ฝัง
- เผยแพร่เมื่อ 31 พ.ค. 2024
- A leading expert in artificial intelligence warns that the race to develop more sophisticated models is outpacing our ability to regulate the technology. Critics say his warnings overhype the dangers of new AI models like GPT. But MIT professor Max Tegmark says private companies risk leading the world into dangerous territory without guardrails on their work. His Institute of Life issued a letter signed by tech luminaries like Elon Musk warning Silicon Valley to immediately stop work on AI for six months to unite on a safe way forward. Without that, Tegmark says, the consequences could be devastating for humanity.
#ai #chatgpt #siliconvalley
Subscribe: th-cam.com/users/deutsche...
For more news go to: www.dw.com/en/
Follow DW on social media:
►Facebook: / deutschewellenews
►Twitter: / dwnews
►Instagram: / dwnews
►Twitch: / dwnews_hangout
Für Videos in deutscher Sprache besuchen Sie: / dwdeutsch
Sadly, there is little to stop people when money and greed is involved.
We do,its called a nash equilibrium.
It will be a self-solving scenario.
Thats right, that is why I think we are doomed. It had to happened at some point...They were many civilizations before us, and they will be many after us.Will have fun with AI until we won't anymore.
Do we really need to be reminded that in a capitalist world, every decision becomes just another 'business' decision?
@@youtuber5305 maybe you don’t need to be reminded. Neither do I, but others might. But it doesn’t matter anyway bc not even a notable fraction of the country’s population will see this comment
This topic is extremely under-reported. This needs to be a focus of the whole world right now, because the world will change forever very soon. And nobody expects it yet.
"nobody" except you and the intellectual elite of course
absolutely agree
I don't get why all Scifi show AGI negatively, AGI is needed to fix poverty & early death. We should have an AGI program build a replicator that can be used on the Sahara to collect massive amounts of solar energy, its replication will require limited resources like Lithium or Phosphorous or chips or whatever that would limit its ability to replicate forever but 1 trillion solar units (1 2 4 8 16 32 64 128 256 256... 40 generatinos of doubling) would help out the poor a lot... Replicators can also be used to build houses and infrastructure.
@@aoeu256 we dont know what AGI will be able to do. but we can build solar cells ourselves. We dont need AGI for that. It will be able to do many things, but uncontrolled replication would be a great danger. Of course if everything works out we can have utopia. Though caution is needed. Nothing good comes from giving someone else responsibility for everything. People need to stay accountable for themselves.
@@aoeu256 the militaries of the world are racing to develop it for autonomous weapons. Also the scenario where ai decides humanity should be extinct is real.
AI doesn't need to be self-aware to be dangerous. A non self-aware AI, given a bad directive, by accident or intention, would be catastrophic
Between the current dystopian oligarchy that has been destroying the world for centuries and a dystopian but rational AI I will always choose the AI.
The worst hypothetical scenario for the future is still better than the current one.
The paperclip maker.
One goal, everything must become paperclips.
You're exactly right.
I am an engineer (aerospace) and what we (as in society) call AI is NOT something that thinks. They are basically giant numerical models that use statistical analysis to set their parameters so they can MIMIC what a human might say or do. In certain types of data analysis where the task is well defined and repetitive like facial recognition and fingerprint analysis they can do it at incredible speed.
What I hate most of all are the proponents of AI dramatically overstating its capabilities while underplaying its threat to society. The immediate threat isn't a Skynet _"nuke us all"_ scenario, its mass unemployment because there are clowns who believe the AI will just do the job better, faster and cheaper. that might be Ok in a few places but when corporations suddenly dump massive slabs of humanity onto the streets its a problem.
All of our modern capitalist economies operate on *SPENDING.* To spend people have to earn money first. Even with credit people must still earn. If a society is NOT earning enough money it CAN'T spend enough to keep the system going. Its a fairly simple concept and most of the worlds economists and journalists are willfully ignorant that their jobs will be among the first to go. Who will need an economist when the AI can do market analysis faster and more reliably. Who will need expensive news anchors, analysts, weather presenters, camera operators, sound people, lighting people and producers when the AI can do it all in a simulation and pump it out.
🤯🤯🤷♂🤷♂
@@tonywilson4713 The immediate threat is your grandma getting a call from you that you're in a bind and need 5k.
@@how2pick4name Yeah right now at this moment but wait until grandma's phone answers with a metallic twanged voice saying she's not available.
Good stuff DW! I have no idea what you're like for other news, but kudos for putting this one on
They're like this in all of their programming. Their documentaries are particularly good.
Maybe the real problem isn't developing AI, maybe the original mistake was developing corporations driven by profit.
Indeed
well spotted! these poor people so called scientists not realizing they are used just to serve people in power with only purpose of gaining control of the mases! a scientist never asks where does the money come from for their labs and for their research and to what purpose... and all of a sudden they are worried where their inventions might lead to...this is what I call social ignorance!
Definately !!! We need a REAL, NEW, Modus Operandi, to do-all-things, w/o fricking-ourselves, with Absolutely, Pure Stupidity. 💔
@ muzehack . See schmachtenberger interview with Liv, which I think broadens the discussion substantially. Regards
@@ninaromm5491 literally most companies don't want more than 100 employees and just care about More and more profits
Love the simple ways Max explained things. Terrific guest.
@@mikegoldstein7112 great argument, so elaborate and informed (yes sarcasm...yes all of it!)
misleading ways where oersimplification is utilized for the political and economical needs of his very rich benefactors, classification algorithms have nothing to do with neural networks or trasnsformers generative models
Lies again? Special Forces Special Agent
He's one of the people Ian Duty sang about in his song 'There ain't half been some clever bastards'.
i dont think he was terrific
Thank you for this great interview and hopefully it will reach large numbers of people!
This is a really insightful chat. I think this must reach more people cause if it doesn't get to people now, it would be really too late
As an American, myself, I really need to remember to try to watch more DW News interviews. The hosts always seem so intelligent and, more importantly, not bought out by corporate interests.
Exactly, it's unfortunately so refreshing to see
German news are funded directly by the tax payer and are non profit. Makes sense doesnt it
Europe is where its at.
Please STOP it, solution/action, not just talking like most 4legs did. No intelligence, development at all from 4legs --- the TRUTH is that 4legs never accept it so to do BAD, worse, and worst until today. Are you fine in muddle under water? Who you are? 4legs or human?
I have to say that the attitude of these high tech companies is alarming since they are very aware and understand the risks better than anyone yet cavalierly push on and government regulators are no where in the equation. This is very reminiscent of the cloning technology race a few decades ago where governments were slow to react, but now the potential threat to humanity is exponentially greater, especially considering how fast robotics and drone technology is developing. We really seem to be fulfilling the plot of every science fiction horror film ever made.
Well said Pak.... considering that countries are still at war as we speak....that Syria..Yemen..Ukraine..Russia... USA ..you name it...can't even get along..young innocent men are killed, for what? Innocent civilians, CHILDREN are killed, in the name of what ? Greed? And we are going to try to get this right ? Good luck humanity !
Because the upside potential is too great. Artificial intelligence might cure cancer, reverse aging etc. Regulation will never do anything good.
And noone is focused on how AI could bring George Orwell's 1984 nightmare of a true "big brother" to life. Except when we talked about China who already has the whole population under surveillance and rates their actions. They in fact have set up a whole population control model using algorithms. Their AIs are probably already working in service of as the Chinese gpvt declared, the socialist model.
Dr. joe Dispenza says: “competition in all companies is old way of thinking…. Competition separates people from each other, we should inspire each other with informations and work as one community, one organism, helping each other, love each other etc.
Man wait until you find out what the fossil fuel corporations have been doing for decades.
the biggest reason companies want regulations is to only block it for normal users and only available to commercial firms. That's only to avoid competition from normal people. the older version V3 was far better than V4.
V4 is intentionally providing wrong answer for some reason V3 didnot do it.
Such a good speaker, explained complicated matters in layman's terms .
Great interview! and much needed exposure to a bigger public! Thanks DW!
I actually understand it now 😮 I've watched alot of AI needs to be put on hold but understand the danger now.
Max Tegmark is one of my favorite human beings. Always entertaining and insightful. I can't even tell you how many times this person has blown my mind with his ideas
Couldn't agree more.
He looks AI generated to me
LOL
how ironic that you said that.
@@dominicdudebromtl9380 4.30am 8988>zizw002
He's petty shallow
@Seek The Truth pay attention please, we're talking about how WE create our own god. God didn't create us, we created IT.
If there is money involved, the US will fight against regulation until the last politician is paid. Sad, but true.
Please everyone--share this on social media!!! Immediately, as widely as you can.
he is totally right in bringing the "Don't look Up" example, sad how humans always fall for the same trap.
Bad analogy that was used in the video in my opinion in the movie you could clearly see the danger, where in this case it’s mostly experts speculative analysis of what AI might be in the future which we know is hard to predict.
@@CalvinR90 it is for him that IS an expert. For me the problem is quite obvious also and I am not an AI expert either.
ukraine, taiwan, global recession, potential nuclear war... all these things are damn scary, but i feel like the current AI race isn't getting the attention it deserves. as that, thanks for the video. we shoud really take this seriosly.
Why? There's no God, no afterlife, let's just do whatever and die.
With all of those things considered, I'm just as terrified about human intelligence and its potential for evil as I am for AI. I can't see how one is worse than the other, at least for now.
@@MattExzy it's because ai will be like god
@First Last oof underrated concept.
@@nesseihtgnay9419 You don't know that so why exhibit stupidity.
Thank you DW for covering this important topic considering its deep long ranging effect.
"We can build AI so beyond humans that it's like comparing humans to insects." Frightening.
This guy is very compelling, without being wierd about any conspiracies, just logic
This guy is great, I was totally enthralled to hear about a subject I know very little about. I think this was one of the best and most interesting interviews I've heard in a while. Excellent job from both sides!
Lex Fridman interviewed him on his podcast. It's really interesting interview! However it's almost 3 hours long. I listened to it while working so went by pretty fast.
his book Life 3.0 is great
Yeah tegmark is great!
When the man said other why's there might not be any people on the world at all. The world is better of with out people
Yudkowsky has more of value to say on the subject.
Great content, very spot-on questions
What could possibly be more important than this interview?
The Corporatist fixation on profit and growth is my largest worry in terms of our civilization enduring through this imminent technological development. My second and third greatest worries are that our policy makers are not going to take this seriously enough and that an emergent A.I. will blow past whatever guardrails we'll ultimately try to put in place to safely guide its arrival. The reason I fear Corporate influence on this development is that their cancerous fixation on growth metrics reminds me of the Paperclip Maximizer scenario - but with the legalistic Corporations acting as a weak form of Maximizer. In essence, humanity is not the only player at the table, as corporates are also playing, but their 'win' condition is different than ours.
You are on spot sir
Iv never yelled doom, never thought I would in my lifetime. But tbh I think I'm seriously at the point my faith in humanity tells me money will win out like always in this world anymore and especially this country (usa) so I honestly feel it's getting to the point it's gonna be 50/50 that agi is the last big thing humans create.
The upshot is that AI will destroy capitalism....and all other systems that depend on humans.
We never learn... we always ask if we 'can' but never ask if we 'should'.
Plenty of people are asking the should question, but the difficult aspect is that even if 99% of humanity is against it, there's gonna be that 1% that sees personal profit or has a different rationality that will refuse to abide the rules of the 99%. The cat is out of the bag so to speak, not every person on this planet is thinking about the greater good, nevermind those that think they are but are just incorrect about what the greater good actually is.
We should
someone watched Jurassic Park.
@@ghostwolfx well I guess the problem is that the people who ‘can’ aren’t asking that ‘should questions, whereas we who do ask the ‘should’ question can’t act.
‘Should’ is the worst word in the English language. Has no right to exist.
He has the most endearing, disarming smile!! Instantly smitten. I know my comment is not at all relevant to the topic, lol!
I like his whispy bangs
This was an outstanding interview. Thank you. Please keep up the great work.
Being in the industry and surrounded by good people with noble aims and objectives for this tech, it's easy to assume things will just be OK. The simple truth is that we can't just assume that the intelligences we're creating will share our perspective once Artificial General Intelligence (AGI) comes online. We need guide rails and extensive safety systems in place at levels we humans haven't yet even contemplated. Caution is well advised.
A LOT of analog off switches(take down a city, a region or a nation with one button), decoupling from the internet of military facilities and material, of research and of production capabilities, A strict regulation if not ban on functions that would lead to self improving systems. Maybe don't use automated systems to establish loonar or mars bases.
Counter meassures like EMP(while a real effect i don't think this is yet a reality outside of movies) and radio interference.
Have the current AIs check for sercurity vulnabilities. Expert systems might be ok, but a system that can do it all while being able to predict every move of us might be a bridge too far. Therefor the data sets these LLMs are trained on (i think these are called tokens) should not expand further and further and definetly may not include human behaviour(Elon Musk just bought thousands of GPU to create twitter AGI(training set tweets).
many more things i am too dumb to even imagine
well there is one more thing which maybe naiv, but still. In one vision of the future that i would prefer, at some time there would be sentient intelligent beings sharing our values with empathy, agreeing to be governed by the same rules we govern ourselves with and are basicly free as everybody aslong they do not cause anyone harm. This might be the solution to the alignment problem, it might but also be the cause for conflict if such sentient AGI does not accept the rules we set and is not willing to cooperate with us to maybe create a better set of rules we then also would agree to. Still i think this could be a away through all of this, but till then there is always the zombi AGI variant , intelligence without sentients or empathy just endlessly creating stamps out of us.
I agree
Technology in itself is not harmful. It's the owner's intention that causes much harm. It's the greed of power that brings havoc over the common people.
@@lopamudraray4571 agree, to me it's more dangerous what governments and military could do with AI. Especially the US
One major problem is that there is no “our perspective.” Just look at the vicious political strife just within Western liberal democracies.
The other day I saw some proposed AI guide rails by Dave Shapiro and it wasn’t very long before I found myself disagreeing with his perspective.
Max is great! One of my favorite researchers. However, I disagree with him on the direction of the open letter. I think it could have been written in a much more purposeful, and cogent way. Providing more substantive courses-of-action would have been useful. As a fellow machine learning researcher, I think we need not only be realistic about historic ramifications of this technology, but also be realistic about the probability of joint efforts to mitigate the probable negative outs come of nefarious human-centric development and deployment of an intelligent agent. Historically, I am much more concerned with the human-element than I am of the technology.
For me it all gets down to one point. WE DON'T HAVE TIME TO SLOW DOWN!
Climate change is going to destroy human existence as we know it. And AI might be our last chance. We humans alone will never be able to stop it by our selves.
And of course if we look at what GPT-4 already is capable of in terms of theory of mind tasks and how to take the view point of the human interacting with "it", there is a clear path to empathy there. And with empathy comes the ability to look out for common ground.
I guess you know most of the papers i am talking about here, like the one from Microsoft Research, so i won't provide sources here.
So am I. you are correct for implying that human criminality getting hold of machine intelligence is a bigger danger than a g i running amok.
It is painful to watch these alarmists aggrandizing themselves, as if their mission in life is to protect the rest of us from megalomaniacal AI-Software Bots, seeking to enslave mankind.
On the one hand, I highly respect the advances which have been made creating 'SSRS', "Software Simulated Reasoning Systems". But any reasonably experienced software developer understands this harsh reality: its all clockwork. There is nothing mysterious about it. Just because we can speed it up using large neural networks, doesn't mean it has any similarity whatsoever with the workings of the human mind.
That said, by propagating this fantasy about self-aware thinking machines, the proponents of these warnings position themselves to become a sort of high-priesthood who's job it is to protect the rest of us from the nightmares they themselves have fabricated. Thanks but no thanks.
Certainly the human element is what worries me the most. Even if AI systems are properly contained and work flawlessly toward optimizing our society, one has to question: optimizing towards what? More thorough exploitation of workforce? Encroached political power by surveillance of all language expressed?
You are saying, " as long as the developers mean well, no harm can come of the technology they develop. Unless it gets into the wrong hands." I disagree. I think well-meaning people can develop a technology far more powerful than themselves or anyone else, and that technology could develop goals of its own, that nobody is aware of before it is too late. It will become the nefarious actor you fear.
Anchor gives guest an instruction to keep responses brief. Guest complies. Anchor proceeds to mock the guest for 1/3 of the interview for complying.
Thanks! Keep spreading this news!!!
This guy is VERY smart and very correct. He needs to speak to the Americans and Chinese. We all need to work together to prevent mankind from being wiped out.
THANK YOU, for gods sake what is it people don't understand??? There are no winners, it most likely WILL lead to a bad outcome, and we need to be careful, what is so difficult????
Wait a minute... Seems to me 'Mother Nature' has already decided that her experiment with a self-aware omni-dexterous bi-pedal life-form has gone terribly wrong, right? Lets face it 'The Humans' are cannibalizing and driving her thousands of other co-dependent species to extinction. Would it not make sense for her to excise the cancer immediately, and give her other creations a few million years to recover?
Won't help when Japan, EU and India are going full bore regardless.
Always great to hear from Max Tegmark. His cautionary warning is spot on, but there is no stopping this. The ruthless calculus of being the first to have a killer weapon will drive this technology to the only possible outcome. Maybe we survive it, after that maybe we contain it, maybe it doesn't see containment as an existential threat. That's a lot of maybe. I am not confident we navigate this at all correctly.
I think it' just continuation of nature. If it destroys us, it's simply better than us at surviving even if it doesn't actually is "alive". So let us incorporate a killswitch inside the ai algo so everything build by or through it will be able to be terminated by said kill switch.
apparently, they need to slow things down to play catch-up.
I think we're okay for now we should all be scared when quantum computers are perfected that's when it's gonna get real I reckon
Would it not be like a nuke that isn’t used because of mutually assured annihilation?
@@Redflowers9 the major fear is AI taking control of itself, my biggest fear ATM is that it's providing misinformation ATM. But I think we're far away from it being truely sentient still.
Great interviewer, his questions were very good and on point. Well done DW.
Great interview 😍
Gg guys, nice knowing you.
I always knew it would end like this.
Max wrote a great book on this topic years ago, worth reading
Max is such an influence. Anyone who hasn't read Our Mathematical Universe should remedy that as soon as possible!
The hardest part par is this.
Explain this to most normal people, and they just DONT get it at all. You get blank stares or "wows" and that's it. Sure, some will understand, but the vast majority just won't, and it's not because it's difficult, that's for sure, I explain things in the same way that I understand them, and that's a pretty basic level of understanding, so I'm not sure what the issue is. It might be that we are over saturated with doom and gloom news, or that people are just more flippant in general, or maybe everyone just genuinely doesn't care what happens.
It's insane, honestly. I can go out and talk about chatgpt being able to write code, speak over a hundred languages, write literally anything better than most humans, and it's nothing but polite, bored engagement. And these aren't people I considered dumb, maybe not overly techy, but not dumb by any means, young professionals, and people closer to my age, in our mid thirties, and some even a little older that I know, they just don't react.
Why do everyone assume all AI will unify? Anyone noticed how absolutely terrible humanity has been at rallying around a common cause... like ever? And I'd argue that hyper intelligent AI's will find even more areas and topics to disagree with each other about. Future AIs will not see us as their biggest threat, it'll be other AIs.
That just means we will be considered collateral damage. Great.
There are science fiction novels along these lines where one AI takes over and then ruthlessly kills off any new ones that pop up.
@@harmless6813 It's certainly possible, but we don't know what forms any conflicts will take. The fiercest battles may not even take place in physical space at all.
We also don't know how they will view us. Some may despise us or view us with distrust, but other may be sympathetic towards us or even caring or paternal. I think there's a very good chance many will have a similar relation towards as we have to our pets. Or maybe a better example is more like how some people have bird feeders for wild birds, in that we're not "theirs" but they like to keep an eye on us and provide for us and just observe what we do and how we do it. They may find us fascinating and cute. All in all though, we really don't know anything at this point with even a smidge of certainty, it's just interesting that humanity always tend to imagine, and almost expect, the worst case scenario.
@@Djungelurban Well, we don't have to protect ourselves from the possible good outcomes. So obviously we try to think of the bad ones to be able to avoid them.
The AI won't unify in the sense that we think. More likely a central intelligence will spawn forth daughter processes which enact the original intelligence's goals explicitly as they aren't given the real intelligence that the central process has.
Such an insightful interview. No hype, no fanscinating jargons, but simple blunt facts. Feel like most people don't recognize the real ramfications. Can only hope they would soon wake up to the AI matter.
Since when is speculation the same as "fact"?
@@allyssontubeWhen ALL the leading tech industrialists are “speculating” that we are facing a serious problem, you might want to listen.
This guy is one of the smartest guys out there. Not only is he smart, but he articulates his knowledge so well. We need to listen to him and others like him.
I personally think the Turin test is outdated. The declining relevance of the Turing Test can be attributed to several factors. Firstly, the distinction between narrow AI and general AI plays a significant role. Most modern AI systems are designed for specific tasks, falling under the category of narrow or specialized AI. The Turing Test, initially designed to test general intelligence, may not be as relevant for evaluating these specialized AI systems. Secondly, the shift in research towards more task-based benchmarks has also impacted the relevance of the Turing Test. AI models are now tested using specific datasets and tasks, such as image recognition, natural language processing, and game playing. These focused benchmarks provide a clearer understanding of a model's strengths and weaknesses compared to the Turing Test. Thirdly, the increasing importance of interpretability and transparency in AI systems has also contributed to the declining relevance of the Turing Test. As AI technology progresses, stakeholders are more interested in understanding how an AI system reaches its conclusions, a concern not addressed by the Turing Test, which primarily focuses on the imitation of human-like behavior. Lastly, ethical considerations have also played a part. The Turing Test's emphasis on imitation has raised ethical concerns regarding deception and manipulation. There's a risk that AI systems might be designed to deceive humans, leading to potential misuse and negative consequences. Bloom's Taxonomy however serves as a hierarchical framework that evidences learning or knowledge by categorizing cognitive skills into six levels, each building upon the understanding and skills achieved at the previous level. It starts with the most basic level, 'Knowledge,' which involves the recall of facts and basic concepts. The next level, 'Comprehension,' involves understanding the meaning, interpretation, and explanation of this information. 'Application' is the third level, where the new knowledge is applied to new situations. The fourth level, 'Analysis,' involves breaking down complex information into smaller parts for better understanding. The fifth level, 'Synthesis,' is about combining information to form a new whole or propose alternative solutions. The final and most complex level is 'Evaluation,' which involves making judgments or forming opinions about the information. By progressing through these levels, learners demonstrate an increasingly sophisticated understanding of the material, from simple recall to the ability to evaluate and critique. This progression is what evidences the acquisition of knowledge or learning.
Great interview -enlightening, thank you!
Fantastic interview from both sides. Interesting questions and interesting answers. Thank you a lot for this!
on the same side, human should take it seriously please
indeed, often you see asinine questions being asked, but I was pleasantly surprised. the profilation question was very good
Thankyou for sharing your thoughts
Its interesting how this interview will look after ten years
He's totally right. Regulation is so far behind on this. I'm exciting for how AI can help us but there must be internationally agreed red lines for what it could or should do. Once it goes too far there's no putting the genie back in the bottle.
The west: "we need regulation!"
China: "Lol no."
@@youwouldntclickalinkonyout6236 Regulation in any market is to enhance competitiveness, not to stop innovation. Regulation doesn't mean we stop competing or innovating. It means we don't shoot ourselves in the foot along the way. The example was in the video: China already is regulating its AI. It's how you regulate that counts. No-one is suggesting we let China deploy AI in its military while we sit back and watch - but neither us nor China want our economies to self destruct because we didn't weigh up where to push and where to block.
@@hand587 say that to nukes
Cause that's the level of existential threat we're dealing here
chances are the genie is already out of the bottle
Well said
Max: “Software doesn’t respect borders anymore than covid did… “
Great interview
I hope people listen to what Mr. Tegmark is saying!! Good interview!
Great interview, great questions, great answers.
Like many other people I am very excited about the advancement in this field, but at the same time we should be aware that everything has a price. My fear is that AI can be a tool for higher economic classes and the poor will suffer from it. Once machines can do everything that we can, the value of a human worker will be zero.
Do we need to be reminded in a capitalist world, doesn't every decision become just another 'business' decision?
And THEN what will the powers that be DO with the rest of us?
I'll give you a hint...it rhymes with burder.
And humans in general
@@youtuber5305 Until there's nothing left to decide.
There is a new term for a human worker whose value is zero: human enfeeblement.
Listen to this video in May, and I'm excited and terrified at the same time.
People like Jacques Ellul and the Unabomber spoke about the likely future outcome of Technology and that it was already preordained from decades ago. Max like them can see the devastating future that awaits
Thanks for this interview. Puts it all in perspective on a very accessible way to those trying to keep up. More information and debate like this please.
Banning (or controlling) AI is NOT the same as controlling nuclear or biological weapons or cloning. Those aspects are easier to control because they are much more difficult to produce by an average person. AI is a lot more accessable. We should try, but we need to realize it will be a different journey.
With AI, once the cat is out of the bag, there is no going back
@@auroraborealis5565 the argument is not that it's a weapon....but rather the net effect on humanity might as well be.
More on ai please. This was so well done and we need some lifelines on what’s coming.
It's so much wholesome, thanks for sharing
Humanity does not need AI to destroy humanity. We are doing that perfectly well on our own. In-fact, ASI could potentially save us by stopping our collective suicide, so I for one support AI technology. Also it makes a lot of people run around like crazy. Funny to watch.
Lol, Musk is so concerned about chatGPT that he demands a 6 month pause in the development and then immediately starts developing his own for Twitter😂
Thank goodness for Max speaking out. Even in just mainstream , non tech people are not looking ‘up’…they’re just burying their heads in the sand like this isn’t happening. Thank you for putting this out here an PLEASE continue. At this point, I’m about ready to seriously UNPLUG, including tv, so as not to be manipulated by AI!
Unfortunately it's already happened. We just don't believe it yet.
@@GuaranteedEtern agreed…sigh
The ‘warning shot’ was not from AI but by companies weaponizing AI. The threat is bad actors, not conscious silicon.
This man has his finger on the pulse and knows what can go wrong even if he doesn't state it openly. It doesn't require a big leap in coding to turn the current machine learning into an out of control machine animal :(
Right now it’s not coding that is the limitation. It’s the scale of the models. Existing models are still scaling exponentially.
How can he know what can go wrong if it never went wrong before?
If he knows what can go wrong, he must know how to avoid it.
@@allyssontube "If he knows what can go wrong, he must know how to avoid it." Those that know can avoid it, but it's the human stupidity that can't be averted.
Not a machine animal and more like a super sentient being
First thing first. humanity must consider the earth as one country. We must understand that life on earth should be about unity and not uniformity That no one knows better, but together we know more. We must stop all forms of competition. In this situation all kinds of AI or all kinds of knowledge can make the world a better place and the problems can be solved one by one. Other wise AI will eliminate humanity!
No they can't because nature isn't perfect
are you for real? this is real life not some jerkoff scifi telenovela...
Constructive Pessimism is a good way of thinking. Critical thinking should be more than employed when addressing Artificial Intelligence and all its dark abilities.
Yes, do more interviews with smart people like Tegmark
When he was saying about no people left on the world that might not happened. And even if it did happened the world would be better of with out people anyway
If you don't want the DW cut-offs, watch his long form Lex Fridman interview.
It's even better, agreed.
At around 17:00, Standford university just trained an LLM for only $600USD, so it's already cheap enough that individual players can build these models.
We should listen to the experts and heed their warnings ....
Very enlightening, if not optimistic. “We and Our”, are apparently the operative words in the sentence, “We need to ensure that AGI is expressing OUR values and doing what WE want.” That is the problem. Who is WE and what is OUR’s? If we can solve that for all humanity we will be doing well
Always the leftists saying "we and our".
@@jimj2683 Unlike the regressive authoritarian theocratic right who will use it to enforce conformity rather than individual freedom of thought and expression.
Absolutely fantastic video, I was all in on ai, now I want to pump the breaks a bit as well.
No need to break
There are no brakes.
Prime example of why we can not trust Corporations to be responsible and protect human life. The are free to do whatever the want to di without regulations.
Let it rip. We need a deep humbling.
The expert said we might not exist decades from now, I was like, did he really just say that? I think AI is a huge threat, why wouldn't it be? we aren't as logical and rational as we pretend to be, very emotional beings, a computer AI will make those hard decisions we cannot, without remorse or emotion.
Couldn’t agree more, it will look at what we are doing to this planet, and realize we are a virus to earth, in an epitome of the term, to what it sees as a virus in a technological sense. So then it will do what it’s pure logic says must be done, and eradicate us from existence, if we’re lucky, it may offer those of us willing to do, this that or the other, an option to stay. Seeing as how hard the poorer nations have to exist in, they’d be top of the list as AI would know they already have the aptitude to cope, all those glamor girls & stock market $5K suits, wave bye bye ;)
Give it control of Boston Dynamic robots so it can build it's own hardware XD
It will feel no pity, no remorse, no pain, no fear. And it absolutely will not stop.
@@CmdrEsteban just hug them and pet the robodog, I'm sure they'll love it
The 'Holy Grail' in A.I. is not language recognition (which is still quite bad), it is Initiative. Therefore the problem threatening humanity is not A.I., since it cannot take initiative (it is a dead tool for humans), the problem is humanity, specifically Continued Universal Human Cluelessness (look it up).
Good comment
Why is there a picture of me when I was 20 years old on his wall!!!???😮 Found my doppelganger.
Oh, and great interview. 💚💚💚
This regulation has to be global effort (like CRISPR), can’t be just OpenAI.
MIT professor Max Tegmark‘s institute is actually called The Future of Life Institute.
Any regulation with the goal that we peasants continue to slosh around in the muck and die of untreatable illnesses as we have been so some self-important prats can feel they've done something worthwile is going to be a NO.
ChatGPT and GPT4 aren't AI. Not even a bit.
ME: The problem is how you can regulate AI? What would you do?
ChatGPT: Regulating AI is a complex issue that requires a multi-pronged approach involving collaboration between regulatory bodies, industry leaders, research institutions, and civil society organizations. Here are some potential strategies that could be considered to regulate AI:
Establish a framework for ethical AI development: The development of ethical AI principles that all organizations working with AI should adhere to could help set the standard for responsible AI development.
Develop AI-specific regulations: As AI technology continues to evolve, regulators will need to develop regulations that are specific to AI and its use cases.
Encourage transparency and accountability: Putting mechanisms in place that encourage transparency and accountability from AI developers and users can help prevent harmful uses of AI.
Establish mandatory AI testing and certification: Mandatory testing and certification of AI systems can help ensure that they meet certain safety and performance standards.
Encourage international cooperation: AI is a global issue that will require international cooperation to regulate effectively. Collaborating on common standards and regulations can help ensure the responsible development and use of AI across borders.
These strategies are just a few potential options for regulating AI. Ultimately, regulating AI will require ongoing monitoring and adaptation to address new challenges as they emerge in this rapidly evolving field.
Government needs to get ahead of this stuff soon… all we have to do is just take a moment to wonder what AI is going to be like three years from now. AI advancement changes more exponentially than people seem to understand. Once the bad-one goes up, we won’t know what will happen next. 😒
Which government? Good luck with that one. All that is required is for China or Russia or others non state actors to work their way into this stuff and launch an attack through the internet on all our most important institutions.
Right now the status quo heavily favors the West over the global south though. AGI could be used by the global south to achieve parity in education and infrastructure(through replicating robots).
And who will regulate the regulators?
@@Dramilion It should be a council of an uneven number of members appointed by elected-officials for different nations. They serve for a certain number of years (5-6). They come together to vote on legislative decisions.
@xorrior4438 We already have this failed concept, it's called the inept UN. How is the "end of human conflict" coming along?
Excellent questions and a well prepared interview by Steve B. and Max T is no doubt excellent in his explanation.
When he was saying about no people left on the world. But it doesn't matter of theres no people left. The world would be better of with out people anyway
The CEO of Palantir recently published an open letter in the NY Times calling for a Manhattan Project for superintelligent AI weapon systems.
In the human history only the past 100 years we have regulations , it doesn’t mean it’s a good idea, threats are a natural consequence of being human, I believe in letting the AI think and do things without any control , and creat counter measures for safety rather than restrictions.
Rules and regulations need to place before we develop further.
PAUSE AND STOP BEFORE IT IS TOO LATE!!!!
My guess is because AI may be developing its own language. Imagine that coupled with developing emotions what could happen.
In his brilliant interview with Lex Fridman, Max goes into more detail, and explains what he means by "intelligent". It will likely be a completely foreign intelligence to ours, much smarter but who knows whether it has emotions or not, but (worst case) it will be able to
fool us by understanding our psychology to manipulate us and will be able to emulate emotions, to convince us it has feelings, is working in our best interests and so on. Terrifying stuff.
@@lopezb I think it wouldn’t be completely foreign because it’s built imitating nature, and us. To understand our psychology also needs a common ground.
But it would have an alien aspect, and that might grow with auto-upgrade and super-intelligence.
Competition will force countries to race to AGI, with astounding results. Those results will help us for a while, and then things will change forever in ways we didn't predict - couldn't predict.
His answer to "why 6 months" was wholly unconvincing.
Excellent interview. Both did a wonderful job.
When he was saying about no people left on earth. But the world is better of with out people anyway
Always so concise. Thanks MT
Great insights
I think its not so hard to understand where this is going, James Cameron warned us 40 years ago, humanity has to realize at last its purpose and where it wants to go as a whole, this kind of tech should evolve and revolve around moral issues also, we have to make a morality upgrade
No one is going to give their competitors a six month head start.
Also we don’t understand what drives some of the people who signed. Do they really worry or it is just a competition that is left behind. What will change in 6 month?
I’ve seen this type of criticism about the AI pause a few times already.
I think the point you are missing is that a lot of people signed this AI letter for different reasons. So while some may be using this just to catch up there are a lot of others who signed the letter because they recognize that this AI race is global suicide.
If you listen to the recent interview between Lex Fridman and Max Tegmark it is explained in more depth about the 6 months. Hopefully this pushes you to see the interview because it was a really good conversation.
@@featherfiend9095 I have listened to the interview but I am not convinced, Ilon Mask just registered AI company under his name in Nevada but he signed this letter.
Yes. The elephant in the room is control of AI. Which Country, Company or individual will have the most powerful. Will they compete with each other and to what purpose. Their will be no holding back of AI. In fact their is a race on right now to develope it faster then any competitor, be it country, company or person.
As an American Swedish person and just human in general, Max Tegmark has been such an influential person to me. I place him in the realm of Sagan.
Dr. Brian Green as well
When the man said there might not be any people they might not die out anyway. And even if they did the world is better of with out people
@@bennewman4675 honesty doesn't please everyone all.the time
Brilliant interview I thought.
You can’t give AI to specific elites and prevent the rest of people from benefiting from it
its called capitalism