Oriol Vinyals: Deep Learning and Artificial General Intelligence | Lex Fridman Podcast

Oriol Vinyals is the Research Director and Deep Learning Lead at DeepMind. Please support this podcast by checking out our sponsors:
- Shopify: shopify.com/lex to get 14-day free trial
- Weights & Biases: lexfridman.com/wnb
- Magic Spoon: magicspoon.com/lex and use code LEX to get $5 off
- Blinkist: blinkist.com/lex and use code LEX to get 25% off premium
EPISODE LINKS:
Oriol's Twitter: / oriolvinyalsml
Oriol's publications: scholar.google.com/citations?...
DeepMind's Twitter: / deepmind
DeepMind's Instagram: / deepmind
DeepMind's Website: deepmind.com
Papers:
1. Gato: deepmind.com/publications/a-g...
2. Flamingo: deepmind.com/blog/tackling-mu...
3. Language Models are Few-Shot Learners: arxiv.org/abs/2005.14165
4. Emergent Abilities of Large Language Models: arxiv.org/abs/2206.07682
5. Attention Is All You Need: proceedings.neurips.cc/paper/...
PODCAST INFO:
Podcast website: lexfridman.com/podcast
Apple Podcasts: apple.co/2lwqZIr
Spotify: spoti.fi/2nEwCF8
RSS: lexfridman.com/feed/podcast/
Full episodes playlist: • Lex Fridman Podcast
Clips playlist: • Lex Fridman Podcast Clips
OUTLINE:
0:00 - Introduction
0:34 - AI
15:31 - Weights
21:50 - Gato
56:38 - Meta learning
1:10:37 - Neural networks
1:33:02 - Emergence
1:39:47 - AI sentience
2:03:43 - AGI
SOCIAL:
- Twitter: / lexfridman
- LinkedIn: / lexfridman
- Facebook: / lexfridman
- Instagram: / lexfridman
- Medium: / lexfridman
- Reddit: / lexfridman
- Support on Patreon: / lexfridman

[@lexfridman]

Here are the timestamps. Please check out our sponsors to support this podcast. 0:00 - Introduction & sponsor mentions: - Shopify: shopify.com/lex to get 14-day free trial - Weights & Biases: lexfridman.com/wnb - Magic Spoon: magicspoon.com/lex and use code LEX to get $5 off - Blinkist: blinkist.com/lex and use code LEX to get 25% off premium 0:34 - AI 15:31 - Weights 21:50 - Gato 56:38 - Meta learning 1:10:37 - Neural networks 1:33:02 - Emergence 1:39:47 - AI sentience 2:03:43 - AGI

[@olebilly]

Get Alex Jones ! Porfavor

[@willd1mindmind639]

A good problem would be for gato being able to dynamicallly create and update a dictionary of knowledge it has learned, where each term has text definitions, visual example such as images and video, to audio examples of the spoken word. And of course being able to dynamically translate it into different languages. If you can figure out out to do that, you are well on your way to "meta" learning. Because a lot of this today requires big data processes and tools to accumulate store and catalog.

[@Hexanitrobenzene]

Hello, Lex. Great interview :) Complete nitpicking, but your podcast playlist is missing five entries: 291, 287, 283, 282, 268.

[@briankennedy353]

Have a man named Hamilton Souther on the podcast.

[@nil42]

0

[@p-51d95]

Lex, your superpower is... asking the right question... at the right level... at the right time.

[@ZettaZone]

Super conversation. This type of guests(AI researchers) are the best! Please do more of that!

[@DeuceGenius]

agree! ai all the way

[@TheHumanPodcast.]

Lex, thank you for bringing us AI fans such great insights into the behind the scenes of DeepMind. Without people like yourself we’d miss out on hearing from key figures such as Oriol and Demis 😃👍

[@Sl33zytheclown]

if by AI fans you mean ppl that are getting an insight into the inner workings of the Cyberdyne terminator program that will doom humanity. Yea, we are AI "fans"

[@codydouglass242]

The podcast is really good information I never read before

[@amandajrmoore3216]

So true I recommend to all.

[@marcguarch3485]

Thanks for the interview. It is great to see a fellow Catalan being so successful

[@nilsdula7693]

🎗

[@ciudadanoanonimo5243]

@@nilsdula7693 He visto varios comentarios resaltando que es catalán... Si Oriol dijera públicamente: pues si, claro, soy catalán y soy español... y estoy muy orgulloso de ello (cosa que desconozco absolutamente, la verdad). Me pregunto si la mitad de los comentarios no desapacerían...?? Alguno con lazo me da la impresión que sería borrado... lamentablemente ya no importaría tanto lo brillante que él sea en su área concreta de investigación... Sorry for leaving this comment in Spanish to all those reading it and not understanding why, but it has a very reason for us to do it like this. Regards to everybody, and thank you for your patience and comprenhension 👋🏻👋🏻👋🏻.

[@marcguarch3485]

Simplemente lo comentaba como un signo de admiración a alguien que viene de la misma región de España que yo. Si hubiese sido andaluz, gallego o riojano hubiera hecho el mismo comentario. No es necesario hablar de nacionalismos aquí, ni el catalán ni el español

[@EduNauta95]

@@ciudadanoanonimo5243 En oriol vinyals es un catala de sabadell

[@ciudadanoanonimo5243]

@@marcguarch3485 No lo decía por ti, hombre, lo decía por el que politiza a una persona que solo ha venido ha hablar de su trabajo con un lacito amarillo. Oriol no ha hablado de política en ningún momento. Ni falta que hace si él no quiere o no viene a cuento. Es totalmente entendible que uno sienta alegría y orgullo de ver a un paisano reconocido por su trabajo, máxime si encima es catalán igual que tú. De lo que me quejo es de la gente sectarea que ni siquiera le ha preguntado a él su opinión, pero intenta apropiarse de su figura y su talento profesional. Yo creo que se entiende, no?? El lacito no pintaba nada con la entrevista ni venía a cuento, tan simple como eso. De todos modos, se agradecen tus palabras.

[@SLAM2977]

Very clear and to the point explanations from Oriol, no buzzwords and marketing, all very real and super informative, an example to follow.

[@ttff-bd2yf]

Not really? Lex would ask him a question then he would just say what is basically word salad for a few minutes.

[@tristanwegner]

Amazing podcast. Loved the clarity of thought in the answers. Looking forward to mindmelding with you guys post singularity!

[@mondeezul]

I love hearing your conversations while working, the morsels of information I gain from this makes me forever grateful for your platform. Thank you and keep doing what you do 🤙

[@ChaiTimeDataScience]

I listened to the previous interview many times! I already know this one is going to be awesome! Thanks Lex, for always sharing these gems with us!

[@EdeYOlorDSZs]

Yesss, I was waiting for another AI related episode! Last one with Demis Hassabis was phenomenal

[@maltsutty]

As a simple man , I am thankful for these conversations between very smart people while still using low level enough language that I can understand some of it . Great stuff

[@abdullahjames7038]

Appreciate everything you do🙏🏾

[@ryanmkeisling9089]

Another great podcast. I've seen every single one. I used to live at 519 Massachusetts Ave. there used to be a bar down off of Boylston st called bukowski, I used to go there often i even cooked there for a while. When I look back, I swear I saw Lex. A deep dive with professor Chomsky on this Ukraine crisis would be awesome to contrast with Kotkin and Stone. I was watching a KZbin of him talking about it and as usual his perspective was completely unique...

[@zanyarzohourian9398]

This podcast is incredible, many years from now people will appreciate what Lex is doing here.

[@wizerd2089]

Looking forward to this for the drive home through DC traffic later. Love your work Lex. Come home to Texas safe!

[@wizerd2089]

I have returned to say that this conversation was so hilariously beyond my understanding. But still, I listen.

[@M4ko1]

Great interview! Can’t wait to be on the pod here soon, keep up the awesome content brother 😎👍

[@djimiwreybigsby5263]

Lex, I'm grateful to you for bringing forward thinking to the internet; and I admire your ethos and skill set. Unlike others in this arena, every brilliant mind you interview says to you: "that's a good question"; and indeed, your podcasts take my mind to places that my lack of formal education typically prohibits ... Thanks again!

[@BossModeGod]

Lucky! Wish a bot would phish for me😔

[@dovienino4309]

Enjoyed your broadcast. And ,Thank you to .💕🕊️

[@1PercentPure]

Damn, lex really pulling out all the stops Thank you lex, love you

[@Omilliyo]

the personality is whats interesting, the uniqueness in each person. The viewpoints. The experiences that they have.

[@arvisz1871]

Great questions, great answers, great conversation. 👌

[@peterszilvasi752]

"I feel I am a bit of like a failed scientist, that's why I came to machine learning because you start seeing machine learning is the science that can help other sciences. I love science, I love astronomy, I love biology but I am not an expert and I decided that I can do better at these computers." This paragraph touched my soul... I was really into astronomy and I wanted to dropout from engineer IT specialist. Then getting involved in quantum physics, health, biology and neuroscience. Obviously, they are all immensely interesting however you can not be an expert for all those field. But you can get closer to them by machine learning. That's a beautiful idea and hope!

[@S1NX]

Been looking forward to this 🤜💥🤛

[@S1NX]

MiAU - MiAH = has many clones, who call themselves MEOW so you know. Sounds the same but it's NOT the REAL GAME 😉 Truth remains unbothered. It never complains. 🤫 #Justknow, Bro. They think I'm insane because I have faith in the human 🧠 building links connecting ⛓️'S. LEX FRiDMAN 💪😎🤳...remember the name. $@#❌️

[@emilylund794]

Wow!!! This is hands down the most compelling and intense opening question to begin an interview, lex you never cease to amaze me and I must meet you one of these days, or is this Deep mind posing as Lex????Hmmm

[@eaf888]

Love you lex!!! Thank you!

[@goldstar4556]

Thanks Lex, you are such a smart guy

[@morganmarston]

Have been listening to your podcast on Spotify. Your show and JRE are my favorites. Thank You Lex. Glad you are doing well in Russia, and have a safe return back to good ole America 🇺🇸

[@CryptoC4T]

What/who is JRE? Can I watch it on YT?

[@freddyfootstomps6557]

Oh hell yeah I’ve got an essay due on this tonight and just found content that can’t get caught for plagiarism 😎

[@manamsetty2664]

😂

[@MrFedX]

Going back to the roots with this one! :D

[@pattty847]

Lex: What makes an agent? Oriol: Capacity to take actions and take in new observations, recurse this function forever. Lex: Reminds me of "What is life" which is crazy difficult Oriol: Action, lets get back to gato tho

[@alexwillett2837]

Lex, great episode! I'd love to see @Yannic Kilcher on here some time.

[@DrDress]

36:37 The Holy Grail in philosophy that Plato, Kant or Wittgenstein never quite grasped.

[@laneseals9859]

@lex What are your thoughts on Large Video Models ?

[@tastytoast4576]

AGI is humanity’s path to accelerated evolution

[@tassospandis9642]

Thanks for Open Source Singularity

[@LockeLeon]

This interview is amazing. I love hardcore AI guests.

[@Kallut1337]

Maybe talk with someone about openAI in the online game Dota 2. It started as a 1v1 feature that destroyed most pro players. It later developed to play the "real game" 5v5 and completely smashed human players. It had something like 99,8% winrate against players. Pro players took big inspiration of the AIs understanding of the game and way of playing.

[@simonnilsson5356]

Thanks for the work Lex. Please try to decrease the lenght of the questions, thanks!

[@FluffyBuritoz]

Thanks you for teaching lex some Spanish there. He’s going to need it being in Texas.

[@michelcusteau3184]

Great interview Lex, please do more AI content like the good old days ;)

[@movax20h]

About working memory. Would something like multi level memory work? I.e. level-1: just what it recently said, and was said to it. 2000 words. Think like few last pages of the book. level-2: more abstract representation of the whole conversation sequence, think like abstracts of the whole book up to this point. level-3: core and repeating facts, things that in general are not removed, but constitute repeating details that are not captured by level-2, or could be removed from level-1 if we temporarily switch a topic.

[@ChrissiX]

Well done. I knew I understood the conversation when I was asking questions in my head that were either asked or answered minutes later. We are a long way from replacing you as the interviewer and further from replacing Oriol as the interviewee. IMHO --- Thank you again Lex.

[@sarveshahuja2385]

Man, i just searched "lex friedman ai" and this banger popped up in my search result !

[@akbaramanov6138]

Dear Lex! Please invite John Mearshaimer :) It would be very interesting to listen! Thank you in advance!

[@kevind4061]

Ty!

[@brianhamm8940]

Lex, are there any AIs out there that you can actually interview? It would be neat to see the progression of those Interviews with AI systems every six months similar to people reviewing Tesla’s autonomous beta.

[@jinoobaek]

Thanks!

[@penguinista]

Benchmarks act like skylights along the tunnel. Helps keep track of progress, keep you on track, and boost morale by being mini 'lights at the end of the tunnel' to break up the marathon.

[@LarsRichterMedia]

If there is some intelligent handling of AI by the wider society, this podcast will have played a significant role in it. Thank you Lex!

[@rockapedra1130]

Lex, you are the BESTEST interviewer with the FANTASTICMOST guests. Hmm... Feed that to a language model?

[@freddyjosereginomontalvo4667]

Awesome channel with awesome content and great quality as always say 💖🌍

[@sofias-tubular]

Timely podcast, and thanks for asking Vinyals about lambda. Top researchers ought to have the best possible intuition. If we think of the brain as a computer*, then universality says that it should be possible to represent it by using more of simpler structures. That is, a hierarchy of complexity is not necessarily needed to produce a sentient AI, if we assume that sentient intelligence is computable. There is a fun book about the birth and evolution of consciousness called Metazoa by a (philosopher) Godfrey-Smith. First instances of components we think core to sentience, such as distinction of self compared to environment, and "stress" as opposed to just pain, are already expressed in e.g. some crabs (which is why they are changing the laws regards to boiling them alive in some countries). Would've been nice to hear where Vinyals draws the line in sentience in the animal kingdom. *Not everyone agrees that it fully is, at least if they believe in free will.

[@PlanetBorne]

Hypothesis: Lex secretly made a sentient AI at home and the questions it asked Lex are the same questions he's asking in all his interviews.

[@FornoDan]

Lex is AI. He's the next generation Sophia

[@jasonwei7211]

When will you get Jeff Dean on the podcast?

[@Chemson1989]

Please do text to image AI like MidJourney/ Dall-E /Disco Fusion and how they'd affect the market of creative industry! Thanks!

[@anonony9081]

I want to hear Lex ask about the morality of including pain like signals in an artificial intelligence. You could design a robot to have feedback that tells it not to do certain things that would damage itself but at what point would we be able to equate that to pain like a human feels? And at what point would it be unethical to give them this feedback system since it would surely lead to great suffering.

[@OMGAnotherday]

✌️Extremely good question!

[@EngIlya]

and when it's unethical to let people go around without painkiller dose ;)

[@Evanderj]

Pain is a misleading word. It can’t feel pain. It would be incentivize vs avoidance paradigms, and the question of associated parameters & consequences, including the duration of a negative reinforcement consequence.

[@dmitrykazakov2829]

In machine learning with a teacher you have positive and negative examples. Negative example is an equivalent of pain. Furthermore, examples can be weighted etc. So it was unethical already 70 years ago. 🙂

[@estevetrias]

In my opinion the response is in the three laws of robotics.. Well, in the fourth one, that never appears; "The machine can't be conscious of the third previous ones...(The same for men, slaves, and so on....): en.wikipedia.org/wiki/Three_Laws_of_Robotics

[@kanishktantia7899]

Lex, you should really check out David bohem and jiddu krishnamurti about space time and attention

[@chockablock34839]

There are 18 states with the Atari input system, not 17. Eight directions of joystick, plus one more for centred. Double those 9 for having the button pressed. 18 states.

[@LakelandRussell]

I loved your question on whether AI is invented or discovered.

[@Radical_Middle]

funny thing is he doesn't even realize how his kind pushes humanity to the end

[@ruipedroparada]

Thanks for reaching out, noble KZbinr; I do not use WhatsApp or any such modes/platforms; keep up the good work, good luck, etc

[@richardpyles1682]

Maybe we should try a different approach like manipulating empty space and working back from that

[@Loveyouflakah]

I love how all of your guests l Give you space to come up with a question. Most times I expect an interruption with your “intermissions”. I can also see that you are thinking so deeply through a thought m, I understand the space. I’m learning how to interact. Its fun to catch the funny points of your personality. Lol you’re cool.

[@mathemystician]

Should AI necessarily be learning from our social media interactions?

[@a.e.1502]

I learned alot. Size does matter! 😁

[@lukekelly738]

I came to answers of the meaning of life and anything that's needs to be answered in this world are what you can imagine them to be .

[@codydouglass242]

How do you use a computer for learning is it microscope office or a computer software isn't that alot of data ?

[@AlphaGamerDelux]

if you can train a model to be trained, in turn being able to learn it higher more complex tasks, could one ask to create a better version of itself?

[@ernestboston7707]

“Why of course, Lex, we can create an AI to do that. Just think how we designed you.”

[@darklight9282]

Dear Lex a A.I robot hand crashed a boys finger while playing chess in moscow, why did this go wrong, did the robot AI Hand confused between human and board chess game etc, bad programming on the sensors etc

[@tracewaybos]

Oriol Vinyals, the George St-Pierre of scientists.

[@vickimiller6991]

Understand none of it but fascinated by all of it.

[@dewinmoonl]

holy cow first question 1:00 not pulling any punches!

[@ruipedroparada]

apologies for a third comment: is the nature of a model absolute? Or rather, is model-building driven towards an ultimate model? How stable does a model need be and for how.long? (cf. Wittgenstein)

[@dvntb]

Is it strange that Oriol is referenced on Wikipedia but doesn't have an entry for himself?

[@floridaLise]

Those who can imagine anything can create the impossible.

[@nilwccm123]

Damn I never thought I'd see a fellow Catalan peer in one of your podcasts, interesting af

[@tbpp6553]

Lex does not even name his sponsors and ask the audience to check the bio. What a badass 😅😅😆

[@ozzell]

Lex, please get Peter Zeihan on.

[@30803080308030803081]

Before I learned how ANN’s work, I didn’t know that they are trained once and then that’s it: they won’t “learn” anymore, they only do what they were trained to do. Knowing that this is how they work is a little underwhelming. I wonder whether it’s possible to program and use an ANN that never stops training: which perhaps has an initial phase of supervised or self-supervised training, but is deployed while still in a training mode, continuing to update and improve itself indefinitely in response to all new data.

[@frr5004]

I recall reading about some... reasoning, if not research, in that direction, maybe during the noughties, while the research into ANN's was still relatively young, commercially uninteresting and generally "on the fringe". Then GPU-based acceleration and "deep learning" and "deep dreaming" came along and nowadays that "continuous learning" aspect seems to be firmly under the carpet. Perhaps unless you talk to scientists who work on biological neurology (see the Human Brain Project). AFAICT, indeed, most "commercial applied AI" works along the lines of "train the model once, as training is compute-intensive, and then use the trained network, now fixed, on a massive scale, as that is relatively cheap." Hmm... Google seems to suggest that there is still some research going on, on continual learning: www.google.com/search?q=ANN+model+with+continual+learning

[@alejandrameza3968]

Of course you throw in that cheeky ‘size does matter’ comment. Haha that’s why we love you Lex.

[@Becidgreat]

1:00:00 there’s a flaw and hubris in a continued thought that the machine will learn at our pace. At some point it will be asked to improve itself or worse yet will improve itself without prompts 1:06:00 is that taught through loops? 1:09:00 I think alphago answered that right? 1:15:00 intention? Subjective Compound learning 1:28:00 right because once we really get moving there will be a TON of interest and monetary benefits and accolades and awards and then it’s almost too late to go back and fix those details. 1:35:00 sounds like there’s no prediction of what information is needed to create interest 1:50:00 I’m more concerned about religion and sentience. That war may destroy us …. Actually that may be the first fight or flight decision the sentient machine experiences. Holy crow! I just had a thought… this interview will be used for learning. Lex - what would you do if a machine wanted more information from you about you

[@mikeavery4098]

Information overload I think I need a zettapoch memory implant to obtain half of what you guys just said. Either that or a faster processor LOL.

[@alexkaa]

In digital audio sequencers, a 'Humanizer' function is a regular thing since decade(s)... 🙃

[@Becidgreat]

Is it me or is there a lack of future thinking about how we’re going to deal with this because we’re still so elementary and it seems silly to predict?

[@lukekelly738]

I ment to say I grew up on the streets and history and documentarys I used to love ..it was in my mid twenties o stoped searchimg for answers I still watch history I connect the pieces to where I can see the answer world mysterious queyit comes to me the answers when I least expect kt it's comes naturall bro woudk love to chat with you givew a buzz back if you can please 💯👍

[@ginogarcia8730]

Philosophy noob here.... I feel like someone who knows about Wittgenstein and something something Tractatus and Philosophical Investigations would be a good next interview after talking about words, symbols, languages, text, etc.

[@MyAnita1976]

Joscha Bach is like that.

[@ginogarcia8730]

@@MyAnita1976 true that, i finished those though. Need moar haha. And they're so dense, one has to keep watching them multiple times in order to finally get to the 1st step towards Joscha's level haha. And I love it.

[@elizabethgoltz4100]

An emoji is reference to a larger. idea, like a #hashtag or a global variable. Blind people use emojis and I don't think meaning is lost much there. Love this convo, thank you!

[@zanyarzohourian9398]

"Those who imagine anything, can create the impossible"

[@HanzDavid96]

I think for a superhuman intelligence, the reward function must be defined in such a way that it results from the transformer model itself. When the model produces an output, this output must be checked for various criteria by the model itself and the result of this check serves as feedback for the model. This can work if the model has already understood the world well enough through observation. For example, the model could generate a text on any topic and then be asked what the model thinks, how sexist the generated text is or how much logical contradiction is contained in the text. In response to these questions, the model should generate a numerical value for it's own output, which serves as a reward function. Perhaps it is also advantageous if different transformer models check each other in this way. But reinforcement learning or a comparable feedback strategy is necessary in any case, because a student cannot become better than his teacher if the student does not try out his own things and gain his own experience. However, these experiments and experiences can take place in the model's mind, i.e. in his thoughts, because it has a good representation of the world outside in its mind. Different Vectors of knowledge could be used to evaluate other vectors of knowledge.

[@romulus3345]

Google are NEVER ever going to admit that they possess a sentient AI. We will all find out when it's too late.

[@markrimkus200]

I agree (FWIW). It seems like a 'the lady [Google] doth protest too much, methinks" moment these days. Really hard to tell from the sidelines.

[@SinDigital]

What would you be trying to make if you made the universe? Open to vague answers.

[@Tiago_R_Ribeiro]

Very interesting conversation, but more importantly... are those plants made of plastic?

[@yeahright3733]

When a neural network is self-destructive.

[@olebilly]

We need Alex Jones on here

[@mattgraves3709]

He better not be one of those people: 'I don't think I can, so I don't want it'

[@mattgraves3709]

Wow, I was in a bad place when I started watching this :( switched to the first talk and enjoyed it very much

[@tonsetz]

Lex, would be great to bring Steven Pinker again!

[@peterszilvasi752]

Orial: I would argue size matters. Lex: I would argue size always matter. But that's a different conversation. 😅

[@TYFILMPRODUCTIONS]

I like the daytime interview setting better than the classic dark one. hopefully there will be more coming :)

[@77FINNBEAR]

When does the human become the tool for A.I?

[@77FINNBEAR]

Nano Nano! Earthling and Company.