This is a clip from a conversation with Garry Kasparov from Oct 2019. New full episodes once or twice a week and 1-2 new clips or a new non-podcast video on all other days. If you enjoy it, subscribe, comment, and share. You can watch the full conversation here: ruclips.net/video/8RVa0THWUWw/видео.html (more links below) Podcast full episodes playlist: ruclips.net/p/PLrAXtmErZgOdP_8GztsuKi9nrraNbKKp4 Podcasts clips playlist: ruclips.net/p/PLrAXtmErZgOeciFP3CBCIEElOJeitOr41 Podcast website: lexfridman.com/ai Podcast on Apple Podcasts (iTunes): apple.co/2lwqZIr Podcast on Spotify: spoti.fi/2nEwCF8 Podcast RSS: lexfridman.com/category/ai/feed/
@Iteration Zero it made sense to me. Here is the joke: an algorithm found a video praising how algorithms are great and unstoppable in the game of chess. If you personify this RUclips algorithm, then you could say it's bragging.
I love the way Kasparov’s explanation. He is saying directly what he means, he isn’t forgetting the small nuances and he isn’t barging when he says his success. He is very informative.
Garry is a really eloquent and precise speaker. It's not just his chess brilliance that makes the man - he's very wise and reflects well on a lot of topics.
I remember in 1997 when this happened I was amazed. Then my friend shrugged and said, "so what? it's like doing a computation contest VS a calculator".
@@prixiusnecrolance8531 He literally said it's not that. If English isn't your first language (it's not mine either) this can be confusing. When he says "it's not just the first time I lost to a machine, it's the first time I lost. Period." That means it's the first time he ever lost to anyone/anything. He probably meant first time in a serious tournament where he was trying his best.
@@PostSovietPod Now, I understood what he wants to say.... He wants to say that.... He lost was not because he play bad.. according to him.He tried to his highest potential.
Wow! First interview I watch with Gary, he is fascinating to listen to and provided more relevant info about AI than I heard so called AI experts provide. Thanks for this!
From the perspective of a professional in Data Science, these seemingly small insights from Garry have profound implications regarding humans making small corrections to save machine learning time. This may be lost on most, but it's very powerful.
I was interested in a closely related point, focusing on the areas where humans still have the advantage. As a professional in the same field, it's both interesting from the approach of "what does AI lack" and in terms of human + machine collaboration (which in a sense all deployment of "live, production" AI is)
Kasparov is super knowledgable about different types of games, their cultural and historical background, and general development of AI. Couldn't have explain it any better.
It's cool that Garry has insights into AI because he has gone head to head with it in an area of mastery. Listening to a radio program the other day they brought up a similar point that when AI is used to optimize certain goals, it can have a lot of unforeseen consequences because the program doesn't know how to weigh pros and cons of different outcomes in an open system. It lacks human values basically like morality and common sense.
@@Alf763 He does not say "you know" with a pure russian accent. A pure russian accent would make it sound more guttural than he does, which is why it's an example of a mixture of both (which is what Brandon Klopp was saying).
@@OneDerscoreOneder I didn’t say it was pure Russian, I’m saying his Russian accent is why it sounds like that, an upper class British accent if anything shortens the word to a dragged out “no” source being I have one
What a great assessment, that a machine will always beat us in a closed system. This also sheds light on the great value of human creativity, the type of mental activity that takes place in open systems. A machine will never "beat" a human at language, because it is an open system. Even the idea of beating a human at language sounds absurd.
@@nisatouseef7833 Language is an open system because all the rules and boundaries are flexible and can be bent or even broken. Take the single word "bad". Its technical definition means one thing but its slang definition can mean the exact opposite. This is the reason that even the best AI still cannot communicate via language as well as a 10 year old child.
@@nisatouseef7833 If a job can be accomplished without nuanced verbal communication or other forms of creativity, then most likely it CAN be replaced by a machine. I would guess that a stock broker can eventually be replaced by a machine, but a job such as a therapist will never be able to be replaced by a machine because it requires a great deal of nuanced verbal communication. The easier that a job can be entailed by a system of closed rules, the easier it can be replaced by a machine. This is just my opinion on the matter, I don't consider myself to be an expert.
Ravi Zacharias said it best: “The idea that Deep Blue has a mind is absurd. How can an object that wants nothing, fears nothing, enjoys nothing, needs nothing, and cares about nothing have a mind? It can win at chess, but not because it wants to. It isn't happy when it wins or sad when it loses. What are its [post]-match plans if it beats Kasparov? Is it hoping to take Deep Pink out for a night on the town?"
Ravi Zacharias is a fraud Televangelist , recently busted for having an affair with a mistress and many more scandals surrounding him. Look it up, its public knowledge know because his court case files got released. So get outta here with your Ravi Zacharias.
Sherief Elsayad Send me some links where it’s been all over the news. I can’t find any. All I can find is a book on the matter, written by an atheist of course.
Big K has that champions attitude of singular belief in himself. This is an illuminating discussion for sure
2 года назад+1
At the end of the video is the most interesting sentence, AlphaZero has weaknesses but it is hard for it to correct them unless it looses many times against an opponent that expose them... so at some point not trying too hard to win is the only way for learning new stuff and explore new territories in order to get better those times that you really want to win.
Gary Kasparov told me how to put the difference between human intelligence and artificial intelligence into words. The AI will triumph when the problem is closed ended but if it is an open ended problem, the machine does not know when the questions it asks gets into the territory of diminishing returns. Then I think does humans have a component beyond the physical which helps in open ended tasks, could people receive guidance in life, in dreams and in psychedelic experiences and does the guidance affect the physical world by determining patterns in how the quantum mechanical wave function collapses. That could maybe guide the signals in the brain because of the high electric and neurotransmitter gradients over very short distances in the brain? And then there should also be a mechanism for taking information out of the physical world. To maybe many other realms which all intersect with the physical world, and where God knits all the influences from the other realms together in the physical world?
IBM did a lot of shady stuff. There were even some that speculate there were human inputs during the matches. It's never been proven but they were being sly.
This may sound odd but my experience with a russian/english accent has been about hearing a russian/american english accent. Kasparov's accent sounds russian/UK English accent. Just an odd lilt to it
He’s from old USSR, but he’s an Armenian Jew from Azerbaijan, so that probably has some bearing on his accent, but it is indeed a very unique accent that I’ve always found odd
Garry understands the vulnerabilities of AI algorithms so well that humans can forget about the fear of AI singularity for good. I understand this too now
We lost to machines when we invented our first car, first steam engine, first calculator , first any technology which could do better than us. But we create machines as our tools so we should not try to compete with it but use it to our advantage. I think it’s stupid on owe side to try and compete against machine and then feel bad about losing. We are not comparable. It’s like trying to race against a car on your feet. Makes no sense
Ah, no. Most people can beat their cars in a race through a thick jungle, or up the side of a mountain, or through rice paddies, etf.. Cars are faster on a smooth road. Wheels don't help much on 90% of the earth's surface. What is scary is the latest robots are getting good at going places and doing things no machine could do before, and progress is accelerating, and breakthroughs are coming faster than expected. A computer can already beat Kasparov at IQ tests, and can invent new electronic circuit designs, and can read people's emotions better, etc. Etc.
@@illarionbykov7401 The only dystopian scenario is humans create robots and AI that have as basic goal to survive and to get out of control.Emotional intelligence and corporality of the emotions,physical pain and that we can think and change our basic goals(we can choice to stop play the "game" or change complete the "game",Kasparov has right) is the difference with any type of AI in the future.
Excellent stuff. GK raises some excellent points. How can humans help in advancing deep learning AI's? Even found out about IBM's neurogammon (learn something every day). Thnx!!!!
It physically hurts me to lose too, Garry. That is why I quit playing so many times. It does feel good to win which is why I still play. Now days I’m attempting to just have fun without the expectations of studying too hard.
Kasparov uno de los mejores jugadores de la historia pocos puedes dejar una huella tan grande y su contribución con la literatura ajedrecista , a mi parecer Kasparov es el 2 mejor jugador de la historia
@@mralion88 los puestos de los ajedrecistas son discutibles es decir dependen de cada persona y en lo que se fije en este caso sería así Robert James ficher, Gary Kasparov ,Anatoly Karpov , Magnus Carlsen , etc
Garry is wrong about the distinction between closed and open systems. In the 80s everybody "knew" that computer chess could never beat humans unless somehow we figured out some way of imbuing the programs with strategic plans. It turns out, you don't need plans if you have solid positional understanding and deep tactics; we just had to make the machines better at what they were already doing. Similarly "it's not about growing in size" (8:52) is the same myopic point of view of not realizing that a large enough quantitative change becomes qualitative.
Great points IMO. At one point he said that humans are more flexible, because it takes fewer games to correct the error, but computers can play through the games much quicker, so I don't know if there's a certain defined edge to either...
When Kasparov says "error" what he's really talking about is "anything other than the best move", chess players are typically ppl who are very hard on themselves. Something to keep in mind when he says that his match with Deep Blue had lots of "errors".
I know this is an old video, but thankyou Gary Kasparov for making it clear that chess is a game! I thought that 'idea' that it is a pinnacle of human intelligence was quite silly.
Fr some people are very obnoxious about chess but in reality it just promotes certain mental abilities which make it incredibly useful, (critical thinking, decision making, thinking ahead etc. the list goes on) but it's not the be all and end all of 'intelligence' which is hard to define and can be different for everyone.
The idea that people should stop playing chess because machines always win is like saying running should be removed from the Olympics because cars are always faster. Chess is better with humans because humans care about and engage in the struggle. Machines don't care. Check it out-many machine games are unwatchable.
Very interesting talk from a GM who ,modestly,explains that his mistakes are the cause of his loosing! Finaly he says the most important thing at the end:the machine will continue making the same mistakes...until a hulan corrects them!Nothing more to say!
I respectfully think Garry is missing the perspective at the end. Human flexibility is (many) orders of magnitude greater than AI, yes. But orders of magnitude have ways of melting away. When he was born, the best chess computer was probably some kind of state machine.
True but I think his point is that there will always remain a gap with current approaches to AI because domain-general intelligence is a distinct problem to anything domain-specific. The point is that if, as we have done, we draw the boundaries of a problem and then optimise a machine solution, we will of course beat humans, and even if those boundaries become wider and wider, they are ultimately still closed. The transition to complete flexibility is something else
@@Overclocked3770K 'always' and 'with current approaches' isn't realistic though. Of course current approaches are still learning to solve domain-specific tasks. But there are no end of ideas and pure scaling will bring new boundaries - eg: GPT-4 might very well be able to play chess (poorly, but again, scale...), and at that point you're starting to really become general.
I know in late but google say garry is 21, 19 and 104 in his 5 world championship matches so idk what he means first time he lost? Much respect he is a true master of chess
Programs like Stockfish are many hundreds of elo above humans at the moment, to put that into perspective, both Kasparov and Carlsen would struggle to beat it once in 100 games, and most probably would not beat it.
The interresting part for me is his thoughts on the subject. It takes AlphaZero hundreds of thousands of games to correct his mistakes. If helped by a human, this can go way faster. This should teach us to research self-aware neural nets that are able to adapt.
@Iteration Zero As for the number of games I forgot the quotes, I was actually quoting what Garry said. But you are right, it would take more than that to fine-tune the network to correct it's errors. As for self-awareness, it's also a hot topic recently. Some of its concepts could be considered adaptive learning. These are just naming conventions so don't pay much attention to it. As an example, for reliability, a self-aware system would automatically fix itself by using redundancy for example. This is an adaptive system but falls under the self-aware category. However, for your point, it is true that human biases could be a limiting factor to help training but I think Garry still thinks that humans would have the upper hand and that's what we have seen with many end-game studies that top engines (such as stockfish) judge completely winning but a human gm can easily tell that it is a draw.
Chess is NOT the ultimate expression of human intelligence. The world's greatest players walk around their day to day lives, no more able to see what's around the corner than anyone else. The skill simply does not transfer into the real world in any meaningful way. It's one amazingly specific skill. Being great at chess just means you're great at chess.
Being good at chess requires excellent spatial memory, which does transfer to fields like mathematics. There are a number of great chess players who were also great mathematicians, more so than say great photographers or great novelists.
Evolutionarily it seems reasonable to assume that a peer with exceptional cognitive ability in something you understand (a board game) will have similar exceptional ability in something you don't (e.g., chemistry). We all recognize that cognitive ability is general and transferable.
great reflection by GMaster Kasparov; Baudrillard said the same: the machine cannot THINK ! so forget about artificial intelligence, that's just computing.
I would agree with Garry that AI was and will be the winner in closed systems and perhaps not so much in open systems such as stock market trading. But with growth with neurons, parallelization, some of the previously "closed" systems such as say driving and now painting such as Wall-E, these open systems might actually be becoming closed. A bit scary
I doubt that computers will ever learn what humor is. Funniness can't be quantified by description or in any other way. Even harder than learning what humor is would be actually experiencing what we experience intellectually and emotionally when we laugh. Its pleasurable too. And pleasure is another thing a computer could never feel. When it comes down to it, a computer is just a data processor and a storer of data. The instruction set is limited, you can't make a human or even a sentient being out of a computer. BUT..and this is a big but...computers are unbelievably good at mimicking sentience, humor, intelligence. Mimicking vs Actual: there is a gigantic difference between the two, though it might not seem so. People who don't understand that computers can NEVER be sentient will one day make laws against unplugging a computer because they will be deemed to be alive and self aware. It'll happen.
Machines already do humor on a low level--puns and insults (yo mamma jokes) and they keep getting better. It's a matter of time before machines can do more sophisticated humor.
Regarding "sentient" you have to define what the word means first. Unless you have a convenient definition like "sentience is what only humans have and no other entitity can ever have" in which case, you are playing word games.
@@illarionbykov7401 By sentience I mean self awareness, i.e. consciousness. Computers don't know they exist. You can program a computer to say "I'm self aware", or otherwise mimic self awareness, but that's not self awareness. At no point in the execution of any instruction in a computer's program does it become self aware, because the the instruction set for any given CPU has no instruction for self awareness, but even if it did, as soon as that instruction was executed, a new one replaces it (such as fetch data and put it in a register), at which point the self awareness just went away. (instructions are executed one at a time in every core)
Something feels off about his reasoning. At least some open systems can be seen as a combination of a closed inner system and an outer layer which in in contact with the outside unknown. I can imagine an AI having a tight grasp on the inner closed system, while also constantly updating the connections in the outer layer based on what it comes in contact with. (This is analogous to long-term and short-term memory.) I can also imagine an AI optimizing the interaction between these two layers better than we can. Dunno, the way he downplays it seems a little reckless.
We can extrapolate what a closed-system is. Think of the world itself as a closed-system, with a lot of rules. I think one day AI will be able to work in much larger fields, like life itself.
I believe there was some underhandedness in how this was set up. We could all see what Kasparov was doing, why wouldn’t they show us what the machine was doing. The set up was already stacked against Kasparov. The moment you have differing conditions for each contestant it’s no longer a level playing field. Especially when one of the conditions favours one of the contestant.
It's the difference in terminology. Game = 1 chess game. Match = series of chess games. E.g. best of 5. What he's saying is that he had lost one off matches to players and computers before, which is normal and unavoidable, but 1997 was the first time he lost a series to anyone ever and it was a computer.
I was never that impressed with Deep Blue's victory. The match was even after 5 games. Kasparov played an opening he never plays in game 6, didn't play it right and lost. It was a bad choice.
He was very curious about the resources DB was using. They were pausing the game basically to let the computer think, but K all but accused them of cheating by referencing GMs at the time in real-time, not just referencing the games from storage on the machine, but literally calling them on the phone and asking what they’d do. So he played an esoteric line he didn’t play to throw that theory off.
@@jigartalaviya2340 Philosophy has been one of the most important aspect of any human civilization that ever existed on earth, it helps the advancement of science, mathematics, literature, logic, art, politics, etc. It is the one that distinguishs human from machines.
![Felix "Unfair" | [Stray Kids : SKZ-PLAYER]](http://i.ytimg.com/vi/Oswujxm2Ag0/mqdefault.jpg)
This is a clip from a conversation with Garry Kasparov from Oct 2019. New full episodes once or twice a week and 1-2 new clips or a new non-podcast video on all other days. If you enjoy it, subscribe, comment, and share. You can watch the full conversation here: ruclips.net/video/8RVa0THWUWw/видео.html
(more links below)
Podcast full episodes playlist:
ruclips.net/p/PLrAXtmErZgOdP_8GztsuKi9nrraNbKKp4
Podcasts clips playlist:
ruclips.net/p/PLrAXtmErZgOeciFP3CBCIEElOJeitOr41
Podcast website:
lexfridman.com/ai
Podcast on Apple Podcasts (iTunes):
apple.co/2lwqZIr
Podcast on Spotify:
spoti.fi/2nEwCF8
Podcast RSS:
lexfridman.com/category/ai/feed/
Please do a video on the Poker AI Plaribus
Is it me or is this guy extremely impressive?
So IMB didn't give Gary chance to run it back after 97 match even tho it was split 1 to 1?
the look on your face when you realise his first lost at chess 😅
Keep up the great work Lex. I love your set of questions. Plus, I am a teal fan of Kasparov as chess player and lately as activist!
if a youtube algorithm suggested this to me is it essentially just humble bragging by the computer?
Iteration Zero it’s a joke
@Iteration Zero it made sense to me. Here is the joke: an algorithm found a video praising how algorithms are great and unstoppable in the game of chess. If you personify this RUclips algorithm, then you could say it's bragging.
Iteration Zero u didn’t have to ruin it for the rest of us man
@Iteration Zero you must be fun at parties .
He's mugged you all off 😂😂😂 poor saps
Kasporov is not just a great chess player, but a wonderful human being with superb insight of the human condition.
He's a wonderful insightful communicator which is why business love to have him at their prestigious presentations!
*Kasparov
He used to be an arrogant, narcissistic person back then.. now he has transformed into a friendly uncle..like Mike tyson
I love the way Kasparov’s explanation. He is saying directly what he means, he isn’t forgetting the small nuances and he isn’t barging when he says his success. He is very informative.
Garry is a really eloquent and precise speaker. It's not just his chess brilliance that makes the man - he's very wise and reflects well on a lot of topics.
I remember in 1997 when this happened I was amazed. Then my friend shrugged and said, "so what? it's like doing a computation contest VS a calculator".
Your friend was an idiot apparently.
@@brunonkowalski not really though
@@brunonkowalski No, it is quite the case. Put that in the mecanic world : would you be sad if a car move faster than a runing man ?
@@antoninatgerauteur9306 It is quite the case only for people clueless about programing, chess and how computer algorithms work :)
@@antoninatgerauteur9306 Do you have any idea how complicated chess is though?
"First time I lost, period." Damn. And people discount context.
Why has Kasparov said that ? He certainly lost other matches before 1997, he refers to not losing any tournament maybe.
@@mateiacd first time losing to AI
@@prixiusnecrolance8531 He literally said it's not that. If English isn't your first language (it's not mine either) this can be confusing. When he says "it's not just the first time I lost to a machine, it's the first time I lost. Period." That means it's the first time he ever lost to anyone/anything.
He probably meant first time in a serious tournament where he was trying his best.
He said it was the first match he lost. Match =/= game. A match in chess is a series of games, like the 6 game match between Kasparov vs deep blue.
@@PostSovietPod Now, I understood what he wants to say....
He wants to say that....
He lost was not because he play bad.. according to him.He tried to his highest potential.
"Capitalizing on our mistakes." That's exactly what modern technology is best at, from making us social media addicts to beating us in chess.
Wow! First interview I watch with Gary, he is fascinating to listen to and provided more relevant info about AI than I heard so called AI experts provide. Thanks for this!
From the perspective of a professional in Data Science, these seemingly small insights from Garry have profound implications regarding humans making small corrections to save machine learning time. This may be lost on most, but it's very powerful.
I was interested in a closely related point, focusing on the areas where humans still have the advantage. As a professional in the same field, it's both interesting from the approach of "what does AI lack" and in terms of human + machine collaboration (which in a sense all deployment of "live, production" AI is)
It means there might be some more fundamental concept to pattern finding that we humans have access to that even we don’t understand yet.
Kasparov is super knowledgable about different types of games, their cultural and historical background, and general development of AI. Couldn't have explain it any better.
Gary could have defeated Deep Blue by pulling the plug.
Would have been legendary if prior to resigning, GK actually did pull the plug.
The computer equivalent would be, Just electrocute/suffocate/starve/crush/immolate the human, no need to play a board game with them.
@@LOVESPACEDREAMS A preview of times coming...?
This guy is not just a GM, techie as well.
I’m sure you’ve heard this, but this is one hell of a resource you’re making for humanity right here with this series.
gay
@@DanielOnFire101 no, but thanks for the offer.
It's cool that Garry has insights into AI because he has gone head to head with it in an area of mastery. Listening to a radio program the other day they brought up a similar point that when AI is used to optimize certain goals, it can have a lot of unforeseen consequences because the program doesn't know how to weigh pros and cons of different outcomes in an open system. It lacks human values basically like morality and common sense.
deep blue wasn't an AI
@@tomhill4738 What was it then, an AI with human influence?
How wonderful to hear Kasparov mention Dota - who would have thought it
Kasparov knowing about OpenAI and Dota made my day
I've never heard anyone that has an accent that is both full out high-society British and Russian simultaneously.
What? I don’t hear any British in his accent at all
@@Alf763 listen to how he says “you knoew”
@@OneDerscoreOneder yeah it’s because of his Russian accent, that’s not how posh British people say it
@@Alf763 He does not say "you know" with a pure russian accent. A pure russian accent would make it sound more guttural than he does, which is why it's an example of a mixture of both (which is what Brandon Klopp was saying).
@@OneDerscoreOneder I didn’t say it was pure Russian, I’m saying his Russian accent is why it sounds like that, an upper class British accent if anything shortens the word to a dragged out “no” source being I have one
I really like his jacket and he’s rocking that ‘deep’ blue shirt too
Great talk. I do love Kasparov.
4:08 'In japan heart surgeon number 1, STEADY HAND'
Sweet. I am 33yrs old with Stage 3 heart failure, ill have to keep that in mind and hope big. :)
@@strikeforcek9149 This is supposed to be a joke reference from the TV show "The office". It's not for real.
@@abhishekkaranath2119 ahhhh... damn, I got excited! Lol
(I'm gonna have to go watch it now. Hehe)
Darryl save life!
As soon as I heard it I went to the comments to look for this!
Agent Smith: "Kasparov? Yes of course he is a legend"
The Oracle said one day a man will be born who can beat any computer
Agent Smith: "...the PEAK, of your civilisation. And I say YOUR, civilization, because..."
African MusicGenuis Garry Kasparov even mentions the Matrix in the TED talk he did where he discusses man and machine ;)
One of the greatest videos explaining AI in my opinion.
Thank you Lex and Garry for this eye opening disucssion.
More like round opening discussion. ;)
@@chrisi237 could you explain what you mean by round opening discussion?
@@pranavsreedhar1402 It was a chess joke.
@@chrisi237 right.. now I get it.
What a great assessment, that a machine will always beat us in a closed system. This also sheds light on the great value of human creativity, the type of mental activity that takes place in open systems. A machine will never "beat" a human at language, because it is an open system. Even the idea of beating a human at language sounds absurd.
Could you futher elaborate that idea??
@@nisatouseef7833 Language is an open system because all the rules and boundaries are flexible and can be bent or even broken. Take the single word "bad". Its technical definition means one thing but its slang definition can mean the exact opposite. This is the reason that even the best AI still cannot communicate via language as well as a 10 year old child.
@@BGlasnost ao do you mean that like a job of a stock broker or salesman cannot be replaced?
@@nisatouseef7833 If a job can be accomplished without nuanced verbal communication or other forms of creativity, then most likely it CAN be replaced by a machine. I would guess that a stock broker can eventually be replaced by a machine, but a job such as a therapist will never be able to be replaced by a machine because it requires a great deal of nuanced verbal communication. The easier that a job can be entailed by a system of closed rules, the easier it can be replaced by a machine. This is just my opinion on the matter, I don't consider myself to be an expert.
Ravi Zacharias said it best:
“The idea that Deep Blue has a mind is absurd. How can an object that wants nothing, fears nothing, enjoys nothing, needs nothing, and cares about nothing have a mind? It can win at chess, but not because it wants to. It isn't happy when it wins or sad when it loses. What are its [post]-match plans if it beats Kasparov? Is it hoping to take Deep Pink out for a night on the town?"
Ravi Zacharias is a fraud Televangelist , recently busted for having an affair with a mistress and many more scandals surrounding him. Look it up, its public knowledge know because his court case files got released. So get outta here with your Ravi Zacharias.
Sherief Elsayad Just a bunch of lies to smear a man of Truth.
@@joverstreet24 dude, it's been all over the news he got exposed , he is a filthy con artist, not a man of truth
Sherief Elsayad Send me some links where it’s been all over the news. I can’t find any. All I can find is a book on the matter, written by an atheist of course.
@@joverstreet24 "Nobody I agree with is saying it, so how could it be true?" Stay simple you religious clowns, stay simple.
Big K has that champions attitude of singular belief in himself. This is an illuminating discussion for sure
At the end of the video is the most interesting sentence, AlphaZero has weaknesses but it is hard for it to correct them unless it looses many times against an opponent that expose them... so at some point not trying too hard to win is the only way for learning new stuff and explore new territories in order to get better those times that you really want to win.
Love the disparity in glassware. The interview has an ultra-tall glass barely filled with water and Kasparov has a short, white, squat coffee cup.
I giggled at this, never thought anyone else would notice.
Gary Kasparov told me how to put the difference between human intelligence and artificial intelligence into words. The AI will triumph when the problem is closed ended but if it is an open ended problem, the machine does not know when the questions it asks gets into the territory of diminishing returns.
Then I think does humans have a component beyond the physical which helps in open ended tasks, could people receive guidance in life, in dreams and in psychedelic experiences and does the guidance affect the physical world by determining patterns in how the quantum mechanical wave function collapses.
That could maybe guide the signals in the brain because of the high electric and neurotransmitter gradients over very short distances in the brain?
And then there should also be a mechanism for taking information out of the physical world. To maybe many other realms which all intersect with the physical world, and where God knits all the influences from the other realms together in the physical world?
How high were you when you wrote this 😭
GK demonstrates an Einstein-like understanding of AI. Absolutely legendary!
So IMB didn't give him a chance to run it back after 97 match? The matches were split 1 to 1
IBM did a lot of shady stuff (often against the wishes of the deep blue development team)
IBM did a lot of shady stuff. There were even some that speculate there were human inputs during the matches. It's never been proven but they were being sly.
This may sound odd but my experience with a russian/english accent has been about hearing a russian/american english accent. Kasparov's accent sounds russian/UK English accent. Just an odd lilt to it
He’s from old USSR, but he’s an Armenian Jew from Azerbaijan, so that probably has some bearing on his accent, but it is indeed a very unique accent that I’ve always found odd
He knows about DOTA WOW
It would be so good to get Garry on Joe Rogan
Garry would get bored hearing Joe talk about DMT and smoking weed.
@@manovoid741 That is how it would go. Lol.
Garry understands the vulnerabilities of AI algorithms so well that humans can forget about the fear of AI singularity for good. I understand this too now
Kasparov has improved. He makes sense now.
Man didn’t lose to machine. Man created the machine. He lost to the efforts of many men and women
You mean the efforts of many men, lol
Kasparov’s book “Deep thinking “ is a wonderful read about AI and what intelligence is. I highly recommend it
thank you very much Mister Gary!! that's exactly what I wanted to know about alphazero!
Out of the Rabbit Hole, into the Lex Fridman interview :)
This was extremely interesting, thanks. 👏👏
eye-opening conversation
Gary is a fantastic talker. Great insight and very interesting.
We lost to machines when we invented our first car, first steam engine, first calculator , first any technology which could do better than us.
But we create machines as our tools so we should not try to compete with it but use it to our advantage.
I think it’s stupid on owe side to try and compete against machine and then feel bad about losing. We are not comparable. It’s like trying to race against a car on your feet.
Makes no sense
Ah, no. Most people can beat their cars in a race through a thick jungle, or up the side of a mountain, or through rice paddies, etf.. Cars are faster on a smooth road. Wheels don't help much on 90% of the earth's surface.
What is scary is the latest robots are getting good at going places and doing things no machine could do before, and progress is accelerating, and breakthroughs are coming faster than expected. A computer can already beat Kasparov at IQ tests, and can invent new electronic circuit designs, and can read people's emotions better, etc. Etc.
@@illarionbykov7401 The only dystopian scenario is humans create robots and AI that have as basic goal to survive and to get out of control.Emotional intelligence and corporality of the emotions,physical pain and that we can think and change our basic goals(we can choice to stop play the "game" or change complete the "game",Kasparov has right) is the difference with any type of AI in the future.
@@illarionbykov7401 drones can beat us in all those.
Excellent stuff. GK raises some excellent points. How can humans help in advancing deep learning AI's?
Even found out about IBM's neurogammon (learn something every day).
Thnx!!!!
I see youtube algorithm wanted to flex by suggesting this video
It physically hurts me to lose too, Garry. That is why I quit playing so many times. It does feel good to win which is why I still play. Now days I’m attempting to just have fun without the expectations of studying too hard.
Somebody should tell Garry about OpenAI's GTP-3. A great read on this is the short text "A Bitter Lesson" by Rich Sutton.
Kasparov uno de los mejores jugadores de la historia pocos puedes dejar una huella tan grande y su contribución con la literatura ajedrecista , a mi parecer Kasparov es el 2 mejor jugador de la historia
bobby fischer es el primero?
@@mralion88 los puestos de los ajedrecistas son discutibles es decir dependen de cada persona y en lo que se fije en este caso sería así
Robert James ficher, Gary Kasparov ,Anatoly Karpov , Magnus Carlsen , etc
He won the first match that's all I care. If he's allowed to undo every move he might defeat an AI.
Garry is wrong about the distinction between closed and open systems. In the 80s everybody "knew" that computer chess could never beat humans unless somehow we figured out some way of imbuing the programs with strategic plans. It turns out, you don't need plans if you have solid positional understanding and deep tactics; we just had to make the machines better at what they were already doing. Similarly "it's not about growing in size" (8:52) is the same myopic point of view of not realizing that a large enough quantitative change becomes qualitative.
Great points IMO. At one point he said that humans are more flexible, because it takes fewer games to correct the error, but computers can play through the games much quicker, so I don't know if there's a certain defined edge to either...
When Kasparov says "error" what he's really talking about is "anything other than the best move", chess players are typically ppl who are very hard on themselves. Something to keep in mind when he says that his match with Deep Blue had lots of "errors".
Such an interesting conversation!
i am amazed he mentioned DOTA.
anyone know where kasparov has got this upper class english accent from? did he go to school in the england?
I love that he mentioned DOTA in games 😀
westerners: chess is the peak of what a human mind can accomplish
chinese: ever tried go?
He saying that a bit later in the video, you attention span isn’t very good is it?!
@@nikolaos1991 no, pretty bad
I know this is an old video, but thankyou Gary Kasparov for making it clear that chess is a game! I thought that 'idea' that it is a pinnacle of human intelligence was quite silly.
Fr some people are very obnoxious about chess but in reality it just promotes certain mental abilities which make it incredibly useful, (critical thinking, decision making, thinking ahead etc. the list goes on) but it's not the be all and end all of 'intelligence' which is hard to define and can be different for everyone.
The idea that people should stop playing chess because machines always win is like saying running should be removed from the Olympics because cars are always faster.
Chess is better with humans because humans care about and engage in the struggle. Machines don't care. Check it out-many machine games are unwatchable.
Very interesting talk from a GM who ,modestly,explains that his mistakes are the cause of his loosing!
Finaly he says the most important thing at the end:the machine will continue making the same mistakes...until a hulan corrects them!Nothing more to say!
I respectfully think Garry is missing the perspective at the end. Human flexibility is (many) orders of magnitude greater than AI, yes. But orders of magnitude have ways of melting away. When he was born, the best chess computer was probably some kind of state machine.
True but I think his point is that there will always remain a gap with current approaches to AI because domain-general intelligence is a distinct problem to anything domain-specific. The point is that if, as we have done, we draw the boundaries of a problem and then optimise a machine solution, we will of course beat humans, and even if those boundaries become wider and wider, they are ultimately still closed. The transition to complete flexibility is something else
@@Overclocked3770K 'always' and 'with current approaches' isn't realistic though. Of course current approaches are still learning to solve domain-specific tasks. But there are no end of ideas and pure scaling will bring new boundaries - eg: GPT-4 might very well be able to play chess (poorly, but again, scale...), and at that point you're starting to really become general.
Chess is just a conversation with rules, after all.
I know in late but google say garry is 21, 19 and 104 in his 5 world championship matches so idk what he means first time he lost? Much respect he is a true master of chess
Completely agree with Garry's views. Computer will always win because it is based on rules and don't make mistakes as human can do in some moves.
Garry legenda he never agreed to serve Putin’s regime for any money!
He is well serving the Western oligarchs and their agendas.
So how many top spot does machine claim in the chess field now?
All of them
Programs like Stockfish are many hundreds of elo above humans at the moment, to put that into perspective, both Kasparov and Carlsen would struggle to beat it once in 100 games, and most probably would not beat it.
Kasparov is underrated.
The interresting part for me is his thoughts on the subject. It takes AlphaZero hundreds of thousands of games to correct his mistakes. If helped by a human, this can go way faster. This should teach us to research self-aware neural nets that are able to adapt.
@Iteration Zero As for the number of games I forgot the quotes, I was actually quoting what Garry said. But you are right, it would take more than that to fine-tune the network to correct it's errors. As for self-awareness, it's also a hot topic recently. Some of its concepts could be considered adaptive learning. These are just naming conventions so don't pay much attention to it. As an example, for reliability, a self-aware system would automatically fix itself by using redundancy for example. This is an adaptive system but falls under the self-aware category. However, for your point, it is true that human biases could be a limiting factor to help training but I think Garry still thinks that humans would have the upper hand and that's what we have seen with many end-game studies that top engines (such as stockfish) judge completely winning but a human gm can easily tell that it is a draw.
Дотеры на месте
3:20
Chess is NOT the ultimate expression of human intelligence. The world's greatest players walk around their day to day lives, no more able to see what's around the corner than anyone else. The skill simply does not transfer into the real world in any meaningful way. It's one amazingly specific skill. Being great at chess just means you're great at chess.
Being good at chess requires excellent spatial memory, which does transfer to fields like mathematics. There are a number of great chess players who were also great mathematicians, more so than say great photographers or great novelists.
Evolutionarily it seems reasonable to assume that a peer with exceptional cognitive ability in something you understand (a board game) will have similar exceptional ability in something you don't (e.g., chemistry). We all recognize that cognitive ability is general and transferable.
Being able to say "it's just a game" is pretty big of him
I love Garry.
Which thinks Garry is the best , deep blue or alpha zero
great reflection by GMaster Kasparov; Baudrillard said the same: the machine cannot THINK ! so forget about artificial intelligence, that's just computing.
"cHeSs iS a cLoSeD sYsTeM" - Greatest chess player, who played and studied the game their whole life.
Kasparov is really smart.
That is all.
Amazingly well rounded guy unlike many chess masters with zero personality
Makes perfect sense. Open system we can just smash the crap out of the computer and then say "check mate. What you gonna do now?"
Skynet comes to mind? If the open system allows pathways for violence for the ai...
@@jeremykothe2847 You ever see one of your old comments and have no idea of the context?
@@TheDantheman12121 More than I'd care to admit :)
0.1° of gun direction change gives 2.8m difference on 1mile :P
He’s copping so hard
Bruh wtf
Great point…peak in Europe, not everywhere
Legendary chat
I didn't realise that was the first match he ever lost. Wow!
I wouldn't say humans are in an "open" system. It's just a series of closed systems, however numerous.
I would agree with Garry that AI was and will be the winner in closed systems and perhaps not so much in open systems such as stock market trading. But with growth with neurons, parallelization, some of the previously "closed" systems such as say driving and now painting such as Wall-E, these open systems might actually be becoming closed. A bit scary
Dota! He mentioned my game, yay!
I doubt that computers will ever learn what humor is. Funniness can't be quantified by description or in any other way. Even harder than learning what humor is would be actually experiencing what we experience intellectually and emotionally when we laugh. Its pleasurable too. And pleasure is another thing a computer could never feel. When it comes down to it, a computer is just a data processor and a storer of data. The instruction set is limited, you can't make a human or even a sentient being out of a computer. BUT..and this is a big but...computers are unbelievably good at mimicking sentience, humor, intelligence. Mimicking vs Actual: there is a gigantic difference between the two, though it might not seem so. People who don't understand that computers can NEVER be sentient will one day make laws against unplugging a computer because they will be deemed to be alive and self aware. It'll happen.
John van Neumann put it the best. You can program a machine to do anything that a human can do...so long as you know how a human functions first.
@Pulimuli It's amazing that you missed the entire point of what I said
Machines already do humor on a low level--puns and insults (yo mamma jokes) and they keep getting better. It's a matter of time before machines can do more sophisticated humor.
Regarding "sentient" you have to define what the word means first. Unless you have a convenient definition like "sentience is what only humans have and no other entitity can ever have" in which case, you are playing word games.
@@illarionbykov7401 By sentience I mean self awareness, i.e. consciousness. Computers don't know they exist. You can program a computer to say "I'm self aware", or otherwise mimic self awareness, but that's not self awareness. At no point in the execution of any instruction in a computer's program does it become self aware, because the the instruction set for any given CPU has no instruction for self awareness, but even if it did, as soon as that instruction was executed, a new one replaces it (such as fetch data and put it in a register), at which point the self awareness just went away. (instructions are executed one at a time in every core)
Something feels off about his reasoning. At least some open systems can be seen as a combination of a closed inner system and an outer layer which in in contact with the outside unknown. I can imagine an AI having a tight grasp on the inner closed system, while also constantly updating the connections in the outer layer based on what it comes in contact with. (This is analogous to long-term and short-term memory.) I can also imagine an AI optimizing the interaction between these two layers better than we can. Dunno, the way he downplays it seems a little reckless.
We can extrapolate what a closed-system is. Think of the world itself as a closed-system, with a lot of rules. I think one day AI will be able to work in much larger fields, like life itself.
The difference is that with chess we know all the rules. With the physical universe, we don't.
May be a set of effective rules can make a system effectively closed. But to determine such set of rules will be a challenge for any programme
I believe there was some underhandedness in how this was set up. We could all see what Kasparov was doing, why wouldn’t they show us what the machine was doing. The set up was already stacked against Kasparov. The moment you have differing conditions for each contestant it’s no longer a level playing field. Especially when one of the conditions favours one of the contestant.
Was it painful because you didn't realize of my premise or was it because you did? My God.
Didn't he first say 1997 match was the first time he lost. Later on he says he lost some matches before that? What the hell?
It's the difference in terminology. Game = 1 chess game. Match = series of chess games. E.g. best of 5. What he's saying is that he had lost one off matches to players and computers before, which is normal and unavoidable, but 1997 was the first time he lost a series to anyone ever and it was a computer.
Garry should create his own microchip
Emotional is what break human also what if ultrasound hidden play on human
Garry is looking good.
I was never that impressed with Deep Blue's victory. The match was even after 5 games. Kasparov played an opening he never plays in game 6, didn't play it right and lost. It was a bad choice.
He was very curious about the resources DB was using. They were pausing the game basically to let the computer think, but K all but accused them of cheating by referencing GMs at the time in real-time, not just referencing the games from storage on the machine, but literally calling them on the phone and asking what they’d do. So he played an esoteric line he didn’t play to throw that theory off.
8:58
Why is chess seen as the pinnacle of humanity? It's number crunching. I'd think that art, philosophy, literature, science, medicine, etc are the peak.
Science and medicine...for sure.
Literature...yeah.Why not??
Art and philosophy??...u can throw them in garbage.Wdnt make any difference.
@@jigartalaviya2340 Philosophy has been one of the most important aspect of any human civilization that ever existed on earth, it helps the advancement of science, mathematics, literature, logic, art, politics, etc. It is the one that distinguishs human from machines.