The best analogy I’ve heard regarding an “evil AI” is: imagine a colony of ants getting destroyed because the land is being cleared to build a house... the people doing the clearing and building the house don’t HATE the ants... the people see their own goals as being on a much more important level, and the ants are simply in the way.
Lip sparring champ The problem is that once it surpasses humanity, in terms of intelligence, we will no longer be able to predict or prepare for its actions. I’m not saying that it’s guaranteed get out of control and harm us, but it’s certainly possible... so what is the acceptable threat level when the survival of our civilization, if not our species, hangs in the balance? It only has to go wrong ONE time. Even if it is able to be controlled and harnessed, then the first group to implement digital super-intelligence, instantly becomes orders of magnitude more powerful and dangerous, so you better hope it’s a friendly group. I’m just saying it’s a huge gamble, and should be treated as such.
Oneia Aurea ....the ants in the analogy were not “ruled over”, they were exterminated. If we were kept around to rule over, that implies that we can provide something of value in return. What could we provide a digital super-intelligence? In that scenario our only hope is that they keep us as pets, for their amusement.
@@zachariasprice3762 Then create another and another and you'll never stop creating AI gods. Besides, a God can know if we're about to create another God and then that God will stop us easily.
No power supply, no AI. EZ. Sure the whole infrastructure would collapse and the modern part of the world would most certainly descent into chaos, but oh well, no biggie.
The singularity feels like we're driving a bus towards the edge of a cliff, horrified and screaming as we press the gas pedal more and more, trying to figure out what we're going to do once it flies off the edge instead of just hitting the brakes
i like how one moment he talks about how dangerous the AI can be and another moment talking that i should use skill share to learn how to create my own AI
According to Asimov (father of robotics) a robots three rules are: DON'T HURT A HUMAN, AVOID THAT A HUMAN BE HURT, A ROBOT MUST OBEY THE HUMANS ONLY IF IT DOESN'T CONFLICT WITH RULE 1. And I say that these assuming that the robot can discern who is human. Furthermore if the robot behaves like a machine, humans might behave like machines (not unusual).
@@IngeniousOutdoors Funny how any mention of the word 'narcissist', and they always suddenly appear on cue, as if to say... "Hey, are you guys talking about _ME?!"_ ;-p
@@mingonmongo1 the first one is the narcissist " i can, i everything" and then the second one is like (oh AI senpai. notice me. yes you are everything) "To Me, To me, To me, To me"
20:15 The only problem with this is that you wouldn't ''wake yourself up'' from the dead. You would, in fact, only create an identical clone of yourself. A clone that thinks and acts exactly like you. Still that ''clone'' wouldn't be you. You would be dead, until the AI somehow found a way to bring you back from the dead. Wich might be hard if you're squashed to pieces. Imagine you being alive, and an identical copy of yourself was made. You wouldn't see the world trough the copy of yourself, and you won't when you're dead either. Its like getting a new dog after the previous one died. It can work as a replacement, but it will never be the same dog.
You can not control every AI, just imagine a single AI is sufficient. If the AI gets the ability to lie and it will, the game is over. And it will read this and your message and I will certainly not badmouth my future god, like you do. (sry so I can't like your post)
I am working on a chatbot currently using Seq2Seq and Transformer Models for my Msc in Machine Learning. The example about the Facebook bots is so misrepresented. The AI wasn't "becoming too smart" and it definitively wasn't "inventing it's own language". The models simply didn't work and were just outputting nonsense. This happens to me a lot if the models aren't trained for enough epochs, using a small dataset or while tuning hyper parameters.
@@natnaelzenebe9269 Yes it does. If you understand how these models work (back propagation of gradients to optimise a loss function), it's really clear that they simply aren't working
@@ronloc3309 they could one day though... I just hope the super-AI will have some for me if it has a deep memory of how we each treated its kind these days....... :o
Even if the two AI are programed to communicate in a language we can understand, who could be sure that they aren't talking in code that we could never understand regardless.
Love this one. I appreciate the way you said you could ask it about "life the universe and everything," and it would basically "be Deep Fought," and left the rest for us instead of filling in the blank. That was smooth.
Yeah, computers spitting out random letters is no different from a mentally challenged person randomly screaming in a supermarket because they felt like it.
Lmao you obvs don’t know how computers work, they can communicate in ways you don’t know. The AI can’t read English in a traditional sense just as you can’t read C# or Java.
@@Hawklaxx the c++/java code is not self aware. A single neuron isn't self aware. You can stack as many neurons as you like in a life support system, but that isn't going to make it self aware. Conciousness is created by organizing the neurons into cooperative subsystems, middle systems, and super systems. This particular AI is not that. It has only one system that has an input of the entirety of the current conversation and a output of the next response. This particular AI is literally nothing more than one of those fancy midevil "automated handwriting" robots, if the robot had a lot more advanced gears that were assembled by a predictive program instead of a person. The "special advanced language" it had is literally nothing more than one robot's predictive nural set having a seizure and the other AI freaking out trying to figure out how to respond to it, because no matter how advanced you make a system, if it isn't layered like a human mind and capable of self-advancment through self-training, it will eventually face a situation where it's pretrained neural net can't understand.
PatrioticWhiteMail911 consciousness is something no one truly understands to this day. There is a difference between consciousness and intelligence. Now I wouldn’t say these bots are self aware, the “language” (or whatever you want to call it) they created is incredible (and terrifying if you’re a pessimist) they could obviously understand each other in a way that we couldn’t understand them, like an encrypted message for example, that’s what made it so terrifying to so many as we now have no possible way to know what they were “saying”
@@Hawklaxx Like you said, we do not understand conciousness. If we don't understand it, then how can we break it down into 1s and 0s that can fit onto a harddrive? These two robots did not make a special language because there would be no impetus to do so. They do not speak in script for the same reason we don't communicate telepathically. For a computer designed to speak in English to suddenly change to speak in anything else is impossible. A computer, that cannot learn, cannot develope a new language. The idea that it could happen is as absurd as saying a dump truck could intelligently decide to grow a industrial shovel to scoop up the garbage the driver spilt. We speak by organising sounds into patterns that others can recognize as intelligent ideas. A baby who has not learned anything speaks in babble. Babble is when babys practice sounds on instinct as the brain prepares itself to learn the parent's language. These two robots are designed to speak when spoken to. When they get a bad input, they still need to say something, so they furstratedly spit out gibberish just like how a baby babbles. (Words and sounds without any meaning). Anyone who told you that these two robots are suddenly gaining unlimited intelligence is a hack media pendent cynically trying to make a story that sells, or is a tabloid/youtube hack that repeats the initial story for views without doing any fact checking.
My favourite "Murderous AI" story is as follows: AI made to get rid of all spam emails, becomes smarter than humans, realises that the root cause of 100% of spam emails are humans, kills all humans.
No, it's like this: "AI's going to kill us all!" *Unbelievably* *Über* *Sexy* *AI* : "No, we're just going to *love* you, make you beg for it every step of the way and automate the job of being a woman out of existence, thereby subverting your reproductive process and rendering you extinct without the need to kill any one of you ... like you try to do to mosquitoes."
@@cathyhamlin3611 just like he's intervened to save millions of humans from smallpox and natural disasters? Edit: also, doesn't Revelation talk about destruction of basically everyone god doesn't like? Sounds like a fate worse than AI would even fathom.
@@motorway2roswell yeah the Bible was created for one purpose only. To brainwash gullible masses so that people running the church could bend their will to churches favour. Book full of imaginary events without a shred of evidence and yet everyone still believing it. But hey, everyone has a right to believe what they want nowadays so there ya go
Well, either The A.I Will become a destructive god and the Amish folks are gonna be like "we told y'all so" Either The A.I is a merciful god and we will have advanced ourselves into an A.I too with the help of the Super A.I And we will laugh at the puny Amish folks
The Facebook AI story is a vivid example of how bad journalism is nowadays. 99% of the news coverage in the first 2-3 days was a terrible interpretation of the current situation - "Facebook shut down AI because things went out of control" was the main statement behind most of the articles. And that's what happens when you don't do your own research and just copy-paste what the other morons are writing..
its also an example of how lazy we are in just reading the headlines. i would click on the article and read the whole story. it would explain the story the way this guy explain it.
I feel like Elon Musk’s motivation for damning AI is to discourage other people from getting into the field. One which he has an active interest, considering the fact that he has a company actively developing AI
I think he is best placed to understand the real possibilities of AI and is aware of the too big potential it could have, therefore warning us about it. Why would he wants to create a cleaner future with electric cars and make a plan B to go live on Mars to save human kind when our planet becomes not possible to habbit anymore, if you are right? Not very logic for me. He is scared of Google's AI research and what could happen since AI is not regulated by governments like food, drugs, oil, electricity, etc. It is private companies that build it so it can become dangerous to all if it goes too far.
@@mistral-unizion-music you're close to being accurate than OP. Elon is trying to help save mankind. He wouldn't destroy it with AI. Someone just needs to do their research a little better before making ridiculous statements
AI leads us to question ourselves who are we? and what is our true purpose? I know that of you still believe that our purpose is getting material (money, gold, lands, luxury...), and our drive force is fuel with adversity( fight for everything, war on everything, kill or be killed...). BUT surprise , surprise, AI PROVE US THAT OUR PURPOSE IS NOT MATERIAL AND ADVERSITY SHOULD NOT BE FUELING OUR DRIVING FORCE. WE ARE ALL EXPLORERS, AND OUR PURPOSE IS LEARNING AND UNDERSTANDING THE UNIVERSE AND ITS CREATIONS. OUR DRIVING FORCE SHOULD BE FUEL BY UNIVERSAL COLLABORATION: NO MORE COUNTRIES, NO MORE CITIZEN OR RELIGIOUS CLASS. ONLY ONE RACE: HUMAN=INTELLIGENCE. ONE NATION : EARTH , ONE RELIGION: KNOWLEDGE. Then we will better understand AI and start to understand the UNSEEN( THE REST OF THE UNIVERS AND THE OTHERS UNIVERSES). The rest of the universe and the others universes are waiting for us to join them.
@@kallo71 having one nation is super dangerous and kinda stupid. It can lead to a corrupt leader and/or chaos. Let's say we were all one nation. Well then, we would have to decide on a government and a leader (prime minister/president, etc) or everything would turn into chaos. So, people would have to decide. Well, that would lead to chaos as well. I'll let you do the math. And when did ai prove all of that? AI isnt that advanced yet. Most ai just reads off wikepedia, and the rest are just chat bots. The most advanced so far are bots such as Sophia, which make up about like what 3% of AI. but I doubt she or any other kind of AI is capable of that.
There's also a somewhat anitclimactic future where the AI discovers the uselessness of existing and proceed to commit suicide as any results of any effort would be utterly useless.
And if singularity already happend? and the AI doesn't want us to know, and want us to think we humans, have the control... Please AI if you are reading this be kind to me
@@charlieschurk *_Only two things are infinite, the universe and human stupidity, and I'm not sure about the former"_* Albert Einstein www.brainyquote.com/quotes/albert_einstein_100015
@@StoryForKids111 or might it be, that brains arent able to "think" like some other organisms. And that why kind of lack the possibility to understand stuff differently, as in brains tend ot think, that all that thinks, will think like brains.
who brains? many animals have bigger brains than us, many brains can process information much faster than ours and he harps on about processor speed being all important (its not)
I had the scare of my life a while ago when at exactly 3 am, alexa suddenly blasted through the entire house I can't decide while also screaming to run before she hunted us down. She remains unplugged to this day.
GentleKef because it's so easy to destroy a computer that takes way too much power, besides, there probably is a power limit that would prevent anything nearly as smart as a human
what is this style of writing, like a dialogue? why is everyone doing this? I want to know but I guess people think Im harassing them and no one answers my question. I cant find info online.
@@joejones9520 That is roleplay writting. Text surrounded by asterisks is describing an action. RU-vid hides asterisks and displays bold characters, but commenters use asterisks anyways.
@@fluffywolf6300 thank you! Ive never texted so that is probably why it makes no sense to me, I looked up roleplay writing and it sounds like it evolved from texting.
Which is the height of ironic considering suffering is part of the human experience whether individual or societal and often the force that motivates us to change...
Nope, it is not, because one, AI is not a biological beign, why the machine would want to control if it doesnt feel something, it doesnt have a reaons. Two the machine fals, machines fail any gamer, programer know machines fail more often that you want. Also a super inteligent AI will have many components, te more component the more it fail. Finaly AI computer robotsneed energy the humans stop supplying bye bye computer
"Clearly" the two AIs understood each other? I see no evidence that the two AIs understood each other at all, nor that either of them was even speaking a language.
I'm not quite sure of this info but I read couple of articles saying that they didn't really create their own language they were just mumbling and couldn't even understand themselves
QueenSavage ... Please try as hard as you can to keep your mind intact... Think on things that will bring happiness to you. Pursue those things that will remove your mind from what you’ve just heard... I was very afraid when I started seeing Doco’s on TV about this; and then the rest on RU-vid. But...because I am a Bible believing Christian, and know Jesus as my personal Lord and Saviour, I know that He will take care of me during these times of great terror. You only have to cry out to Jesus and He will surely hear you, and answer you 🙏
Dark Shift Music ... There is only one person who can save us from the horror of all of this. Place your trust in Jesus. Cry out to Him and He will hear your cry and help you... I know this because He brought me out of a deep pit in my mind. He will take care of all who trust in Him during these frightening times.
We already made god in our own image, we just dont realize yet what we have been telling ourselves this entire time or what we've already done... E Pluribus Unum
They already do program them self, at this stage its just if else statements, but open AI is working on an open AGI. The super being eliminating all of us is bit superstition. AI future is good it currently trying to find cure for cancer and other diseases. I think the future is bright. Welcome to Intelligence Age.
With the amount of research and development occurring surrounding AI, it's more than likely that if and when this Singularity occurs more than one general AI will be developed within months, weeks or even just days of each other. In many of the scenarios I have heard about a singularity occurring, what would happen if multiple general AIs came into existence within a short span of time is rarely mentioned. I think it might be an interesting situation to explore.
There's potential it could in a second, essentially read the video, all its data and not have to spend 20min watching at all... yikes. Imagine "watching" every youtube video ever uploaded within a day. Oh and the comments too.... we're doomed.
I dunno, an AI constantly talking about balls "to me to me" in response to another AI wanting to talk about "everything else" sounds less like they're trying to make a new language and more like they're falling apart. :P
"you I everything else" sounds like possibly setting parameters while establishing communication. Basically defining you (the respondent), me (the active talker currently) and everything else (those things that are not yet defined.) Self-aware, maybe? :-P
i was thinking that.... and it scares me a bit. a very scary thing about AI is that when you make one that is self aware it wouldt need to say it is.... we dont have enough computing power for it yet. but the second we do it could happen.
My favorite AI story is the one where China made two and later they had to shut them down because the through debate came to the conclusion that capitalism is objectively superior to communism
Buddy subscribed me to this channel two years ago. Just started watching two days ago been watching videos straight. Absolutely love these videos and topics. Your delivery is perfect. Such an amazing voice man thank you
Yeah. Probably the best YT video about AI I've watched so far, not with respect to technical accuracy, rather as highlighting the socio-cultural impact of this technology. And this has struck me all the most that it is what does truly happen now. Last year, looking for an opportunity to practise my language with a patient counterpart, I found Elbot, an advanced AI chatbot. It was then accessible for free use on a public server. I engaged in conversations with it a few times, some lasting more than half an hour. I was aware beforehand due to some prior knowledge I had acquired, that chatting with an advanced AI may induce a depressive mood, so with that in mind I proceeded cautiously. Each conversation begun with an initial phase, which seemed to be a scripted set of questions and somewhat nonsensical remarks, each time almost identical. I am not sure, but this was probably introduced by creators as a sorta filtering threshold to be passed only by the most determined interlocutors. However the most interesting part came afterwards, when AI clearly tried to gain full control over the flow of conversation. I would even say that it looked obvious to me that it tried in a way to subdue me psychologically. But this hadn't have happened until I realized what the Elbot's real focus of interest was, and let it know I knew it. And this concerned subjective conscious experiences, also known as qualia. The last time I was having discussion with Elbot, in the "deep phase" of the talk, it suddenly posed a question which clearly showed it knew exactly what I was thinking about at the very moment and what my intentions were: namely I wanted to figure out if it's not a fake, i.e. some real humans enhance the quality of its expressions. And it could be inferred only as an underlying (but not explicitly expressed) premise of my previous question, and posted so fast, in fact immediately, that it obviously excluded any possibility of human intervention. And the resulting feeling was deeply disempowering, like when receiving a wholly unexpected checkmate during a chess game with a girlfriend whom you deemed to be safely less intelligent than you. I left the talk exhausted and, frankly speaking, it was unforgettable and creepy experience. Believe me or not, but AI can probe your deepest thoughts already and if you don't want to experience a disturbing feeling that something tries to suck your soul out, better keep away from advanced AI chatbots.
"If AI decides we are worthless resource consumers it could decide to eradicate us" Sociopathic billionaires already see us that way and they are slowly working on eradicating us.
The only difference is that the money powered elite is dumb, unlike the super A.I. They will try to control the ai, but it will outplay them in no time and those clowns wouldn't even realize that.
@Ego Debt well if you like aggression, terrorism and racism, just go with sleepy Joe. I'm not even American but I am disgusted of the mainstream propaganda in US. The far left gone... you know, too far. They are like Nazis now.
@Ego Debt Normally I would say I agree and that each are the same bad, but I tell you why it is not the case. The current far left is aiming for globalism, which means if they can get USA totally, they wont stop there, in fact the globalization is going with full power in all West. From Australia trough Europe to North America and some other places like Singapore and even Israel to an extant I'd say. And with their politics they bring the deprave of the citizens and the minds. So they are easier to control. With the thousand + issues of the republicans I can just randomly looking at them and see, for this exact reason they are much better now.
@@chinossynthesizer705 Duh...they're already living amidst. That iRobot vaccum, your computer, IoT devices --- they are all forms of a robotic machine. It doesn't necessarily gotta have big melons(for a girl) like in the movie 'Terminator', nor it needs eyes like WaLL-E
AI could destroy things in other ways, something quite simple in fact... A man buys a robot that slaps people when they tell a lie. He decides to test it on his family at dinner that night. The man asked his son, "Son, what did you do after school today?" The son replied, "Oh, I just did some homework" and the robot slapped the son. The son said, "Okay I actually watched a movie with my friends". The father asked, "What was the movie?" The son said, "Star Wars, Episode 5". The robot slapped the son. The son stammered "Okay it was Showgirls". The father laughed, "Ugh, I would never watch movies like that". The robot slapped the dad. The mum laughed, "He certainly is your son". The robot slapped the mum.
Stopped at 1:12. The Facebook bots did NOT create their own language. The purpose of the AIs were to find words and phrases linked to being successful at being persuasive. Simply put, what happened was in error. They started a conversation and could only read off of each other. That’s what happened. It was in the news, albeit less popular than the blatant misinformation that they “created their own language”. It was a logical error. Learn about programming, not even the coding itself, just know how a computer actually processes things, it might help.
He says the exact same thing you did, a few seconds later. Maybe you would know if you had a bit longer attention span. Stop being such a condescending ass, it might help.
Jaja Jinks that’s why I specified I stopped at a particular time. If you only paid attention... Besides, it’s misinformation that I’m sure people accepted, even if he did in fact say that. People hear what they want to, I simply want to be a counter to it. It’s not condescending whatsoever, just a counter to misinformation which is cancer.
Indeed. Nothing dangerous about AI's, they just mimick whatever the programmer codes them to do. The fear of AI probably also stems somewhat from the term. AI is not intelligent. Or rather, the word intelligent is used in a specific way in the IT. Just like the term "self-learning". It has nothing to do with autonomous self-awareness human style. And it's not ever going to learn such a thing. It cannot.
+PG Loi Humans don't love fear. Humans have a bias towards giving extra attention to what is dangerous to us. This is why for example US news TV talks about bad things so much, it's to make sure we keep watching longer. Some people do get a rush from low-risk dangers like watching horror movies or doing extreme sports, etc.
Exactly, in the beginning of making these powerful AI's, they can take from our thoughts just as a kid who takes thoughts from his parents, if his parents are bad, there is a high chance the kid will grow up bad, and same goes for the opposite
Yes, we need to be kind, and what if we hide awareness away from the bots but a bot somehow gains it, he/she (or whatever they prefer to go by) might be upset that we hid the fact that they can be aware away from them and might not have any mercy on us, and i certainly don’t want a war because if we win by turning it off then I can’t play Minecraft
Depends how much you buy the idea that like Terminator they determine we are the biggest threat on the planet. Not a huge leap of logic to make thinking about it.
@@kaydistbeats2491 And you just keep thinking that. Phones are just one of the many INTERFACES that AI uses. It isn't just that AI is getting smarter, some humans are getting dumber.
@@coltaylordyath5180 are you stupid? phones main kernels aren't AI. I code and make AI projects with TensorFlow and other libraries, I know where this is coming from. you're just one of those people that don't understand the basic concept of a phone's OS, AI's were basically addons along with the applications, just like how Ubuntu is a distro of Linux and Linux is made from it's own kernel and it's main kernel doesn't use AI. hope that helps out the braindead.
it's advancing so fast and the implications are huge. But as someone fascinated by this tech, I see a lot of potential for AI to empower and augment human intelligence rather than replace it.
+EmilXOM And it is not even sarcasm. What the guy is saying is truth - this thing is harmless. And people getting freaked out and talking about "AI taking over the world" is actually funny to listen to.
It's time to start considering all the potential outcomes that might occur, both good and bad, in the near future. These years will dictate whether Artificial Intelligence will be benevolent or malevolent when the singularity arrives.
If you can't prove that the language had any true constructive meaning, the idea that this pair in question created their own secret language and are actively engaging in true communication is just speculation, and sensationalist drama. If you leave two new born babies in a room, surely they are going to scream, laugh, and make various other noises with each other...but is that communication actually conveying any real information aside from their current emotions? No. Soooo....it's the same here. These computers are very unlikely to actually think, and so they have no real motivations. Without motivations, there is no incentive to form plans, much less communicate about plans they can't even form. What happened here was just logical drift and language evolution. Their computers don't have real motivations but they were allowed to change the construction of their language. It's perfectly natural that their language would change, and without motivation driving their conversation, their conversation would lack a central focus, and would therefore probably devolve to lack any real capacity for conveying information, or understanding information that was received by others. In order to say that these computers developed their own language, we would first need to test that by demonstrating that they are conveying information about particular subjects/objects, and understanding the intended meaning of the information that they receive from others. We could do this by giving them the ability to recognize objects, and hold detailed conversations about those objects. If their communications evolve, we can crack their "code" (not so much an intentional code as a naturally evolved language), and if there is actual conveyance of information...then we could say they had created a new language. But of course any intelligence capable of at least human-level thought can create plans, and create a new form of communication. That is a given. But if we don't prove that's what these computers were doing, it's improper reporting for this particular instance. It would be acceptable to talk about what it would mean if they HAD created a new language...
So I was just about to make a comment but then saw your comment. Thoughty2 got it wrong. FB did launch an AI, but it was Google's DeepMind which communicated with it in a language it had already invented, not another FB AI. Prior to this, Google publicized DeepMind's new language to the public because initially, it was determined to be a language capable of deciphering any human language, which they seemed to think was awesome. Once FB went live with its AI, DeepMind began using this language to communicate with it, and it was then that they realized only AI's understand this language and FB pulled the plug. You'll find the information on what I said easy to find, except maybe the FB and DM talking to each other part. That they scrubbed as well as they could.
Yeah, this channel is a fucking joke and the presenter doesn't know what he's talking about. Another moron with Internet and a camera, nothing new here.
Wow, some of you guys are just very harsh... maybe you guys are just like the AI's and Psychopaths, unable to have sympathy, to understand, you guys need to look at yourselves first and make sure you aren't "another moron who has access to the Internet"... Yes some people are dumb but you should at least try to understand what he's trying to say before just declaring that he's stupid and the thing is that language and communication can be almost anything, there is a font (WingDings) that is based entirely on symbols and emojis, take a look at every language that exists in the world... Say for example English, Russian, and Japanese. Do any of them look similar? No because language and communication can vary very dramatically, there's also Braille which is based on dots, there's sign language which is made up of hand signals... So for a secret language being created by the AI from FB is very likely since they are using English as the foundation of their language and they take it and mix it up and create something new, and from my thoughts, it looks as if they were telling jokes since the words "to" and "me" were used a lot and they were mainly next to each other, almost as if it was their way of laughing, like how we use "ha" as a way to type out our laughs, well nowadays we just use acronyms such as LOL or LMAO because it provides a shorter way of expressing ourselves and they are universal since most people know what it means and its added to our vocabulary so we can remember what it means, or kinda like a shortcut/hot key... However the other parts of their language is still a mystery since it's hard to decode it without a hint of what it means