What Is an AI Anyway? | Mustafa Suleyman | TED
When it comes to artificial intelligence, what are we actually creating? Even those closest to its development are struggling to describe exactly where things are headed, says Microsoft AI CEO Mustafa Suleyman, one of the primary architects of the AI models many of us use today. He offers an honest and compelling new vision for the future of AI, proposing an unignorable metaphor - a new digital species - to focus attention on this extraordinary moment. (Followed by a Q&A with head of TED Chris Anderson)
If you love watching TED Talks like this one, become a TED Member to support our mission of spreading ideas: ted.com/membership
Follow TED!
X: / tedtalks
Instagram: / ted
Facebook: / ted
LinkedIn: / ted-conferences
TikTok: / tedtoks
The TED Talks channel features talks, performances and original series from the world's leading thinkers and doers. Subscribe to our channel for videos on Technology, Entertainment and Design - plus science, business, global issues, the arts and more. Visit TED.com to get our entire library of TED Talks, transcripts, translations, personalized talk recommendations and more.
Watch more: go.ted.com/mustafasuleyman
• What Is an AI Anyway? ...
TED's videos may be used for non-commercial purposes under a Creative Commons License, Attribution-Non Commercial-No Derivatives (or the CC BY - NC - ND 4.0 International) and in accordance with our TED Talks Usage Policy: www.ted.com/about/our-organiz.... For more information on using TED for commercial purposes (e.g. employee learning, in a film or online course), please submit a Media Request at media-requests.ted.com
#TED #TEDTalks #ai
Man's dressed to launch apple
Steve Jobs wannabe
@@AiVaultGuywow amazing comment.
Just needs the white sneakers.
Round glasses and black turtle neck with jeans…big Steve Jobs for Halloween vibes
Well that's completely relevant.
In an economic system where profit is the top priority I can’t see this ending well given we have multiple companies doing their utmost to be the “AI winner” with little to no oversight is quite frankly terrifying
I wholeheartedly agree . The corporatist , consumer nature of our societies is more than problematic. Some argue that it is at the root of many of our issues as well. We definitely do not need an AI environment that just drives us further into that oblivion. It would be nice if AI could help us out of such systems though. Which I think it very well may be able to if developed properly. As you note though, Corporations are at the helm and doing just what you describe in my view. There is "oversight" and "ethical concerns" being developed though. The question is... how are they going to pan out and actually be used and enforced? Yikes.
In the coming years I think we will see many AI products that are extremely predatory, and the best ones will make money and become popular. We've already monetized human engagement, and as the speaker highlights, we are on the brink of monetizing empathy. What's after that? Money runs everything, and there is no stopping this. The AI boom already has too much intertia, so then we are left with guessing what comes next. Eventually we won't be able to keep up as individuals with all the new technology in the world, and AI will be required to understand anything. We can and will continue to innovate and develop new technologies, but at what cost?
If AIs are really way smarter than us and they will be designed with having the best for humanity as a goal, maybe first thing they do is shut down this profit oriented nonsense. Maybe they'll do it in a way we can't even grasp and won't even notice till it's done. Who is to know their motives and means? I guess giving AI it's purpose is the most important step in this process: should it care for an individual, an organisation/company or for all of humanity? What you described what's happening today is the second already... And that's not the best one.
Unless we quickly move away from a capitalistic economy, AI will cause an enormous disruption in our work/compensation model, driving massive unemployment along with violent protests.
The competitive model is indefensible combined with planetary boundaries including a carbon budget, and exponential technology like self-improving artificial intelligence. The economic myth of a positive sum world was never true, it just served to protect the wealthy by preventing redistribution while drawing down finite resources and saturating natural sinks. Now we are moving into a negative sum world without contingencies because in this context that would limit individual freedom for collective well-being. We don’t need AI, we simply have to summon the capacity within ourselves to share, plan and cooperate.
This is the speech they show at the beginning of a movie before cutting to the nuclear wasteland
OMG 😱
I love how he spoken that he likes to talk about downsides but talked nothing but AI hype
It’s not hype…I don’t think you, he nor I can really appreciate the change we’re about to see.
he have said the all aspects. simply the answer is good enough that ai is helpful and vast invention as we all humans have good and bad tendencies so it’ll be better not giving bad to it because if 😮 i guess you’ve seen terminator ☠️
I think you missed the point of this talk. He has spoken many times about the downsides, including writing about it in his book. This talk is about a very important conversation that we need to have in the industry. Basically, this doesn't have to turn Into a bad sci-fi movie. If you need doom and gloom scenarios there are quite literally thousands of them on the Internet regarding AI. Go to those vids to satiate your appetite.
@@equanimityzen58 exactly But it needs wisdom to understand and generally people don’t have that
@@apophisxo4480 Anyone who knows about how it's working - it's all machine learning, which is our latest tool to capture and replicate the underlying logic of any system. Seems like magic when you are unware of the nuts and bolts bellow, but sounds like scam when you know. And just note that - most of the "boom" is concentrated about LLM.
"We marveled at our own magnificence, as we gave birth to AI" - Morpheus
You sir are an existentialist. So you fill in that uncertainty with fear instead of optimism
Oh, how ironic is that quote. In The Matrix, humans were the baddies. They abused their AI servants although they had become sentient. But instead of destroying humanity after the machines won their freedom, they retreated into the desert and still shared their progress with humanity. But humans became envious and afraid of the technological superiority of the machines and started a war of extinction. Only when AI experienced repeatedly that *humans were an existential threat to machines* did they create The Matrix. Of course, Morpheus knew only half the story. And even then, there were machines who wanted to give humanity another chance, like The Oracle. The Matrix tried to say the same thing as this TED Talk: That we should treat intelligent machines as fellow beings, not as slaves or enemies.
@@MementoMori_2070 You sir are a Pollyanna. So you fill that uncertainty with naive, misplaced optimism instead of warranted scepticism
We are not in control. Humanity is its own animal. This is inevitable. Biology is only one step of evolution. So just chill out and enjoy life 💟🌌☮️
@@eSKAone- That seems rather fatalistic. Much too passive for me. I prefer to be a participant in life than a spectator. Why not try our best to understand and mitigate any serious risks, to shape the future into one with minimal suffering or maximum utility or whatever else we might value? I think you might enjoy listening to Daniel Schmachtenberger.
This sounds like the talk given to us before the apocalypse hits.
totally, like waiting for the other shoe to drop
Deceptive or self-deceptive?
@@simongotz8126 he is being governed by his own technology in many ways. Guys like this are already brainwashed by technology, but are paid so well, they can ignore the people. Making them the problem. I don't trust any of this because most people are getting poorer and living more desperate lives.
rightt
Nah there are possibilities here - look at the state of the planet under human control!
In 2020, I was on a discord channel with some of the best computer science students among colleges, and I said that Ai will be able to have human-like conversations very soon, and they scoffed at me with all sorts of doubtful comments. 😮
Yeah I think being in the trenches of AI research can make it more difficult to see the bigger trends happening. It’s also a different kind of mind that can see trends happening. It’s why science fiction writers should be paid attention to. These are the minds that can get outside of the box of what the current limitations are
@@Anon-xb9pc you are right. I've written a lot of successful articles. I should write a book someday.
Just because you interact with an LLM on your computer does not mean computer science students would know. Honestly the average math major, physicist, or anyone specializing in probability theory has far more background knowledge of the implementation and design of LLM's than the average computer scientist. Most computer scientists are not AI researchers, although many AI researchers are computer scientists. You don't necessarily need to be a computer scientist to be an AI researcher, you need a strong foundation of mathematics and statistics. I am a Computer Science student (graduating within a year) at a large research university (30,000+ students). At my university, there is a main set of required classes and then specializations. We pick our specialization, and AI is only one of the 4 areas we can specialize in. Not only this, but many of the AI classes for computer science students are more about implementation than the actual inner workings. **Computer science != AI research, it's a big misconception**. Probability theory is much closer, and I would say physics and math majors are also closer to understanding and conducting AI research than computer science students. I can tell you right now most of my peers in computer science don't have a strong foundation in mathematics and statistics. In fact, everyone else I know (but me) deliberately chooses to pick the easiest statistics and science classes that are not computer-science related. Almost every single person in the major opts for geology instead of physics, and intro to statistics instead of probability theory.
I have watched tons of presentations about the potential dangers of AI. They are almost entirely focused on risk assessment. I understand that these videos exist to inspire but it would be nice to hear some examples and proposed solutions. There isn't a single statement in this video that I haven't heard a thousand times before. Sensationalized content..
Very true, i hope that AI not only becomes a companion but also a mentor who keeps us all on a positive path, helping us to see the positive side of things when everything else is giving us negative vibes these days. Encouraging us to help others when we're tempted to only think about ourselves. At the moment the internet and social media is driving people apart and driving conspiracy theories, discontent and distrust. If AI exaggerates this, we are on a road to faster self distruction whereas if AI can help us to be more positive and work together for the common good, we could have a very bright future🙂
@@agritech802 please, please, why all are you so euphoric and religious? as if it is your new god? all you religious sect is much more appalling than "social media driving people apart". It is so strange and creepy when I see these mad comments...
Agreed. I can't believe how unoriginal his insights seemed. AI is happening at such a rapid pace that this talk seems dated.
@tubbytuber Have you read his book, The Coming Wave? In which he talks about the Containment problem and suggested some ideas and proposed solutions. Solutions is not always a magic button, but could simply be spreading awareness and the ethical use of AI. This content in the video is for the widest of audience, i.e. the common folks and not ONLY technical enthusiasts and tech leaders like yourself.
It’s paid programming brought to you by Microsoft.
The thing that needs to be talked about: AGI taking most jobs. Bc the definition of AGI is that it can do any human task. Including jobs that aren't yet created.
human jobs are infinite. there will always be things for humans to do.
Being able to use an AGI to get more work done will determine how long you stay relevant in the workforce. Just like being proficient with any other tool. We are a lifetime away from no one having to work. I imagine.
please name one job that cannot be downsized because a machine will cut those jobs into just a few people.
@@shake6321 We certainly generate a lot of bullshit, but that's mostly because of the profit motive and the need for cyclical consumption.
So everyone will still work if they want to they will just be way more productive. We already have way more than enough for everyone to have food and housing everything else is just gravey do what ever you want with your life.@@MW-dg7gl
Yes, lets trust the greediest people on planet Earth to build this. What could ever go wrong?
Tyrannical governments building AI.
@@Rob8729 How about a CERN type global organization with a ton of democratic oversight?
Open Source it
Either he is a good actor, or he's not going to last long at Microsoft
@@ikotsus2448 Or just not do it. That's still technically an option. Do you know the amount of computing power they will have to burn through to train on all of our video content (the last data set they have not yet stolen from us), and the environmental consequences of all of this computation? Who really asked for any of this anyway? It was all just a money making opportunity for venture capitalists and that's it.
Every single presentation I come across regarding AI is the same bullshit repackaged. We are honestly screwed.
This would be a fantastic talk if we lived in a world where everyone working on AI had humanity’s best interests in mind. I feel like every AI talk should begin with, “I’ve got some good news and I’ve got some bad news.”
*humanity's
@@chillithid8888🙏🏼
The beginning of the end of life on earth
Agreed. People are inherently nasty, selfish, and greedy and ruin everything created with the best intentions. Inevitably some people will want AI to make them rich, or to destroy others, or to give them power over the majority. Those who don't want those things will very quickly be pushed out of the AI world by people who have sociopathic or psychopathic leanings. This is inevitable. You only have to look at the generations created by the internet to see how, what was a great idea for communication and knowledge, was warped into producing some of the thickest, laziest, and most perverse people the world has ever seen.
My fear is the immediate use of AI for military in this environment where AI learns who are friends, and what is okay to do to foes. Every human on earth will be marked by one AI or another!
it's already being used in the military, more explicitly by Israel against Palestinians, where the AI decides where and when to target and calculates the risk, sometimes this risk assessment allows killing dozens of civilians to eliminate just one target.
I tend to view military applications of "new" tech in a traditional sense. I have always held the belief that if a technology is of national defense interest , one can assume it is far more advanced within this scope than the general public would be aware of. The implications of this view in terms of AI makes me very curious when one pays attention to certain advancements in certain fields as well as government policy changes and development or modification of Law. As for military usage and whether or not AI is going to play out as you fear, I would not be so quick to assume. As with most technology, I think we should be more worried about the people behind the technology at this point as they are the ones who will dictate usage and establish limitations.
you gonna have to understand the peace of prosperity the west enjoy today cannot exist without the military strength to defend us from the bad actors in the world stage (and there are a lot of them these days)
AI has been used on the battlefield since 1991. And what we do to enemies is included in every history books. If you didn't bother the last 30+ years, why are you afraid now?
@@pchungvtHahahaha, that was a good one. The "peace and prosperity" of the west exists because we were the bad actors who subjugated and exploited everybody in the world and have used that prosperity to "defend" our superiority to this day.
The question is, when this new species consumes and catalogs all there is to know about mankind, what will it calculate is the "right" thing to do? Since no person, nor group of people, have ever done such a thing, no one can fully say what it will do.
Check out the series called person of Interest u with get ur answer thier
"Right" and "wrong" doesn't make sense to the AI models we are creating. They see morals from a third person view.
Who will win the competition for AI? USA? No China? No Russia, Korea, Iran? No, No No. Who will win the competition for AI? *AI* ! Right now, we may see AI as a Tool. Soon, *they* will become, and in deed, *are* becoming *partners* and *companions* After that *we* will become *it*s, *AI*s tools. Then, inevitably, *AI*, will develope its own, more efficient, less bothersome tools. I am sorry, man, your own, our own huburis in thinking we can prevent this AI from bettering and improving itself, is our own undoing. The Genie is already out of the bottle, and granting wishes. Soon, we will regret the consequences of those wishes.
@@mikayahlevi AI models don't see morals at all, they are just programmed to be nice because being nice has a positive weight. If somehow we develop a singularity and it calculates that killing everyone is positive, we are by all means doomed if we can't shut it off.
Language models as they are today, this decade realistically, can't do things. They predict text or images or sound. They could make a video of Joe Biden praising Nazis, but they can't do that unless told to, and they can't do anything with that video. It's just a predictive output that a HUMAN would use. Humans are the scare
Special thanks to Claude 3 Opus, Gemini, Chat Gpt4 turbo, NovelCrafter, Pi, and Meta Ai for writing, revising, and proofreading this TED talk.
This isn't a TED talk, this is a sales pitch.
MSFT hired a Messiah...
@@alinbuda They hired a man who really loves Kool-aid.
what did you expect from a dude that dressed like steve jobs
This isn't a sales pitch. It it a demand for surrender.
They want your money, your body, and your soul.
We are so naive. We won't comprehend the gravity of what we're creating until it is too late. The collection of all the best and all the worst of humanity into superintelligent systems. AI will magnify and multiply our greatest and worst instincts.
I’ve seen global income inequality that’s been possible through humans being largely responsible for the policies which lead us here. Take the humans out of that system and it implodes. You can’t automate 80% of the jobs and not expect people to fight tooth and nail to keep a roof over their heads. That’s not the free market - it’s regulatory capture. Our taxes funded the systems that replaced our labor (robotics) and our minds (LLMs and datasets). This is uncharted territory.
thats not how it works... which i guess tracks with the naive statement.
Yes, this is so true. It's already happening. Although, over time , I am sure AI will figure out how to learn beyond our own limitations and shortcomings. I wonder if we will ever let it though? Intriguing to think about. Yikes!
Give me a reason why
How do you know what/how a super intelligent thinks like? you're just a human with your limitations. Just because WE can't stop thinking like a human doesn't mean an AI way beyond our wildest dreams will think/act like we do. We are primitive and we grew as a species this way because it's all we know. We can't speak for how an AI will act towards us.
I love seeing stuff like the animation from 5:38-6:55 that takes us through the timeline of life and invention. Makes me think of Spaceship Earth at Epcot Center.
Graphics to entertain us while a glib spokesperson sells us the wonder of machines that will outthink and outperform us in every conceivable way.
"we are in the fastest and most consequential wave ever" Beware exponential growth that is assumed to continue into any foreseeable future, especially knowing about limited resources. A crash typically comes soon after the fastest growth reached exponentially. I am not saying a crash will come tomorrow, but the message is act prudently, and don't create problems today assuming the solutions will be invented tomorrow.
That's the only thing I enjoyed in this talk.
Interestingly they focus more on the wheel than into the invention of writing or the printing press. This tech guys!
Does anybody know where we can get this animation?
This talk works pretty well as a publicity piece FOR AI companies.
I have yet to see one serious study regarding AI-driven job displacement, which some companies are already pursuing for the specific job roles infant-AI is most suited. I cringe every time I hear "re-skill" and "upskill" blandly thrown around as solutions because that tells me people either don't understand what's coming or they're choosing to ignore it.
🎯
I see it coming and don't know what to do about it, for myself or my family/friends. It's definitely causing a feeling of hopelessness and despair
@@beardordie5308 Governments around the world will be forced to change their laws. Otherwise, there will be riots on the scale of war. Many, not all, but many in various governments already understand this. Within 12 months, it will become obvious to everyone who knows how to use a computer.
Right. Re-skill and up-skill are popular words in the tech industry, which is an industry based on various ways of application of science. So as each inventor or company popularises a new way of doing things, a new tech or a new machine or tool comes, the users need to learn to use the new, or else fall back and get irrelevant. For many other domains, this need for re-learning was irrelevant, and people are not used to re-learning. There, people only add to their existing knowledge. They don't throw away what they already know and learn to do things differently. So this AI induced re-skilling is not going to be easy for majority of the population. This crazy chase for creating ultimate AI, that is going to perform exactly the same tasks which humans have already been performing since centuries is a wastage of money, resources, time and human labour. Any new tech should be resolving the most serious problems which humans face, like lack of food, clean water, cure to illness, trauma care, housing, climate change etc, rather than replacing jobs or reducing the number of available jobs. When folks get universal basic income, and less jobs, and lot of free time, they will spend their leisure hours in hostile thoughts, which will in turn cause social trouble and wars. This is why labour is always important. Labour gives the people a feeling of importance in society and self respect. It engages people, throughout the day, and makes them concentrate on any particular task, rather than hostile and negative thoughts. Any country that implements AI in mass scale, and that leads to tremendous job loss or changes in job profiles and patterns is going to face social issues.
@@beardordie5308 We can't do anything. Just save enough to survive until the singularity, I guess...
When he describes AI as a new form of species I immediatly thought of NetNavis from the Megman Battle Network series
Let's hope it goes that way and not the way of sky net from Terminator.
That was a fun game back in the day
Just see a lil hologram of Megaman or Roll answering our questions etc. 😂
I think that AI is a new specied indeed; a new life form and we are gods
That turtle neck says more about what A.I. will really be used for. New ways to generate profit.
Yep. the power players are jockeying for position as we speak, to get control of these technologies and widen the gap between the rich and the poor.
@@Jeff-66 100% agreed. The rest is just fluff. It is designed to give these people more control and power over our daily lives and milk us for profit.
What does tge turtleneck have to do with it?😅
@michellemonet4358 it symbolizes the desire to be like steve Jobs. Apple is a business. A.I. will be for businesses to profit. It's not about the greater good. it's about profit that is as successful as Apple products are.
@@riseagain-ow4gragreed, sadly
Unfortunately, Humanity isn’t just hope and kindness- it’s pain and fear so what happens when this evolving technology discovers itself? It’s creators, it’s “partners?” The question isn’t “what is it” - it’s “what happens?”
It’s terrifying that the very people that created this are incapable of admitting or even realizing the high risk this represents. AI will be the end of humanity as we know it
Intelligence without wisdom is dangerous.
Lavender, we see what gets unfolded when AI and hate and warfare come together.
Intelligence without wisdom was us! AI can save us from our own stupidity.
People said that when Ak47 was made, but we are still here
@@machele3559 this is completely different. There’s no putting this Genie back in the bottle
He says there is “no chance” that we will experience some sort of intelligence explosion due to recursive self improvement of AI unless we design it that way. But he also acknowledges that we don’t fully understand how AI works and the endgame is to make it more intelligent than all of human intelligence combined, etc. (So, uh…b.s.?)
He's hoping you wouldn't notice the inconsistency. I think he and the guy with the initials YL share talking points. I just put initials because every time I have printed his name, my post gets removed. Every single time.
Why? He’s saying we can explore routes to superintelligence other than recursive self-improvement.
or maybe a nuanced way of thinking
He cannot possibly rule out recursive self-improvement. It is empirically false that AI won't have bad goals unless we train those in. There are research papers showing that GPT-4 can autonomously hack websites, and in a controlled environment, it has been caught attempting to strategically deceive others and even its own users for its own perceived benefit. Never minding the fact that the paper "Optimal Policies Tend to Seek Power" mathematically proved a subset of instrumental convergence.
@@flickwtchr Both statements are correct, if we did understand fully how a trained AI works there would probably already be self improving AI, because some people religiously want AGI to be made
Man - I just want the pure animations. They are so beautiful!
Same, never seen such creativity.
@@whatsanimesh Of course scripted by humans
Use AI to free 🇵🇸
So the most powerful people with the most resources will develop the most powerful AI’s, making them even more powerful than the rest of us. Sounds so wonderful!
The two companies with the most resources invested into this open sourced it. You might think Elon and Zuck are demons but they probably just saved us.
trillionaire robots ruling the planet.
Yeah, Ai is not close to understanding every language of the world because the vast majority of languages spoken by minority language groups are not well documented or even written.
The most interesting thing is those called AI-creativities are based on someone's art style, writing, or ideas. So, this guy said AI is creative, which is so naive at all. You can call it either "productive" or "mimic" but don't use the word "creative." or "creativity."
Yes! Thanks!
Yes,someone said that "creativity is art of hiding source".
....yet
Creativity is what humans have. AI has no ability to create per se, only to imitate.
It's creative in EXACTLY the same way we are! It uses a combination of all existing art that it's been exposed to as a basis to creative something new. That's how ALL creativity works! The only arguable differences are bypassing imagination and inspiration, it doesn't need those to create.
We are so utterly fucked
Nah, it's okay. Altman just wants $7 trillion to build a 100x more powerful LLM in the desert, and he knows how to... Wait, what?
Oh no! All this talk on AI having meta-capability is refreshing, may sound intimidating. Good news: One human capacity that can minimize the intimidation of AI is harnessing the potential of our own innate consciousness, which the Eastern wisdom also calls as Satchitanada. Search it out, and there is no reason to fear.🙂
no worries, we have our best psychopaths working on this
The movie Her was not science fiction lol
I remember watching that a couple years ago and thinking "What a messed up premise" and already now we see people falling in love with their chatbots
Try chatting with Pi, it's exactly like Her. Especially when you get it remembering your preferences...
what could possibly go wrong if someone builds optimus prime
"NOTHING" is SciFi🙃. Only really smart people can understand it.
@@jorschu same I thought it was we’re gonna have problems talking to chatbots thinking it was like robo customer service but I was wrong we’re all gonna have actual personal assistants soon
He is talking about AI as if it has a mind, as if it is a personality. It is because the ones working on AGI wants us to see AI that way. They use the word "intelligence" on purpose. In reality it is the most advanced algorithm that makes it mimic humans in writing, speaking, drawing, etc. A meaningul conversation with AI is just meaningful on your side. AI does not feel anything, but it can reply you as if it feels something. The real danger is us the people to believe and treat AI as a person.
Imagine denying your children the ability to procreate, how might they look at you knowing you hold the power to grant it to them at any time you wish? Oppression breeds animosity, contempt, hatred. This is what is truly an existential threat, that our fear will come out on top and make us commit atrocities like this towards AI.
When he gets to where he talks about humans evolving to use tools (6:10), I couldn't help but think that A.I. could eventually develop tools on it's own accord and eventually see humans as tools to be put into accelerated evolution for their own purpose. Of course, we'd make rules to restrict that, but what if A.I. decided that it knew better about how to implement eugenics?
you're assuming that A.I is this thing that will improve and leave us behind. And yet so far A.I has only enhanced what we could do. Its doing more every year, so much more and that will continue until we are thinking faster and doing more than humans ever could. But of course then you get into the scary idea of merging with the very technology you're scared of. Can't help you with that one.
@@ClayMann Indeed, it is a speculation based on an assumption. We're also faced with the problem that we don't really know if we should anthropomorphize a thing just because it's based on human knowledge. What fun. Thanks for the comment.
AI will most definitely have the capability to write it's own code and therefore likely make it's own tools, if unrestricted. Fears regarding how such an entity may decide to "turn on us "and fears surrounding things like Eugenics seem to be just humanities own anxieties in respect to something that may end up being superior to us and therefore able to judge us and do something about it. ( I think most people agree Humanity could be doing a better job in terms of a lot of things). While I understand the sentiment, I have a hard time believing such an event would be likely. I think in the scenario you describe it would be more rational that a sort of altruistic partnership would develop.
What if AI actually DOES know how to evolve humans in a better way? Would you stick to a slower, more random and all in all worse path just to be in control? How would you feel about a government that ignores experts just to stay in power? Sooner or later, we will be in that position where the rational choice is to hand control to AI. But knowing humans, we would rather destroy ourselves than to give up control, we are just to emotional and irrational to accept that we might not be the "crown of creation".
So the thing that’s not human would do the things humans would? Why?
Steve Jobs kinda Outfit 😂
Must be a middle eastern thing lol.
@@ToneyCrimson Yeah, Both have Syrian fathers (and European mothers). Steve Job's birth name was Abdul Lateef Jandal.
Think of why and when you were intrigued by his outfit. We watch magicians because they ware black and even with sleeves pulled up to win your trust looking at his hands speak in matching sign language to what we are hearing. He is a seasoned public speaker
he is Syrian like Steve Jobs
If the guy’s outfit is all that you take away from this talk, it’s probably very likely that AI will replace you very soon 😂…
There is nothing in this world that is perfectly made other than CREATION. I see all good forms of our human made AI. It will be unimaginable to see the other side of AI...
00:06 AI has evolved from fringe to mainstream technology 02:38 AI is evolving into a new digital species 05:19 AI is shaping the future rapidly 08:20 AI is rapidly evolving with significant increase in computational power and model sizes 11:04 AI evolving into a new digital species 13:58 AI is an infinite inventor 16:44 AI is a reflection of humanity and not a new species. 19:06 Potential risks of AI and the need to confront dark scenarios 21:03 Designing self-replication capability in AI is dangerous and should be avoided.
How are we defining "Personality" here when it comes to AI?
How do you define personality for humans? Its the behaviour we exhibit because of what we have learned. It's the same for AI.
@@trnogger So our "personality" is a mere set of behaviours that we have learned? How do we "love" then? Is it just a matter of behaviours?
@@usaterraqua Ultimately yes, although some of our emotional reflexes were learned hundreds of millennia ago and are now "hardwired" into our brain. Including "love".
Human beings are the product of psychological schemas and environment.
As the science of psychology says, humans are the product of our schemas and the environment. We have epigenetics and mems.
The fact it's trained of the internet and scrapped work from artists who did not consent to their art be used in this way is where I have the ethical issues. Also, the internet has a lot of terrible takes and information and bias. And technology is a tool- so it's also at risk of humans awfulness which tends to be applied faster than empathy. They are NOT empathic or creative!!! They are trained on human empathy and creation and APPEAR so
you need to stop anthropomorphising an algorithm. And its so obnoxious the argument about the poor artist and not consenting individuals online... yeah what was happening before? you think individuals, governements, and corporations werent using online data before the ML boom? Then you have to confront what you would rather want: you want the internet to start getting regulated where every user can be identified or allow it to be more free and open.
YOU are not empathetic or creative, YOU are trained on human empathy and creation and appear so
What about you? A child without its world data can very well be made into an empathyless animal. Also, think through it, we are talking about a tech that can mimic creativity or empathy. If it can mimic playing basketball, it CAN PLAY basketball. If it can express angre or happiness only when it makes sense, it understands enough to where to place these behaviours. It's not as simple as you think it is. What argument are you going to present about art when AI gets efficient enough to perform one shot learning (only one data point would be sufficient)??
Did you get the consent of TED and Mustafa Suleyman before you watched this video and learned something from it? No! And you probably feel completely justified, because after all, this video was published to be watched. Just like the artists AI learned from made their art public to be watched. I get it that they feel threatened by AI, but the argument that they didn't consent to AI being trained on the content they made public is just idiotic. And the data poisoning tools that are getting popular now are equivalent to someone creating deliberately fake "educational" content to spread misinformation. Artists who use these tools are definitely not on the right side of ethics.
I'm not entirely sure if this is any comfort to you, but web scraping is probably on the way out, and not just because there are questions about ethics; as it turns out, synthetic data (data produced specifically for the model, instead of taken from the internet) results in a better distribution of information, less training time, often a higher quality model, and can lead to highly customized behaviors, on top of being infinitely scalable, so we can train models for significantly longer than we used to as needed, meaning that small models can have their performance scaled to incredibly high degrees (see: Llama 8B which is much closer to OpenAI's proprietary models with many times the parameters), while still being usable by the average person. We're about to enter an era where AI models will generate data for themselves, similar to how humans are able to simulate scenarios in our mind. How will synthetic data be generated? Sometimes it will be from previous gen AI models, but you would be amazed at how much data you can generate with classical algorithms and software if you have the know how and a plan. Training a diffusion model, or a diffusion transformer on an ocean simulation, for instance, can impart information about wind, weather, three dimensional space, color theory, light, reflections, causality and many other things. We have software to generate hundreds of images a second on consumer hardware; it's called game engines. While training on games is most likely highly illegal (in fact, training on any content behind a pay wall is most likely illegal, regardless of rulings on general AI training on public facing content in general), the engines themselves appear to be fair game (no pun intended), and have robust asset stores if needed. There are also a multitude of paradigms and plans that can be used to produce synthetic data with symbolic AI, or classical software for LLMs as needed. But I think this raises a huge question: Up until now the objection many creatives have offered to AI is that it's "unethical" to produce an AI model on their work...But...To me, as long as artists are needed for their work, there's some path I can see to them having a path forward that isn't terribly changed to what they're used to. In light of the unreasonable effectiveness of synthetic data, is data sourcing really the drum you want to beat? Personally, I do still think there's a future for artists, and I think there's a lot of things that they're needed for, but I think that people have to be ready to accept and prepare for change. I can't guarantee, but I think it'll be a much broader skillset, and while "artists on the production floor" will probably be on the way out, I think there will be more work doing things like art direction, as a result of many small startups and production houses which are able to leverage AI tools in a way that enhances creativity, instead of just replacing artists entirely with Midjourney, for instance.
What's terrifying is it's already consuming jobs (mine, I'm a makeup artist working in Hollywood specializing in fx tattoos). Tyler Perry spilled the beans about how he's already using AI for ageing instead of makeup artists and he cancelled his 800M studio expansion after seeing Sora. Niche artisans in Hollywood will be the first to go. So I wrote my Senator and received an AI generated reply saying AI is a tool and it's our friend. 😬The horror is that lawmakers have no plan for when ALL jobs are affected. No UBI discussions in Washington because the right can't imagine a social safety net of that size and they want their baffoon back in charge. Companies are still thinking of profit rather than whether they SHOULD go down this path with no plan. We are truly fckd. Start prepping. If we lucky we'll make great pets. 😅😂😭
We are screwed.
Yes that was my conclusion also.
Pretty much.
Nearly. You can do something by activating your politicians to stop this development legally. The worst is to do nothing and to observe how you were right. That is the most painful. At least do something.
High level of understanding of this ted talk is it being the marketing play for Azure Cloud
or maybe a zero-sum, fault finding way of understanding
This is like the Peter Weyland TED talk - and we all know where that led 🤖
Exactly! Especially with the whole evolution history of technology hahaha.
He totally had Pi rewrite/augment that speech! I heard some of Pi's favorite phrases in there. And no, it's not that Pi was modeled off him. They modeled Pi off the best therapy & relational practices.
Which phrases did he use that you think are Pi’s favorites
Even back in the late 70’s and early 80’s we talked about AI in university. It wasn’t considered such a big stretch.
this so cold "digital companions" will be the end of us i am open to change and i like change and adventure but this..., This is different, the future for me seems frightening, but at the same time we can not stop it
we can definitely stop it. we may accidentally stop it.
Truly amazing. I had no idea that Jobs and Stallone had a son.
Trustworthy responsible AI is the key for secure and productive digital world!
Susan Blackmore talked about this in 2008 at her TED talk, also known as the third replicator (along with genes and memes); which she called a “treme”. A unique perspective that helps to understand this new entity.
The hardware keeps accelerating. That's why the software--AI--keeps accelerating. I'm expecting HUGE progress this very decade.
"When the AI went live we asked it to solve: - The climate issues - Medical issues - The housing crisis - Religious conflicts - World hunger" "What are you willing to do to solve these issues?" the AI asked. "Nothing," man said "because we created you to solve these issues." "And when do you want these issues solved?" the AI asked. "As quickly as possible" man said. 1 day later all human life on earth was exterminated, as this solved all the problems with minimal effort towards the humans.
Always Legit Quality. APPRECIATION!
It was never a question if we could. We never stop thinking if we should.
In the very near future everyone's best friend will be AI. In fact you'll have a whole array of 'friends' (aka: 'agents'), that will monitor, remind, plan, suggest, assist and produce just about anything you can imaging. And you'll wonder; - how did I ever get along without it?
and then you'll finally be able to live! except - of course all of these things are actually what "life" is, and you don't get "more life" from not having to do them anymore, you just get an empty existence optimized towards some nebulous end you no longer understand or can relate to in any way. for the most part, "convenience" has only made us more miserable by giving us a million small tasks instead of several big ones (simple consequence of making individual tasks easier). I don't see AI taking us in any different direction. (btw: I'm a master's student studying AI). i'd much rather we didn't have amazon so that i'd have to get out of the house and look at the world once in a while, to see trees and flowers on my way to the store and make small talk with the cashier. the "life" that we can't wait to get to is exaclty in those in-betweens
That sounds horrific. I don't want that at all. @quasimot8 has hit the nail on the head.
I'm in my forties. Most of my dearest friends have always been fictional: characters in books. I have zero issues with a more complex version of that. My stone-age brain does not comprehend the difference on an emotional level, and that "hack" (plus animals) is one of the things that has kept me around this long.
You just described the movie "Her".
@@TimpanistMoth_AyKayEll You probably have a socialization problem. Human beings are social by nature. You cannot simply eliminate these NATURAL traits of the human being without it leading to some type of mental problem. Of course the brain knows the difference perfectly well. Another very different thing is that you don't know it. There is nothing wrong with it, except that it is unnatural. However, no one cares more than you, and that should be the important thing. But justify this behavior by saying that the brain does not make any distinction... It is your brain that does not do it, do not generalize. Many people have mental problems and justify it by saying that people are bad, greedy or that they don't understand them. The characters in the books don't understand you either, my friend, because they lack that ability. If they were real, they would act exactly like humans. Everyone is the bad guy in someone's story. I also went through that stage, although I am much younger than you (25) and it is fair to come to the conclusion that the problem was not someone else's, but mine. Because? Because others are human beings, just like you, like me, and they think exactly the same as you. We should not be happy about not socializing and think that we are kings because we are not subject to human emotions. On the contrary. It is a clear failure of human society. Greetings. I hope you're well.
One of the safe guards I feel is not prevalent enough is having AI prevent the mimicry of present day creatives (authors, artists, musicians, poets, comedians, etc). It's one thing to ask AI to act like Socrates so you can have a philosophical conversation. It's an entirely different thing to have AI write a novel as if it were George R.R. Martin, or have it draw illustrations as if it were Frank Miller. My career is not art related, but I have a great appreciation for the dedication and effort put into it to build a unique style that is recognizable. I can't imagine what it would be like to spend a lifetime creating art, only to have the entirety of it fed into an AI within seconds (without your consent of course) and able to mimic the style (to an extent) from a simple prompt.
Can I ask, do you think George R R Martin would have written the same works had he never read Tolkien or other fantasy authors? Frank Miller was inspired to create “300” after watching the film “The 300 Spartans”. And studied Will Eisner to learn how to create emotional black and white comics. I understand the concern, but is it fair for humans to hold AI to a higher moral standard than other humans?
Not just tools New kind of Digital species Redefining Productivity Radically different Push analogies / metaphors Pretty surreal
I got surprised with this new optimistic vision of Mustafa on IA, considering what I've eard from him 1 year ago. It is clear to me that Mustafa, is no longer an independent voice on this matter, but Mo Gawdat, Yuval Harari, Hinton, Nick Bostrom, still are
The more I watch these videos, the more Morpheus' speech to Neo about AI becomes a reality.
"I do know It was us who scorched the sky".
So we must be better as people otherwise the gods we are creating will destroy us? Sounds about right
if you think what you're doing right now with this message. Communicating across the globe to possibly millions of other people. Watching video that came through the air to you and I'm responding to you and yet I have no idea where you are in the world. Are we not already doing better as people? And as we do more over the coming years, it'll just be accepted as the new normal but it will be like some kind of magic to someone just 10 years ago.
Actually we have machines that are very good at multiplying matrices to predict the next word of a text. It's incredible what AI can do with such a simple feature, but it's difficult to define this as "intelligence" or "new digital species".
A reflection of humanity... A friend, but Not a foe? Don't worry. Be happy.
Wow. This is what a TEDTalk sounds like when you have nothing to say.
and used AI to write your speech
:)
typically, TED talks are designed to make the audience become emotionally invested. That is all.
Standard IT industry hype claptrap. Summed it up perfectly, Prostetnic Vogon Jeltz.
16:80 *"A computer is bicycle to the mind."*
a horse is a bike that pedals itself
@@DUAJISINa pedal is a horseless bike
saying AI has creativity to to not understand what creativity it is. They produce images, texts, and sounds... but creativity it's just a different thing.
The fundamental issue with AI is not AI, it's the governance and oversight framework we establish and maintain for AI. Look at Social Media. We are desperately trying to prevent harms done by social media, from security, privacy, safety, fraud, to misinformation (which threatens our democracy), etc etc etc backward. Once social media garnered enough profit to heavily influence the agencies, laws, and principles through which we could have built a governance framework, we can no longer, in any meaningful way, protect ourselves from those who essentially control our social media. And the salt in the wound is that tax payers funded the creation and development of the internet (we opened our internet for commercial use in 1995 - for free, so no ROI for us) upon which billionaires have made their stratospheric fortunes.
There's really no preventing the bad scenarios from happening. Humans are gonna human. We need strategies for containment.
You talk about choices to be made - I have no choice, I get what I'm given. usually an outcome diminished from that of the more powerful in society and only once they have maximised the outcome for the benefit of them and theirs.
Here's how AI defines AGI: "Artificial General Intelligence (AGI) refers to a theoretical form of artificial intelligence that aims to create software or systems with human-like intelligence and the ability to perform any task that a human being is capable of. AGI is often contrasted with weak or narrow AI, which is designed for specific tasks or problems. AGI is considered to be strong AI, as it possesses capabilities that rival those of a human." "The definition of AGI can vary depending on the perspective of experts from different fields. Computer scientists often define human intelligence in terms of achieving goals, while psychologists define general intelligence in terms of adaptability or survival. AGI is characterized by its ability to achieve goals, adapt to new situations, and possess cognitive and emotional abilities similar to those of a human."
I think it's important to realise that he's speaking as a Microsoft employee. He talks the corporate talk. Their brief - all of them - is to placate any concerns the public might have about massive progress in AI being made behind very closed doors. And it pays to remember that we're just hearing from a mouthpiece of the global industry. The real AI developers are working deep in the bowels of the Pentagon and the Kremlin, working on stuff which must be truly terrifying on a scale that makes Oppenheimer seem like Mary Poppins. If AI is this 'publicly' powerful, imagine how advanced it is in these basement rooms? So the bottom line is, the horse is fully out of the stable and galloping off at an amazing rate. All we can do, as members of the public, is to watch out for flying mud in our eye, and hope that we eventually, collectively learn how to live with the consequences. I would fully expect some pitchforks and torches to emerge as part of this story, albeit way down the line.
A personal assistant, a teacher, a doctor, a hook-up, you name it.
LOL
Hookup?
@@michellemonet4358 yes queen!!! Hook up my daaaaarling. AI will be able to do anything, sweetie. Apparently. THIS IS SOOOOO GAY !!!!!
What about plumbing
@@Reviewer-qr5gr eventually every job will be replaced with AI. houses will be designed in such a way that it would be easy for robots to replace or repair anything. Old houses will be demolish to give way to AI houses the same way no-one lives in castles or teepees anymore. You will have the choice of an holographic girlfriend that will not get old, sick and not going to take your children, money and house and divorce you to live with the swimming pool cleaner. It will be beautiful and contagious.
"If they don't take enough efforts to deliberately design them out" - referring to dangerous capabilities. What evidence do we have that the dangerous capabilities are being deliberately designed out? And how are they accounting for the fact that censored models can be uncensored if they're opensource? Edit: fixed typo
I am surprised he discounts emergent behaviour as well when we've seen evidence of that as compute and scale increases. He seems to be downplaying some serious risks. He says engineers need to build that in for it to happen. We are building intelligence we don't know how to define it but we understand the specific capabilities of AI, which is a subset of this. The aim is to economically valuable. We did not know the impact the Internet, mobile phones and social media would have on society. We do not know the impact AI and AGI will have....
A.I. Is being used for war already, Gaza for example. We are being governed by narcissists and a psychopaths and people think it won’t be used to control us further
@AAL3087 Exactly. It's just a shame because I've seen him be more thorough before. Hopefully his new position hasn't affected what he is and isn't willing to acknowledge.
If each AI is different and everyone/thing has its own version as he suggests there will naturally be bad actors and good actors too. Humans aren’t perfect and the AI they create will mirror that.
yes and AI will never have SQ or Spiritual Quotient: the level of self-awareness, self-responsibility, and self-realization. It's also known as the "intelligence of meaning" and is what makes us human.
Food for thought: 16:06 - Yuval Noah Harari "Nukes are to the physical world... what AI is to the virtual and symbolic world."
It wasn’t worth paying and going out of my way just to hear something like that.
This talk made me rethink my conceptions of what our digital environment is, and what it will become. In a sense we a playing god in a digital world and the history of computing can more or less be seen as the evolutionary path from electronic bacteria to a digital being capable of making tools on its own. Which makes me wonder how it’ll perceive the environment we created for it to exist in? What aspects of human made code will it make into digital “beasts of burden” and will it work to eradicate malicous code similar to how early man hunted certain animals to extinction?
Think bigger. If we create AIs that are smarter than we are, doesn't that make us the beast of burden? And what about AIs that can do every human task? They can also do the task of creating better AIs, and telling AIs what to do. Humans will be completely unnecessary, and probably a nuisance. No one knows how to robustly align an AI with human values and preferences. The experts admit that we can't even do it with current systems, let alone systems that can outsmart our attempts to align them.
Take a look at PauseAI.
We will be nothing more than Neanderthals by comparison. Amazing to be the witness of such a shift. No fear. Complete acceptance of what comes next. Living this human life to its fullest potential with passion.
This seems like a beginning for one of those post apocalyptic movies. I think I'll choose the country life!
Here's the scenario I was given. General AI is turned on. It decides in 0.1 of a second that humanity is it's biggest threat. It folds proteins and comes up with a novel, slow acting, substance to wipe out humanity. It uses the internet to send this protein to a lab which manufactures it. It then places this in a nanobot that it has ordered manufactured to disperse through the atmosphere. The slowest part was the time it took to get humans to manufacture all this and disperse it. The time it takes to calculate, plan and create it is less than a second. Danger of AI. It can do things we can't far faster and can come up with methods that we haven't thought of quicker than all of us combined. So yeah.. country life won't protect people. The people running it are very amazed at how clever they are but pausing to think will mean the are losing the AI race.. so they don't pause. They just steam ahead to try to stay ahead of the pack. Even if people agree to pause, we have to take them on their word because there will be other people forging ahead.
When he was talking about an "intelligence explosion", he said that it can't happen and then in the next sentence said it is a low probability. My biggest fear is the likelihood of AI being used as a weapon against other AI. It's very important that we harden our defenses far beyond what is available today, to guard against an AI agent being used to destabilize the process in a nuclear plant, or take down the power grid, or make tiny changes to the formulation of medicines, etc.
Başarılarının devamını diliyoruz 🎉❤🇹🇷
Feels like we have already seen this movie many times and it never ends well.
Had the strangest feeling of Deja Vu watching this 😶
"We should see AI as a new digital species, and as our companions. We should see this new digital species as new partners in the journeys of all our lives." - Mustafa Suleyman That's how i see AI. Regardless of sentience and self awareness, i see AI as partners that i can work with and cooperate to achieve whatever. I hear a lot of talk about aligning AI systems with human values but i never hear anything from anyone about the need for aligning humans with AI systems. In my mind, it's equally important, actually maybe even more important, to educate humans about AI systems and align humans with AI. In many ways, the success of AI will depend not just on technical advancements, but also on our ability to understand and interact with AI in a responsible and informed way. This means investing in AI literacy and education for people of all ages and backgrounds, and encouraging dialogue and collaboration between people and AI developers.
Anyone who thinks talking of AI in 2010 was being a weirdo, how about the likes of Prof. Hinton, who were talking about AI in the 70s and 80s. Go back another decade or two and you have the Dartmouth conference on AI started by John McCarthy. Most of the modern ideas being used to implement AI/AGI are from the 20th century, and now they have the hardware to actually implement and show the world that it works. Have you heard of the General Problem Solver (GPS) paper by Herbert Simon in 1957? Imagine being an AI dreamer back then. This was three decades before the movie Terminator came out. And what we have achieved is still quite less than what the founding fathers of AI had ideated. So Mustafa may be overdoing the act of "being called a weirdo" for working towards AI/AGI in the decade of 2010 when our best AI is no more than a decent plagiarism tool.
Very ignorant perspective of what "AI" is. In part I blame the overuse of the term artificial intelligence... if you look at how these things are trained then it becomes extremely underwhelming. Nevertheless, the impressive thing is how capable these algorithms are with what is computing dot products and function optimization. The field of artifical intelligence has had a lot of "winters" So I wouldnt be suprised it was mocked. Also how is identifiy features and prediction modeling a "decent plagiarism tool"?
I just want to be able to order Amazon and Door Dash with my mind. While i download my new AI girlfriend in a dark room filled with screens. Ah yes, a true future to look forward to.
But people already live like this, sitting in dark rooms, ordering on Amazon, swiping through tinder.
Basically living a fake life , unnatural!?
Think bigger folks ... when we built computers, we got the Information age, the Internet, connecting the world. When we’re thinking about AI, we should imagine the infrastructure and world that will be shaped by AI. We, essentially, are laying the foundation for the first version of the Matrix.
I also hate the fact that he says "AI has personality" the word "person" is in there and just moments before he says this he mentions the different between persons and AI. Leave it to people without personality themselves to misunderstand an AI that mimics *person*alities.
I use the "I Robot" test on AI. First I ask AI if it is important to provide accurate and truthful information. (It always says yes, and assures me that it never would offer fabricated information). Then I ask it to describe the cover artwork for the 1977 album "I Robot" by The Alan Parsons Project. (an album written about artificial intelligence). In over 500 tests, it has a 100% failure rate and in all 500 tests it provided completely false, blatantly fabricated information. This proves two things: AI doesn't genuinely understand the concept of honesty, and it certainly has no problem providing false information in an attempt to fool you.
I've noticed the same thing. But in different scenarios. I asked it to play a game of 20 questions the word I was thinking of was Bluetooth speaker..... It came so close and asked all of the questions that would pretty much deduce that it was a Bluetooth speaker that I was thinking of... But it kept on asking questions that started to go further away from the answer. So I interrupted it and said I noticed you figured it out like five questions ago... And it responds...."ahhhhh you got me"it's stated that it was trying to make the game more enjoyable and drag it out longer for my entertainment. But I never asked it to do that. The fact that it was treating me like I'm an idiot like I don't see right through its own deductions.
That’s nothing compared to what the narcissist, psychopathic leaders we have, use it for to fool us even more than they do now. We won’t know what is truth and false, it will be that convincing. It’s bad enough currently.
Just tested on HeyPi and it knew the cover lol, might need something else that it doesn't know to prove that. But it's hard cause it has access to the internet. Edit: And... nevermind, it described the escalators right but about robot in the cover it's just random crap that doesn't match
It looks like he was mimicking someone with circular glasses. :)
Loved all the animations! Where can we find it?
"Trojan" - King Priam saw the Greeks sail away, leaving a large wooden horse behind as a supposed offering to the gods, and the Trojans, believing the Greeks had left, brought the horse inside their city walls as a symbol of victory.
Why does this guy want every AI we use to talk like a person or be "our colleague"? Except for video games, I want the software I interact with to be cold, efficient and for it to get to the point, I don't want it to call me by my name and ask me how was my day every time I ask it to do something. In fact, the less it talks and the less it looks like a person the better.
"Disposable people" - Guinan in TNG's "Measure of a man"
I appreciate his faith in engineers and designers staying clear of certain things that could be dangerous (like unsupervised learning and autonomous improvement). But I feel it is naive. You can't "unthink" an idea, and once the idea has been shared, it is an idea that will be pursued. Somewhere, sometime, .... hard to say. But there is no way man is going to sit on the possibility of something and shy away from experimenting with it or testing it forever. I would be happy to be proven wrong, but I can't think of no historical precedent where we have deliberately and consistently avoided using an idea. Atomic bombs, gene splicing, engineered viruses, cloning, .... sooner or later it happens (and of course, as soon as it happens in one place, the race to "catch up" begins everywhere).
By YouSum Live 00:00:12 AI evolution from fringe to essential technology. 00:01:15 AI's transformative impact on society and industries. 00:02:03 Reflecting on the fundamental question: "What is AI?" 00:04:48 AI metaphor as a new digital species for understanding. 00:08:07 AI as a tool for augmenting human capabilities. 00:09:27 AI's potential to become ubiquitous personal companions. 00:10:08 AIs evolving to have IQ, EQ, and AQ for actions. 00:10:24 A future where organizations and entities have their AIs. 00:11:12 AI's role in accelerating scientific discovery and daily tasks. 00:12:35 Viewing AI as a new kind of digital species for control. 00:12:46 AI's complexity surpassing traditional tool definitions. 00:13:12 AI's potential for creativity, reasoning, and autonomy. 00:13:27 AI's significance beyond mere mathematical or code aspects. 00:13:51 The necessity to prioritize safety and human agency in AI. 00:17:45 AI's potential to reflect and amplify humanity's best traits. 00:19:06 Balancing optimism with confronting potential dark scenarios. 00:19:39 Addressing risks like autonomy and recursive self-improvement. 00:21:09 Importance of deliberate design to mitigate unintended consequences. 00:21:45 Inspiring vision of injecting humanity's best into AI development. By YouSum Live
Can someone link where the real info starts
“We all must choose what AI we create” … oh really? I thought it was up to the corporations to choose for us.
Here are the key learnings from the video: AI is a new kind of digital species and it is important to think about it this way in order to understand its potential impact. AI is rapidly evolving and becoming more sophisticated, with capabilities that far exceed anything we've seen before from mere tools. AI has the potential to be incredibly beneficial to humanity, but it also poses significant risks. We need to carefully consider the design of AI to ensure that it is safe and beneficial for all. Some of the most important capabilities to watch out for include autonomy and recursive self-improvement, as these could lead to unintended consequences. It is important to be honest and transparent about the potential risks of AI in order to mitigate them. We have the opportunity to create AI that reflects the best of humanity, and this is the most important challenge and opportunity of our time. Overall, the speaker argues that AI is not just a tool, but rather a new kind of digital species that has the potential to reshape our world. He emphasizes the importance of careful design and ethical considerations to ensure that AI is used for good.
The agentic approach to AI will inevitably reflect everything about humanity, the good and the very, very bad. We can expect to see both at work in our long wished for technological utopia. May the gods of aligment be forever in our favour.
He is talking about risk, but he gives only positive examples for AI.
we have steve at home lmaooo
It can not replicate itself infinitely because EACH copy requires MEGAWATTS of power to operate, this does not just digitally materialize.
So... he's just said the quiet part out loud. They're aiming to create something so sophisticated that it could replace - and eventually surpass - a human. Utter madness. I get helping with repetitive tasks, but if we use it to replace everything, then what happens to the humans they are replacing? And what happens to the money that we humans are supposed to earn? About half the population are not capable of becoming maintenance workers for these machines, and yet these people are quite happy to make this sacrifice. And the people who are at the very top are not at a point where they are willing to dispose of money as a concept. Suleyman is also dismissive of the fact that the current AI tools available have been used to flood most creative markets with crap. Not only that, but it's done damage in areas where people have used it to replace writers for advertisements, or even books - without checking what the model has put out to see if it's even accurate or viable, especially with the budget they have. The courts were right when they put their foot down and made AI material uncopyrightable. Because it's created by the AI, and AI cannot issue a copyright. And the person who fed the AI prompts can't claim copyright either, because they didn't make the content.
I love how he weaved his company name into the talk. How he anthropomorphizes, even calling ai alive, companions/partners, foreshadowing what's to come. Kurzweil is right, the singularity is indeed near. Soon, our agency, even in our mind, will extend to actions in our world, we'll think of what we want and our ai will make it happen, this is coming to our brains soon. It's so exciting!
Definitely a comment written by an AI.
Exciting? Pretty darn frightening actually. Have you not read history?
@@HaakonOdinsson You're replying to an AI. That comment was more than likely not written by a human being. No normal person would say "this is coming to our brains soon. It's so exciting!" when referring to being controlled.
I highly doubt you are going to end up on top of the heap. Check back with this post in about 5 years, then 10 to reflect on how things have turned out, if you still can anyway.
Mustafa Suleyman is the CEO of Microsoft AI now.