stitched together api calls, langchain, and a UI to tie it all together? They don't specify the 'how' or 'what' of it at all - at least that I can find. This has been a notoriously difficult bar to surpass for some time now, for some reason I just am not on board with it yet.
Similar like in banking, in AI we see smart people work with good intentions within an environment that has poor governance and unintentionally sets us up for major larger disaster.
I thought I was the only one who noticed. This interview makes me question his intentions. At the end of the day, he seems to be more concerned about profits than the ethical implications of OpenAI's work.
“Everyone gets to choose how they spend the extra time gained from productivity increase” - wrong. People are working the same time or more with wages that have been stagnant for decades while corporate profits keep increasing “You can just sit down in the plane and reset your brain on a trip to japan” - lol ever been in economy?
There's nothing more unnerving than watching techbros who are designing AI talk about the future of AI. The fact that this guy didn't give an unequivocal yes when asked if artists deserve compensation for their work when AI samples it signals the end of good original art as AI will just cannibalize every good piece of art into uninspiring distorted copies
Imo it's a democratization and revolution of art and the creative process. As an artist I'm inspired by all kinds of things, including ideas from the internet for which no-one pays fees to creators. Also for a long time I was planning to create paper art, and watching just the one Sora video of a paper underwater world, I got an intense creative boost from it in my mind. And the Sora toy gnome working in his zen garden, really magical! And AI-art that can be 3D printed and then manually worked on further. The entire economy will change, affecting everybody's security of income. Governements and global financial institutions should work together to regulate this transition.
An enlightening discussion shedding light on the fascinating interplay between AI and humanity. Peter Deng's insights into the role of AI as a tool to enhance our humanity, rather than replace it, are truly thought-provoking. As we navigate the ever-evolving landscape of technology, his call to stay curious and embrace the opportunities AI presents resonates deeply. A must-watch for anyone intrigued by the future of AI and its impact on our lives.
"slowly deploying"!? what is this snake-oil salesman talking about, they have released adjustable foundation models for image generation since day one, knowing full well Loras could be implemented. Zero safety. A tool? just like dynamite is a" a tool"
oh it will be horrible for regular people. your near future car: I am sorry, due to licensing agreements i cannot take you to Publix, rerouting to Whole Foods Market." A corporate tyranny beyond this current corrupt horrible economy.
In order to instill other culture's/ individual's values requires hiring people from different fields of study as well as from other cultures. It doesn't seem like companies are doing the work to make this happen. If companies are continuing to hire very left brain oriented people only, then they are setting themselves up to make the same mistakes. Also talking about AI as if that AI wasn't built and coded by people allows my previous point to be easily glossed over and some of the talking points in this video become just another way of saying in a generic way "we are building a better world." Another talking point which shows that something is being missed or that past mistakes are not being learned from is the idea that AI will free up time, which has always been the goal with tech, but it hasn't proven to be true in many areas. Also, without a very awake and aware design and implementation as well as awareness of the greed which will attempt to hijack the new tech, we will continue to make the same or even worse mistakes.
it's not the engineers you silly goose, it's the internet, the internet's data as a collection - and the output you see, should indicate to you that the majority of thought is not the same 'far right' thread of your own (apparently). get over it, intelligence is left. elon is a businessman, piping his words on random youtube comment sections is doing his bidding btw, his words are nothing more than pandering to the right. If you think about it, no right-winger just 2 years ago would buy a tesla (*such a left thing going electric) and now there's cybertruck... he's appealed to 'that' audience, now they line up for them AND him... problem solved.
I realize that it is SX, but do we really need such a liberal host asking BS questions over and over to try and get a different answer? There is a name for that....
Great talk! Love the way Peter Deng thinks about AI. Notes only 18mins in: - ChatGPT as a "thought partner" augments our thinking naturally - AI enables us to get into "higher level" thinking and doing
guy was self absorbed, not charismatic, just asked none stop questions and it was more question answer than a discussion. He just loved hearing his own voice.
find it weird the host talks about AI in a lose term. why not highlight the different stages of AI, (AGI , SGI ) where we are right now, what's to come and how long. Both the guest and host at time are talking about different stages of AI. What happens when we get to true AGI, this is what the host alludes to but doesn't make the distinction to current state of AI. what happens to humans when we have AI that surpass us.
I wonder when someone will actually ask tough questions and follow through.
@@slowmissouri205 as an AI engineer I can tell you that we don't know much. We know the math but the output is emergent. Like music is emergent from a record. Looking at the grooves on the vinyl is understandable but until you experience music, you won't ever know what the vinyl is truly capable of. Imagine never hearing music and then bam. Your life changes in ways you could never imagine.. . hope that scares you. That's why I changed Fields. I'm scared..
This guy is a great interviewer. I don't remember seeing someone this good. Well done and good job!
Where’s Ilya?
Probably around
Ayyyy, I dunno
Locked in OpenAI’s basement?
he's feeling the agi, they connected him in the basement to the mashine :D
He’s in the attic.
what a good interviewer, not just asking easy void questions but actually tackling on the problems and the changes that this will bring to the world, and sharing his own perspective with the interviewed, this is a nice interview
This feels artificially upbeat and scripted.
welcome to tech talks, geez..
The job title checks out.
I'm an artist and I don't know any artist who wants AI to give feedback on their work
Current AIs are tools, but the idea that AI in general is 'just a tool' is ludicrous. You don't see people scrambling to figure out how to align other tools with our interests so they won't take over control of the Earth from us when they become smarter than we are. There are good reasons to believe AIs will develop complex world models and goals of their own. It seems likely they will eventually cross the threshold of 'personhood'. We are setting ourselves up to have a hard time acknowledging it because, not only will we be relying on their labor, but it will also mean we have created a slave race.
I think the idea that that AI will be just a tool is myopic. We are at the early stages at the moment, but agents that can plan and do complex tasks by themselves are already being developed. The one I saw demo of today (Devin) was quite impressive. It could even problem solve and correct some mistakes.
Wait till AI like Devin and others learn to use chat gpt and “lesser” AI to complete tasks
I wonder if we will see fast progress and then stall. We may hit the bounds of what is easy and people may overtrust AI or people may be overly vested in the idea that AI will solve everything. That said, I think it is more likely that we won't stall out.
@@slowmissouri205 They already see a smooth function for the scaling law. More parameters and more data alone will keep scaling intelligence. Unless we believe we are at or near the pinnacle of possible intelligence, ourselves, AI will be superior in every domain in 2-5 years. That's not even factoring in architecture or algorithmic improvements. We will have a good feel after GPT-5 and GPT-6 how fast this is going to go.
is 'devin' really ai itself, or just a bunch of cleverly stitched together api calls, langchain, and a UI to tie it all together? They don't specify the 'how' or 'what' of it at all - at least that I can find. This has been a notoriously difficult bar to surpass for some time now, for some reason I just am not on board with it yet.
I noticed that, too - Deng was very careful to talk about AI as a helper.
The talk was great advertisement, almost as good as Tuckers Putin interview.
well, propaganda is very effective
Similar like in banking, in AI we see smart people work with good intentions within an environment that has poor governance and unintentionally sets us up for major larger disaster.
Rule of thumb: if someone points with their thumbs while they talk they are hiding something
Hmmn...okay Mr scientific truth 🤦
Yes, the end of their index finger....
Another rule of thumb: people that are thumb-watchers are jealous of a person that points with their thumbs while talking
Whoever is in charge of audio should be the first to be replaced by AI. Can barely hear them on max volume
"Portuguese and Spanish language translations for SXSW 2024" - LOVE FROM BRAZIL
Has trouble just answering questions. He’s like a politician.
I thought I was the only one who noticed. This interview makes me question his intentions. At the end of the day, he seems to be more concerned about profits than the ethical implications of OpenAI's work.
Yep, that should tell us everything we need to know.
all the good intentions of social media did not go well. this will not either. greed and profit are terrible optimizers for common folk.
“Everyone gets to choose how they spend the extra time gained from productivity increase” - wrong. People are working the same time or more with wages that have been stagnant for decades while corporate profits keep increasing
“You can just sit down in the plane and reset your brain on a trip to japan” - lol ever been in economy?
1:58 second time he's touched upon it, which leaves me the impression it might be his only talking point: AI makes us more human
3:05 AI lets us go deeper?
4:39 humans as focusers: we always were
4:53 what is this shift?
"AI is just a tool"....AI in 2029.."Hold my beer"
Why is this guy too scared to speak about anything interesting?
There's nothing more unnerving than watching techbros who are designing AI talk about the future of AI. The fact that this guy didn't give an unequivocal yes when asked if artists deserve compensation for their work when AI samples it signals the end of good original art as AI will just cannibalize every good piece of art into uninspiring distorted copies
Imo it's a democratization and revolution of art and the creative process. As an artist I'm inspired by all kinds of things, including ideas from the internet for which no-one pays fees to creators. Also for a long time I was planning to create paper art, and watching just the one Sora video of a paper underwater world, I got an intense creative boost from it in my mind. And the Sora toy gnome working in his zen garden, really magical! And AI-art that can be 3D printed and then manually worked on further. The entire economy will change, affecting everybody's security of income. Governements and global financial institutions should work together to regulate this transition.
An enlightening discussion shedding light on the fascinating interplay between AI and humanity. Peter Deng's insights into the role of AI as a tool to enhance our humanity, rather than replace it, are truly thought-provoking. As we navigate the ever-evolving landscape of technology, his call to stay curious and embrace the opportunities AI presents resonates deeply. A must-watch for anyone intrigued by the future of AI and its impact on our lives.
His multitasking is amazing. I was impressed by his note taking abilities.
"slowly deploying"!? what is this snake-oil salesman talking about, they have released adjustable foundation models for image generation since day one, knowing full well Loras could be implemented. Zero safety.
A tool? just like dynamite is a" a tool"
Devins AI made this interview feel old in less than 24 hours. AI made 2025 will eat almost all AI developed 2024 to breakfast.
oh it will be horrible for regular people. your near future car: I am sorry, due to licensing agreements i cannot take you to Publix, rerouting to Whole Foods Market." A corporate tyranny beyond this current corrupt horrible economy.
I have to say, good questions but common answers...
In order to instill other culture's/ individual's values requires hiring people from different fields of study as well as from other cultures. It doesn't seem like companies are doing the work to make this happen. If companies are continuing to hire very left brain oriented people only, then they are setting themselves up to make the same mistakes. Also talking about AI as if that AI wasn't built and coded by people allows my previous point to be easily glossed over and some of the talking points in this video become just another way of saying in a generic way "we are building a better world." Another talking point which shows that something is being missed or that past mistakes are not being learned from is the idea that AI will free up time, which has always been the goal with tech, but it hasn't proven to be true in many areas. Also, without a very awake and aware design and implementation as well as awareness of the greed which will attempt to hijack the new tech, we will continue to make the same or even worse mistakes.
it's not the engineers you silly goose, it's the internet, the internet's data as a collection - and the output you see, should indicate to you that the majority of thought is not the same 'far right' thread of your own (apparently). get over it, intelligence is left.
elon is a businessman, piping his words on random youtube comment sections is doing his bidding btw, his words are nothing more than pandering to the right. If you think about it, no right-winger just 2 years ago would buy a tesla (*such a left thing going electric) and now there's cybertruck... he's appealed to 'that' audience, now they line up for them AND him... problem solved.
Grok goes open source and the world changes for open AI
48:40 when have people ever voted against a company with their wallet and made a difference? Is this dude asleep?
.... just wait untill the Gamedev of existence reboots the game ...
meaningful conversation 👏🏻
I realize that it is SX, but do we really need such a liberal host asking BS questions over and over to try and get a different answer? There is a name for that....
He isnt an really a Illustrator by using A.I. closer to a supervisor.
Where's my dad?
Right here, son.
@@r34ct4 Daddy! 🥺
Great talk!
Love the way Peter Deng thinks about AI.
Notes only 18mins in:
- ChatGPT as a "thought partner" augments our thinking naturally
- AI enables us to get into "higher level" thinking and doing
I feel this! He has some nice thoughts of perspective into the topic of AI and the upcoming possibilities.
this guy not being sure if artists should get paid
They should get paid according to the value they create for societey as determined by the free market.
Art is worst thing happened to Humanity
Why are you so sure artists should be paid?
@@NotRUclips-cp3qg What the fuck are you doing here then?
@@Isaacmellojr Why should carpenters be paid? Fucking do the work for free, fucker! I deserve it.
Host is projecting his agenda way too much. Let the guest talk…
guy was self absorbed, not charismatic, just asked none stop questions and it was more question answer than a discussion. He just loved hearing his own voice.
Great conversation! a lot of inspiration。
find it weird the host talks about AI in a lose term. why not highlight the different stages of AI, (AGI , SGI ) where we are right now, what's to come and how long. Both the guest and host at time are talking about different stages of AI. What happens when we get to true AGI, this is what the host alludes to but doesn't make the distinction to current state of AI.
what happens to humans when we have AI that surpass us.
Super Litty
We can always just pay teams of artists making training data for AI agents. this would fix the issues.
only using that training data tho.
☺️☺️☺️☺️☺️
Very pollianish
Boring. I was looking for death and dismay 😅