Stephen Wolfram: Computational Universe | MIT 6.S099: Artificial General Intelligence (AGI)
Вставка
- Опубліковано 1 бер 2018
- This is a talk by Stephen Wolfram for MIT course 6.S099: Artificial General Intelligence. This class is free and open to everyone. Our goal is to take an engineering approach to exploring possible paths toward building human-level intelligence for a better world.
INFO:
Course website: agi.mit.edu
AI podcast: lexfridman.com/ai
CONNECT:
- If you enjoyed this video, please subscribe to this channel.
- AI Podcast: lexfridman.com/ai/
- Show your support: / lexfridman
- LinkedIn: / lexfridman
- Twitter: / lexfridman
- Facebook: / lexfridman
- Instagram: / lexfridman - Наука та технологія
"I used to do particle physics when I was a kid." Genius.
The crazy thing is he’s not lying. He got is PhD from Caltech at 21 😂
@@infinityinf1 A lot of Computer Science people started out in Maths/Physics, e.g. Dennis Ritchie, Donald Knuth, etc.
Talk:
00:00 - 10:30 introduction & motivation for making the question answering system (Wolfram Alpha)
10:30 - 26:15 ultra cool Wolfram language demos
26:15 what is in principle possible to do with computation?
44:45 how do we make use of all of that stuff
Questions:
58:55 knowledge representation within Wolfram Alpha, how do different domains interact?
1:03:45 who or what are your key influences?
1:09:20 wolfram language for describing purpose/philosophy?
1:15:54 do molecules compute? what do you mean by that?
1:21:00 elaborate on searching the computing spaces?
1:24:00 how should we teach kids about AI/computational thinking? (great answer! IMHO)
1:38:20 how far away are we from learning machines to explain stuff to humans?
1:41:10 follow up question: personalized learning, can we do it?
1:44:10 Newton seashell quote and what would Franz Liszt say :)
1:52:10 the evolution of purposes
Thanks for the amazing series! My try to give a small contribution back :)
Thank you!
Thanks!
thanks
amazing thanks
my mans was flexing his projects for 26 min
this man has an enourmous amount of knowledge and thoughts in his mind
loopuleasa *Born with an exceptional IQ.
And an enormous ego.
... so much that he get's lost in it/them
@Vicktorio Alhakim You obviously haven't met many top academics or scientists. They all have egos. Haven't you read any history of science? Besides, who are you to judge anyone else? What have you achieved or accomplished? Your comment is a reflection of your own resentment and envy.
@@randalltaves6033 On the contrary, I work in academia, and I interact with people with giant egos on a daily basis, myself included. As for my achievements, you can always find my work in academia. In no way does my comment imply that having a giant ego is a bad thing. In fact, it is a strong asset in academia, if you want to survive as a successful researcher. However, in many times it can cloud your judgement, and one should be very careful in this regard.
Nor do I judge Wolfram in any way, and in fact I admire his work.
It's funny and ironic how you can judge me and my personality based on a single comment on UA-cam. Your mischaracterization of my comment however, reflects your true character more than it does mine.
I don't know till now that he is the brain behind the concept of language in the movie The Arrival. He is definitely one of the greatest minds of this time.
Steven wolfram's cellular automaton and his book "a new kind of science" is genius. I love the idea of "mining" infinity for new programs etc.
Feeling so grateful for this channel.
Thank you for sharing, Lex.
This is fascinating! Excellent course!
Fast forward the demo, but please listen to his answers to the questions (1 hour in) - the answers are impressively better than the questions :+) - superb clarity of thinking built on a career of building useful stuff.
Richard Hopkins .
Could not agree more
Thanks for sharing on UA-cam. I hope you keep sharing lectures. They are so inspiring.
Mr Wolfram is a true genius. Wolfram Alpha helped me a lot in my Applied Maths studies.
Wolfram is a very deep, under-appreciated, thinker.
On the road to learning science, Stephen Wolfram is the light that leads me forward.
What a great talk! Thank you very much for sharing
Mind blowing, thank you for sharing 🙏
Amazing lecture. Thank you
Wolffie! I'm your groupie since I was 17! We love you! 😍
Great man, thank you so much
Amazing lecture, thanks for this kind of content for anyone brave enough
The ending is profound... and he was just getting warmed up
Great lecture, thanks.
I benefit every single word in this series so I am enjoying
Thank you for the video :)
While listening to this talk, which was out of my league, I couldn't help but think of Jorge Louis Borges' short story The Library of Babel. I think there is something to be said about the program Mr Wolfram was using and "mining the computational universe" that parallels nicely with the story. Hope you like it if you get around to it!
Real interesting. I didn't think Wolfram Alpha also made use of neural networks. Awesome stuff!
My hero!
Wow, very impressive lecture and I'm amazed by his vast knowledge. Now need to check out his books! And ditto on the problem with our math education! Teaching analytical thinking is a way to go instead of how to compute integrals...
are you kidddding me!!! why the hell am I discovering this gold mine now.....thanks man
Loopuleasa I am not talking about time ,I am talking about Wolfram language.It's 😍😍😍
Edit:I posted comment on that same day.
dhrumil barot I first found out about Wolfram when I bought a Raspberry Pi several years ago. I really liked Mathematica and was really bummed when I saw the price for the Windows version.
Just imagine, hypothetically.
-Lets say AGI, or any kind of marketed (multiLayered-integrated algorithmic systems) competition leads to larger investments in companies building the smartest systems, in the future these systems get so complex they harvest up all elemental molecular resources that the planet holds.
-Building vastly more complex systems that are in direct contact with similar systems as competitors for matter and energy, selection pressures emerge only allowing certain arrangements of intelligent matter to exist because of this "new super lifes" pressure on the solar systems local environment.
-Maybe 5 billion years in the future, the milky way might be a completely different environment with entropy increasing because these new Hyper Life Forms are constantly consuming.
Yah,mayb very soon ppl look back and thought why the heck bother 2 discriminate,no need 2 multiply with meaningless fake weigh var
@@dhrumilbarot1431 right?! same here!
Excellent
Perfect next guest for the podcast.
While I am appreciative Stephen took time away from his work to give this talk, I couldn't help but notice that this presentation moreso than any other is not really about pedagogy, but about idealogical promotion and advertisement of his product. I wish Stephen all the best in his pursuit of fame and economic success.
This was an amazing lecture!
Great talk :)
Wow this was a fantastic talk. I an not a programmer or mathematician but I love playing around with cellular automata, and now I really want to check out Wolrfram Language!
As an aside I think I recognized Tim Tyler as the second questioner. His youtube channel is really cool. He hasn't had new content in years though. I hope he is doing cool stuff. His chanel was about darwinian evolution, memetics and various replicating systems of one type or another. Fun stuff! Of course I could be wrong, but the questioner's voice really does sound like him.
Stephen Wolfram!!!!!!
This... was... splendid. Thank-you.
nice one
I'm just a humble tool maker that developed a new kind of science all by myself.
Whether or not you like the way he presents his information, it’s still correct in its accuracy and functional power
52:00 reminiscent fundamental aspect of Chris Langan's CTMU. The universe is self computational.
If you are interested in where contemporary philosophy is in respect to Gödel's theorem you should check out Alain Badious work, who elaborates explicitly on the consequences of this set theory. As Steven said, apart from Badiou (who follows Lacan - the only man who actually tried to formalize psychology - exactly the mentioned "I want" part of sentence) there is little to non thought put to formalization in philosophy. Badiou is the best you can get.
He'd better release/opensource /public domain the whole wolfram alpha engine. That way, volunteers from around the world will contribute, and help make it the ultimate world computable encyclopedia, the grand world simulator for anyone , anywhere. That would be his greatest possible impact.
exactly
this is like a university hacker club chat, all drunk, love it!!
I'm thinking that the camera-man is actually a machine attached to an AI algorithm that can predict where the lecturer is moving and smoothly adjust its direction. it can also store the "movement hobby" of the lecturer such that for different people acting similarly, it knows some of them are about to move and some of them are only using gestures. AMAZING.
yes what ever its is good
yes
Mind=blown
This guy talked at my school. He said his biggest life regret was choosing not to get his PHD in high school in particle physics. Wolfram was in high school around 1960 when particle physics was a blooming field.
Me Myself and I - this guy has done it all on his own !
Fascinating. Stephen attempted to create a version of Wolfram Alpha when he was 12, and the ambition never left him... there's something here.
It's all very interesting. But I take it all with a healthy dose of scepticism and a pinch of salt.
1:31:02 "in the end it will not be possible to know how it works inside"....blackbox calculator at its best,
I'm his one fan of his. He spent thirty years on one thing, it's very hard for the normal.
Interview Demis Hassabis, please!
So, are we close or not? By looking at the demo it seems like we're close, but by the talk not that sure of that. So, are we considering the Symbolic way or not?
I used to do particle physics... when I was a kid.
20 is a kid to him
Index he started at 14 and got PhD at 21
@@JameBlack He started much earlier, about age 8. At 14, he'd already published three books on QM and at twelve, a compendium on QM.
Buckminster Fuller was pier less until now. Shame they’re not contemporary. Maybe I should add Newton but never heard him lecture.
Any real world application of cellular automata besides what these guys might have baked in their technologies (Mathematica, Wolfram Alpha, ...)?
Where does he talk about AGI? Or demo it?
Stephen "that's a different issue" Wolfram
So to speak
33:45 to 34:15 is me trying to BS my way through an answer I should know.
"do the right thing" = the answer to life= maximize sum(individuals(future freedom of action))= the maximization of the sum future freedom of action for all individuals.
granted; how we define the set of 'individuals' is always going to be an issue for ANY moral system.
Mathematica demo always stars from 2+2.
Give Deep Mind access to Wolframm Alpha as a resource skill.
How about we win if Google wins, and we win if Wolfram wins. If they collaborate and then individually lose - we lose
When I ask Wolfram Alfa it does not give me any useful answer. Salary distribution in Sweden shows me salary in the US based on 2009 data.
Is 'deep learning' theory the same notion as 'neural net' learning ?
Yeah basically. Some people use "deep learning" to mean systems that learn a hierarchy of features with increasing abstraction, but most people just mean neural nets with more than 1 layer.
yes
Is wolfram planning to work with a research team working on Homotopy Type Theory ?
However I'm still years away of from this, lots of years of studying and development.
He did that hand sign at 1:35:45
Could you please elaborate on the "that" part?
He's carrying on his majesty Lucifer's will
26 minutes list in the weeds ... and *then* has something interesting to say. :)
i think wolfram language is so deep. and i am so weak to grasp or follow it and its the same problem with Stephen wolfram that he could explain to people in a single lecture i watched it a lot in ted x, Harvard and other university he talked but for me its not easy to grasp
don't worry, he always repeats the same demo over and over because it is shockingly impressive. Real wolfram language takes a bit to learn but it is not harder than the domain languages it implements, so if you know about economy and programming then you could find wolfram language easy for economic related things, but not for chemistry.
Ever notice how Steve, unlike most of the other intelligentsia that I listen to, rarely mentions the people who have laid the ground upon which his stuff is built. I also, whenever he has to answer a question that would coney a sense of ""knowing wtf I'm talking about" get the impression of that I am witnessing the human equivalent of a rock trying to float. Which leads me to wonder, when he joke about one day learning how his thing works, that maybe he should not try so hard to make it seem like at night he's under the hood of Wolfram AlphabigdickSteve with a wrench getting greasy 14 hrs a day since he was 2, all by himself ...
1:56 'No, Marvin, this time it actually works'
.
20:30 'Proc...' Well, Procyonidae is a family which comprises raccoons. Does raccoon have anything to do with panda? Panda is of the Ursidae family.
Seriously, what's the use of this toy?
I have discovered what is the core algorithm of A.G.I. in 2015. I'm still deciding what to do with it. Scientific journals are so bureaucratic. Any ideas?
Post it here, imagine the headlines: "World's most important problem solved on world's shittiest comments board"
make a blog,
Win the Alexa prize
Ya post it here
Why it is crystals clear until 16:54 for me I understand it very well and am afraid because it is easy I stop watching it isn't suppose to be
Scoot the computer 1 inch to your left please, it's driving me insane XD
A cellular automata with recursive hierarchical pattern recognition units is what it's going to be. Question is, who will put it all together first.
I don't think that made sense. A cellular automaton doesn't have units, they model complex systems using simple rules and they already are recursive.
I feel utterly useless as a human being.
I'll feed the monkey.
AGI AGI AGI! EVERYONE HODL AGI SINGULARITYNET TOKENS!!!!!!!!!!
I would say I'm pessimistic about artificial intelligence. But that would be misunderstood because it's not that I don't believe it can exist it's that I don't particularly want it to exist. In any case what mr. Wolfram is talking about in this video is not so much artificial intelligence as it is a metadata access machine. I don't see how a calculator for metadata is a piece of artificial intelligence. I just don't see how that is. Artificial intelligence to me is something that exists as a video game opponent for example. That is, single player.
Interesting that some programs that are very efficient are impossibly complicated. This is opposite of what fundamental physics hope for in their quest for the secret of the univers. Why should the univers be simple.
Imagine one day, we are running one the universe like this, and when running we get to the point of us watching the simulation. But, we can't go further because it doesn't exists yet. We'll be pretty fucked !
What Newton did was playing with Descartes's 'XY Play Station' and stole the Credits for The realized plays.
Of course, The Primate with the more spread publishing apparatus tends to steal the credits from the Source Code developers. ( Classic Ape's behavioral Patterns)
I’ve never seen a student teacher who looks like an express manikin
Grey looks so much better on Lex than black.
Ingenious data management skills man. Keep up the good work. Jesus Christ Loves you all..
Poly Mathematica
Leibniz 2.0
"Can you explain why you didn't come prepared to class today, Mister Wolfram?"
Will Wolfie try to stole Juergen's credits for 'The Master Algorithm'?
Why that obsession to put himself over everything that he talks??
In talking Primates, When/Where A 'great' Undercover and Simple Core Idea is Showed, Accepted, ... and not yet going for implementation ... Narcissists tends to think: Oh, That Great Idea was thought by Me when I was a Kid , therefore, It is 'My Idea'.
(Disturbing, Don't Sync The system with that being. We don't like Hungry Worms at The ISA's Core.)
The most shocking truth in this video is that pandas are carnivorous.
This was pretty good except for the mythology.
He ran into the Transcendence Movie Problem - The program doesn't have free will to purse its interests.
Mr., The only reason you could build a version of you from all the metadata you've collected is because you have, over time disciplined yourself to fit into that particular box. This has happened to you over time and by natural process of the life you chose to lead. Unfortunately for you, that is neither the most efficient nor the most correct way for a human to think (maybe not even for machines).
Computer science guys are so used to assuming that it is second nature to them. Well, at some point, you will have minimized out of existence something that really is essential to humanity. When, not if that happens, I wonder whether their will be enough real evidence around to analyze what it is and how to correct it.
It may have already happened and we don't know it. Modern life is filled with stupid redundancies and problems that never were a real part of life before and no one even discusses them. We're all so anxious to abandon human failings that we haven't the insight or the humility to preserve them. You, for example, are a geek. You're shedding your humanity like a snake sheds it's skin and I find it revolting. That is not important to you because I'm nobody, but I reason, and there WILL be another you and another me in this.
half man, half computer
Have you noticed his voice sounds a lot like Elton John?
The instinctive knickers natively strengthen because active formally fence beneath a mysterious fibre. fortunate, abaft insurance
He is clearly a brilliant mind with lot's of awesome knowledge of algorithms, mathematics and computing in general. But, this is not even an AI system talk more of AGI. With respect to that, I'm calling this presentation bullshit. You can't wake up in the 21st century and attempt to sell the world a supposedly 'intelligent' system that's clearly confined by design. We should first sit down and define what intelligence is. It's cool if he's calling his product an advance algorithm for processing data and whatnot - that'd be sensible. But this? AI? Funny yo!
this looks a lot like a game engine.
I'm sure Wolfram is a genius but he's way too grandiose. Alpha it's nowhere near what he touts it. It's an natural language parser and search engine with some extras
Stephen Lee it has a lot of sophisticated algorithms, if that's what you mean little extras. Do you even know what an integral is, for example?
These guys are nowhere near solving AGI
Getting there slow is more fun. Slow as in say 2 decades unless you are in your 70s or 80s
little demos and lot of talking
Is it just me or is this guy repeating himself a whole lot???
little demos and lot of talking