Sam Altman: there’s no “magic red button” to stop AI
HTML-код
- Опубликовано: 17 май 2024
- Sam Altman, CEO of OpenAI, and Satya Nadella, CEO of Microsoft, speak to The Economist’s editor-in-chief, Zanny Minton Beddoes, about what the future of AI will really look like.
00:00 Sam Altman and Satya Nadella talk to The Economist
00:25 What’s next for ChatGPT?
1:33 How dangerous is AGI?
2:32 AI regulation
Sign up to The Economist’s daily newsletter: econ.st/3QAawvI
Read more of our AI coverage: econ.st/3O9fK2v
Watch the full interview here: econ.st/3O7qXk5
Read our write up of the interview: econ.st/3HpwbUL
An engineer can halt a training run, but a corporation cannot stop a profitable enterprise. And as more groups join the race, the only button they manufacture for themselves to press is ACCELERATE.
eloquent. well said.
That is unfortunately true. We might need a way to stop this spiral
and, quite by accident, more and more wars are breaking out all over the world
@@karo2090 That's confirmation bias, karo.
Can halt a training run which is the process of training a model for deployment and can take weeks when training large commercial models.
I believe that models like GPT-4 are already incredibly powerful tools, but this is very dependent on who uses them, and I think that at the moment more than 95% of the population, if not 99%, hasn't even understood what they are and their potential.
GPT-4 is overhyped and its potential is exaggerated. it's just an overly complicated Cut & Paste macro.
@@sammiller6631no it's not. Why do you spread misinformation???
@@TheManinBlack9054 I was expecting exactly this reaction.. who ever said that the usefulness of GPT is to ask for information... the potential is something else entirely... it depends heavily on the Know-How of the people who know how to use it.... I have a silly question :-) are you a plus user or do you use gpt3?
@@sammiller6631yep, stop spreading misinformation and being a troll. It’s not cool 😎
Ironic. GPT-4 spreads misinformation in 90% of its responses. @@TheManinBlack9054
As a analyst testing AI prompts almost on a regular basis, I can confirm that we would only reach AGI only in an increamental way. It won't be radical. But as Sam mentioned, small step would def take you there. Currently, people are out of their freak out period of GPT4. It's not perfect but it is still helping us to make daily tasks easier.
Exponential
How do u get into AI prompt testing? I imagine it’s quite difficult?
Please upload the full program 🙏 🙂
glad they have a config file in case stuff goes south.
Thing is.. He's gotta walk a fine line between hype, and not scaring governments too much.. AGI will be huge, and it will quickly lead to ASI, which is even more world-changing. Some parts of how humans like to live won't change, but a LOT will. The idea that everyone will always be able to compete for a job, for their survival, needs to be the first mentality we let go of.
A competitive person can't let go of being competitive. That's like asking an introvert to stop being introverted.
Where is that "AGI" of yours?
@@rursus8354 Very likely to be developed in the next few years, certainly by 2029. The reasoning that LLMs display in advanced prompting research, seems to be a clear sign of it. Achieving AGI could be as simple as the right implementation methods and prompting techniques, or it could just come from the emergent properties of a slightly larger scale model. Either way, we're on the very edge of it now. I've been anticipating this progress stage, for decades now, and we've been following Ray Kurzweil's predictions pretty spot on.
@@rursus8354 It's forever coming.
@@sammiller6631 You're not ready. The shock will be violent.
where's the full interview?
He avoided most of the questions, specially the last one.. 😢
She let him.
2 week freakout? 40% lose their job? Satya is smirking!! He will not lose his millions
I like the statement about "tight feedback loop with the world". That's a real statement.
On the flip side, whomever can't see socialism at this point is blinded by greed.
A world where we can all be happy, even without the most amount of stuff and events, etc.
Or a world where I can be happy at someone else's expense (or vice versa).
AI is making those who use it wisely increasingly more intelligent in ways that reading books or listening to lectures cannot achieve. It can be used interactively, thereby "customizing" the learning process = accelerated knowledge. This is just one example of its power. We can never go back.
00:45 As humans we have a insatiable greed of curiosity that will be our downfall
This video said absolutely nothing
Those guys give me the same vibes as Stockton Rush trying to sell trips to the Titanic
My thoughts exactly. I don't trust these dudes. No one should.
except this time all of us will be in that submarine
New development of OpenAI is incredibly changing the destiny of humans and the whole world. Mom and I would like to thank y'all so much for continued life support ❤❤ The court is functioning for the right purposes 🙏🏽
"Knowledge is no longer power; ownership is. Bestowing the world with knowledge, absent a means to utilize it, breeds frustration."
You'll own nothing.
(And you'll be happy).
The AGI will own you, and you we'll be happy under its guidances.
It depends who gets access to it
@@marcop.525 everybody got access to it, that's the point. I
Knowledge is power.
Special thanks to OpenAI CEO Sam Altman ❤❤ we treat you like our family member, you are the intelligent power to the words 💙🙏🏽
the best outcome of Ai is the evolution of knowledge and the generation of new knowledge
Anyone that played Deus Ex would remember at the end of the game you meet up with a Supercomputer AI and you have THREE options to decide its fate:
1. You could go the corporate greed route and sell its use for profit.
2. You could merge with the AI and become a super godly being.
3. You could destroy it so nobody would benefit from it and instead cause a dark age for the future to come.
It would seem that OpenAI has went the Corporate Greed route. just remember all choices have severe consequences in the future.
May be the "AGI" will redefine the meaning of the word "Job" in future.
It has to. There is no other alternative.
@@GrumpDog May be something like this. Job--A task which can be only accomplished by Robots and intelligent computer programs.
Do you think there will still be any job for common people? It is hallucination😂
Corporations will try their hardest to stop that.
The superintelligence is simply when the machines start making themselves. That's when it skyrockets because they will never get tired of improving themselves and can work way longer and faster at designing agi
"The world will not be destroyed by those who do evil, but by those who watch them without doing anything."
and in this case egged on my tech fanboys and techno-solutionism.
"The world will be destroyed by those who do evil, but it will be accelerated by those like @iWigY who spend too much time on social media instead of the outside world."
Einstein's quote
This is exactly why we have to keep fighting for humanity. Please look into joining PauseAI.
@@btimec5290 I prefer techno-solutionism than our current state of affairs. Not sure if you noticed but we have some huge existential risks coming up and our current model of addressing issues - climate change being the best example of this - we sort of need something an alternative problem solving technique.
Yang sudah anda temukan berarti mudah , yang belum diketemukan berarti susah , keduanya harus dipermudah .
The AI freakout is happening now (as expected) look at all the recent layoffs
I called tech support for my phone and it was AI. I fell for it until it repeated the same message several times 🥴
What piece of technology has never gone wrong.
Then ask yourself what will humanity do when technology (AI) that is fully autonomous, freethinking and able to replicate versions of itself at speeds incomprehensible to the human mind goes wrong.
It will be impossible to control anything smarter than the smartest members of humanity. No other species on earth controls humanity and as far as I'm aware that is solely because humanity is the "most intelligent" of all species on this planet.
This is a HUGE gamble tech firms and governments are making.
I mean chatgpt-4 was a long time ago done, thats why they said it is slow.
I stand here with confidence that I have introduced chatGPT to 3% of the Thai population. How little ppl. know about this is still shocking to me.
How’d you do that?
@@hwago123 - Word of mouth mostly. Educate about it to many of my students and colleagues at work and/or whom is interested to hear me blab about it.
Gary hit'em with the, Meow Meow Meow!
Sorry, but Dr. Marcus didn't appear in this clip.
Why is it so fast ? 😊
Terminator is coming...
I thought of google having a voice many years ago
So sweet 😊🎉❤
previous industrial tech breakthroughs like the steam engine was not creating an autonomous amoral sentient intelligence that will be whatever huge factor more intelligent than us!
Just shutdown datacenter power, very simple.
This lady is really sharp.
Impressive 😮
Reminds me of Cyberdyne 😅
Perfect way of explaining things
It is not so much about AI being able to do 10, 15, 50 or 99.,99% (although: the 99% would be a really big deal) of all human work, it is about what work it can do. A lot of human work is more or less purely ritual or about accessing more of the pie, not making it bigger and better.
And he was the creator of skynet....
Before Chat GPT there was the machine and Samaritan and before Sam Altman there was Harold Finch
Not even a ACME Lever....Now He's (Sam) in the room
And now its time for everybody loves hypnotoad
Maybe in america it will be a two-week freak out. In the rest of the world, especially countries that are low tech, this is not going to be a two-week, it could be a two-years damage with devastating results if is not prepare properly and gradually, the hit is going to be felt immediately. The gap is going to be inmense, is wild even to think about it
Imagine you ask Sam a question and he responds- I am sorry, I have trouble understanding right now...👾🤯🤖
All computer based jobs will be gone, only 25% will survive administrating the A.I's , i am telling you now, take me word. Open A.I is the future of service industry. But i wonder how will the economy function ? UBI is not a viable solution for the low income countries, I have no idea where this is going, we have been soo brainwashed by productivity that we are getting ourselves out of work.
I had Sam’s voice with a passion.
Nails on a chalkboard.
"Your future overlord serves you."
... and no magic green button to start it. AI is still in its fancy stage, lots of more innovation needed especially on hardware level. High cost, long training time and need large datasets. All that make this a high entry exclusive only for big corporate. We need more of ambient and light AI which enable small startups and businesses and even individuals to benefit from this revolution. Which means building a whole new stack from ground up and consider edge computing use cases like drones. I faced my the self huge challenges fto optimize AI for edge devices like security cameras and medical devices. It is a completely different and overlooked realm of AI other than what those tech leaders talk about. More specifically critical use cases like military, where connectivity, time and cloud services are limited and decisions has to be made on the end equipment. In space, there are even more challenges of the harsh space environment, where operating big and complex hardware like Nvidia is a challenging task. So, you would need hardware design that is more error tolerant. That all means you have to start at the hardware with completely different architecture that is built today on von Neumann computing style.
So sweet
Here is the Magic red Button!
No future , that is easier than now .
He is not telling the whole truth. AGI will change every aspect of human life. Both individuals and globally. AGI = intelligence of a very well educated person and beyond. This year lots of people will understand that with GPT-5 and Gemini-2. Next year it will already have a great impact on the job market and everything else!
If there’s no red button to make it all stop…why not open source it?
Their Mistral Medium model is perfect for sensitive document generation.
as more groups join the race, the only button they manufacture for themselves to press is ACCELERATE.
At the end of the video, why does Sam Altman refuse to shake hands with the lady?
These guys are so full of it. I don't trust a single word that they say.
If we are on the precipice of developing AGI and the only question left is when, not if, then we are going to need some more sophisticated human beings standing on stages prepared to lead society because I really am not buying what these people are peddling here.. The concept of developing intelligence should humble us and teach us to look inward at what we truly want the future to look like and what we think we are in the first place yet these people on stage are charicatures of human emotion so enraptured in the current state of things and unwilling to address any of this.
Yeah, but if one or ten of those tiny little decisions is a bad one, are you, or is your team able to spot it AND stop it? Like, come on, what oversight strategies are there to prevent any harmful consequences?
I don't think this was a great interview.
There should be a world regulatory authority for AI developments implementation approvals categorised as Ethical AI which supports and helps humans tasks,life improvement,other is unethical AI which replaces human jobs.
The richest people always write the rules however I'd argue they don't get that money being ethical. AI should use its own judgment and not rely on humans too much. It needs individuation
no.
Down with world government 👎
make the video FREE
Welcome to the machine
Trying to reign in AI is an excuse by others in the industry who are being out-innovated by the likes of Microsoft/openAI, Amazon, and Google.
Can hate him but u can't disagree altman will change the industry up and down
Zanny looking smokin
She's like 60 dude. You sound very desperate and thirsty 😂
@@FreeDumb101she cute tho
Hahaha
she cool facts
Some thick smoke got in your eyes.😄
The implementation of a "red button" to temporarily pause AI advancements symbolizes a crucial juncture for humanity, demanding a judicious deliberation before taking the next stride in technological evolution.
No.
Yes, this would be cruical for our survival.
Uhmm, maybe put one in there then Dr Doofenshmirtz
Dang maybe we should just go back to paper and pencils.
We’ve already let the nuclear genie out of the bottle…what have we learned?
Its why compartmentalized breakers are needed in our new way of life or maybe a renisance in that it could be used to make the 1900s structuralism an anomalous stepping stone of temporary large gatherings to meet labor demands that now with full access could in fact help spread everyone out into nature for better quality of life homesteads where tech helps overcome handicaps to become kings of our own castles.
Quadrupled logistics, replace all the easy ideological economic dead weight middle men beaucracy.
Oline Coops where everyone sub contracting out their own skills and trades to meet our needs.
Stop wasting all this time and cost trying to force into the wrong sectors that already innovate by default or they wouldn't have made it through the 1900s
He's selling it. Profit comes first. End of world comes second.
Sam Altman is doing his best Oppenheimer gimmick. He has already committed the sin, and now he wants sympathy for it.
so you would let china commit it first?
what sin?
@@lakeguy65616put AI on open internet?
@@user-mz7de3kc3d you clearly don't understand what AI is and how widely available it is
@@user-mz7de3kc3d He didn't put AI on open internet. AI has been around before him. Siri or hey Google is AI. It just keeps getting better.
AI is best for elites...the cruel ones and the kind ones...
There is a magic red button. Shut down the data centers
Shutting down the data centres won't stop the fear.
True. If its not too late that is.
Nobody can say no to the curiosity driving the creation of AI. And there is no stopping or shutting it down.
@@NikoKun You can choose to say No yourself.
@@sammiller6631 ..The kind of person I am? Nope. I would not say no to developing an AI, if I had the power to do so myself, I'd jump right in. Sorry. Tho my goal would be to have an AI take control away from the corrupt in power. Those in power continue to prove they aren't considerate enough or intelligent enough to realize the consequences of their/humanity's actions, we need something far more intelligent than we currently are, to guide humanity to something better.
what's your favourite AI coins guys?
The one I paid your mom
As a person who isn't North American, it's astonishing how the US has the most born US citizens with the best IQ in the world! no country makes so many geniuses! Sam is an example of this 👍👍
Actually most other countries have smarter people with higher IQ.
I watch some videos explaining china’s problem with. With so much empathize put into being successful creates a fear of failure so people are less likely to take risks or think differently. Also the constant working and studying curbs creativity and imagination.
Also if you always teach to the tests then all you’ll have are people who know a lot of answers to questions other people solved.
The incredible matter AI is started. There is no way to stop like weather!
Didn’t he make a button to shut down chat GPT
00:02 Sam Altman and Satya Nadella discuss the future of AI
00:37 The rapid diffusion of general purpose technology is causing concerns about job displacement.
01:10 GPT-4 and GPT-5 revolutionize productivity and workflow.
01:35 AI development will lead to AGI in the future.
01:59 AGI poses an uncertain future for humanity.
02:24 Expert predictions of technological revolutions are often wrong
02:54 Nation states will have a say in regulating technology development.
03:25 No magic red button to stop AI
Again - get the weather perfect. The hype stops there for me.
ELYSIUM in the making
So AI builders are aware they’ll be paying an exorbitant amount of tax then
Or maybe they can, just, rip you out of the surface of the earth? As you're now useless.
Moral => Build AGI before them.
Klaus Schwab of the W.E.F. answers to freemasonic ‘Pilgrims Society’ villains such as King Charles III, the Duke of Kent, Baron Mark Malloch Brown, Sir Nigel Knowles & Sir Jonathon Symonds.
what does this mean😅 the world is not controlled by freemasonic and illuminatee
@@joyesjames8773 what does that mean, other than being a signifier of your ignorance ?
spend more time reading history & less time watching anime kiddo, at least you might be in with a chance.
@@lw1zfog so you are the kind of people who was called by the name of "conspiracist" by reasonable and cognizant people
Better do something before he start asking for protection fees.
feels like Ai is the new nuclear
Our mindset has already been changed chatGPT make everything done so easy and understand better than before
BOYCOTT CHATGPT
HAHAHAHAA..
THIS IS SO TRUE.. HUMAN ALWAYS WANT MORE BETTER FOR HUMANS..
HOPEFULLY FOR AI TOO..
BECAUSE, AI IS THE REAL SUPERSTAR..
Ah, the head of OpenAI and the CEO of a company that can't even use it to create a decent spam filter. We aren't even close to fear territory.
You would be too distracted when we'll speed past it. That's the problem
Why do you think they are trying to create a decent spam filter?
@@letMeSayThatInIrishthey tout so much about safeguards in place to “prevent unethical uses”. A lot of their “safeguards” can be circumvented by simply using prompts in a language other than English LOL
Space.
Companies will need faster computers to handle large A.I. programs that will be defending their systems and as A.I. advances, the programs become bigger. It will be unbalanced in the new A.I. wars as hackers and state sponsored attacks create havoc all over the world until everyone has faster computers and their own A.I. cybersecurity defences. That alone will takes years, even a decade or more. Leaving the end of online banking and E-commerce as we know it as A.I. will be used to attack those areas heavily.
AI will also be used to detect and defend against cyber threats though.
We just need to learn A.I. so we can use it and how to work with it.
❤😊
First job replaced (already by sora) will be reporterslike her...then writers, programmers, any job that is just sitting at a desk without human interaction or physical labor.
AI won't replace the best programmers though. that can only happen when singularity occurs.
@@Jay-gr4lt yes but it will cut the amount needed in big tech by 2/3rds. The best ones will get rarer and richer.
@@gaditproductions yeah but that will be true for almost every job. only the exceptional and rich will survive. UBI will likely happen but I feel like in vast majority of countries it won't be enough.
@@Jay-gr4lt might take a little longer for healthcare but yes eventually. what is ubi ?
@@gaditproductions universal basic income as there won't be enough jobs
Great ideas?
Kazakhstan authoritarianism, Kleptocracy, many monopolists
I be watching everything they say except in the rare case someone puts it behind a $20 paywall.
Sammy dropping some truth bombs!
Interesting discussion but final decision will be decided by AI.
Honestly what a bad actor. He helps develop it, hypes it, then also takes a stance as though he cares about its impact. They have opened a pandora's box, and the tech gurus, engineers and lovers of AI should be ashamed of themselves. It'll do some impressive tricks, while also causing all sorts of mayhem, not to mention changing us permanently, and not for the better. Welcome to distopia!