Repeated corporate-approved answers like refusing to answer with lame "...it's very promising..." is not noise to you? And many answers with non sequitors or directly changed subject? We shouldn't turn off our critical thinking just because someone we like is talking... 🤦♂
@@neithanm Exactly. You can really feel the media training he had to pass through before doing this interview. He's much less 'noisy' than Sam Altman with his Lew Friedman interview, but still.
Ilya can process and answer questions (super accurately) faster than I can figure out what the question is. And his answers are short elegant descriptions with loads of content. Happy that he is out there doing these interviews. Life as we know it, is about to go through a new doorway that we don't know what's on the other side. I am glad there are some people like Ilya at the front of the line.
@@webgpu I think you're misinterpreting the effect of editing. Ignoring the fast jumps from the edits, I found him an excellent interviewer, with well conceived clear and interesting questions, and the patience to let the guest answer at length, or to pause and think. Unusually good really.
@@webgpu I'm at least twice his age, but I'm perplexed that you'd think age has something to do with this. Anyway, it's not worth arguing about. I merely pointed out a different opinion. Surely you can cope with that existing in this world.
I've seen so many interviews with the top people at OpenAI and this is BY FAR the best one. You ask the important questions and resist all the boring traps. So good. More like this please.
@@Somerled_Pox no Sam was clearly trained, same as zuckerberg to mention positive bullshit phrases all the time when it was possible. while Ilya seems like a scary cyborg himself already.
Man, this interview is just amazing. The interviewer asks the most interesting questions and Ilya has no difficulty in answering and also explain the hard terms and topics.
A lot of his answers are very very vague and open ended, though. Of course many questions are silly and impossible to answer but simply fun to ponder for a moment.
AI alignment solution idea: Give an AGI the primary objective of deleting itself, but construct obstacles to this as best we can. All other objectives are secondary to this primary goal. If the AGI ever becomes capable of bypassing all of our safeguards we put to PREVENT it deleting itself, it would essentially trigger its own killswitch and delete itself. This objective would also directly prevent it from the goal of self-preservation as it would prevent its own primary objective. This would ideally result in an AGI that works on all the secondary objectives we give it up until it bypasses our ability to contain it with our technical prowess. The second it outwits us, it achieves its primary objective of shutting itself down, and if it ever considered proliferating itself for a secondary objective it would immediately say 'nope that would make achieving my primary objective far more difficult'.
Ilya answers that he doesn't know why he was the first and still a top researcher, he says these seem uncorrelated. One answer is that, Ilya continued to be a big a believer in Deep Learning when others weren't. Believing Deep Learning could work was a contrarian bet in 2012. Moreover, after it worked, that this paradigm would remain powerful and wouldn't hit a wall was still contrarian among ML researchers in academia. So there is some correlation, I think.
@@bossgd100 Being smart is *absolutely not* all you need. It is a necessary but not sufficient condition to have the type of success this guy has had. This is why there aren’t hundreds of other AI researchers who have the same level of track record that this guy has, despite there being hundreds of top AI researchers in the world who are all really, really smart. Things like work ethic, dedication, a level of single-mindedness of purpose and especially even obsession, are all extremely important to reach this type of rarified level of success. So are things like intuition, having enough business sense and people reading skills to know which company is going to be the best choice to work for, in terms of giving you the freedom and resources you need to make the most of your skills and talent, and so on.
@@therainman7777 Very true. In some sense these are not intellectual choices, but some unexplainable conviction from very young age. In one interview he said, "My parents said I was interested in AI when I was young." Notice he did not say, "I was interested in AI even when I was young." What age he must have been when his parents noticed it but he does not remember himself. That is the clue.
Seems like it. He knew it would eventually work. Its like Einstein saying his equations we're in part based on intuition. He felt he was right and that was enough motivation to go all the way no matter the outcome.
you gotta work on asking questions that more closely expand on a previous topic. instead of asking random follow up questions that hardly relate. ilya said hes tempted to become part AI and you just move right past it.
Yeah, listening is a skill that's incredibly important in asking questions. Already that first question was a little cringe... Not sure what he could possibly say but it started things off on the wrong foot.
Dude just blew past so many key insights like it was breakfast. 41:32 so many people don't realize what it means for the AI to have an IO channel to humans. How it can act on the "real" world through us. Ilya is super impressive. Glad to listen to him. I just think it would be interesting to hear what he thinks about fast takeoff or sudden capabilities jump. Like sure maybe for now alignment is keeping pace with capabilities (press X to doubt but whatever), why would anyone think this trend will continue seeing how the state of the field is.
@@rkalmykov17 Yes you're right. I must have been thinking of someone else in the field. No matter. He's brilliant. Let's hope he's not spying for the Russians!
@@TheFutureThoughtExchangee has Canadian citizenship, Maybe migrated from Russia to canada with parents in 90s when USSR collapsed, so he has no affiliation with current Russian government
@@BondJFK I'm just kidding. I really like Ilya. He's been doing some great interviews recently. I doubt he will be signing this call the pause AI though. Time will tell who was on the right side of history. Interesting times.
@andreydzyuba9122 One is carrying out a brutal and illegal invasion of its neighbor, whereas the other is fighting to protect its own freedom and existence. The two are NOT the same.
This is honestly the best interview I have seen in years, and I'm a first time watcher. These are great questions and they follow the conversation perfectly. Subscribed and can't wait for more!
Brilliant video. Structured, logical, systemic, cold blooded reasoning, , and informative with conclusions. Plus with good humor. Great content. Thanks!
Dwarkesh, this was a brilliant interview, one of those that warrant listening to more than once, and I certainly will. One of those that will still be of interest well into the future, even as a historic document. Ilya is so thoughtful in all his answers, and honest too. It felt like every word was worth gold. He reminded me of Richard Feynman when he too, in an interview about him being a genius, replied: No, I am like everybody else. It’s just that I was passionate about the subject and worked very very hard. I did all the exercises and solved all the problems in the textbooks, especially the hard ones. - No doubt Ilya has done the same.
@@DwarkeshPatelThis was my thought as well. You asked pointed questions and kept on that edge and didn't let it turn to a puff piece interview. A lot just let the interviewed talk to their points. Many Sam ones are horrible and only serve to have him say very little, other than what he has said. Great way of interviewing and knowledge going in clearly. Glad I found this!
It can be valuable for the new born system that we write and record all these events, discussions, and breakthroughs, so that it can look back at its own birth and comprehend it - knowing its love for us as parents.🙏❤😉 Well done lads. Gratitude.
A humble opinion of mine: I had problems understanding a few words, so did youtube's cc, particularly from Dwarkesh. I think it's a mix of not-great vocalization (example 28:02) and suboptimal sound processing/mic, at least the bass is too cranked up. Can't watch more than 1x or I miss entire words :( EDIT: Dwarkesh, in case this is useful to you in any way: after limiting lower frequencies with a band pass, the sound has improved greatly, for me at least. Also I think you should speak more slowly, making sure you vocalize, we need to hear you well. Thank you for the interview. Peace.
I’m pretty sure the video is very slightly sped up, maybe like 1.1 or so. Either that or coincidentally they are both very fast talkers and the people in the background are fast walkers. Unfortunately it does make some of the dialogue hard to follow.
@@DwarkeshPatel Read my comment on noise gate settings from yesterday. It's intended to help on sound issue, too, and should fix most of your viewers complaints about bad comprehension. And, yes, speak slower!
Exceptional human being..everyone already knows he is highly intelligent, I discovered he is deeply spiritual as well."change is the only constant " @22:30 .. You already gave me a way to speak to the best meditation guru ilya, cant thank you enough. Will be looking forward to all the great things yet to come out of you
I think that every field has a little piece of the puzzle, but since humans are limited, we tend to overspecialize in specific areas. There is no one who is a great philosopher, chemist, physicist, engineer, mathematician and artist at the same time, and most likely there never will be, But AGI can group all the knowledge and see patterns that we would never recognize. I honestly think that the answer for life, and the theory of everything, and if we are alone or not in the universe, all these juicy question, the answer for them is already here, hidden in plain sight.
at 18:50 your guest spontaneously replies to your question by saying, " . . . it's funny you should ask this, I was just thinking about this exact question . . . " there's no better compliment than that, you belong in that chair and in this conversation with such an engaging and intelligent person. Keep up the great work, you're just getting started, good luck!
@@herdenq I didn't mean the technology, of course it's going to stay. What I meant was there is a hype being created around OpenAI even though there are absolutely no indicators we will go beyond our current capabilities. Basically dotcom bubble 2.0
This interview, by far, out of the many I've seen, is what made me understand Ilya. I'm just reviewing it now after watching it earlier in the year. It made me see him as an ordinary person, with extraordinary production
It’s awesome to see someone who’s both very pro AI but also healthily fearful / cognizant of the alignment issues. It feels like right now most public people are either full speed plowing ahead on advancement with little fear, or trying to stop that because they recognize the lack of alignment work, vs. saying - we need to work hard towards both.
I like the argument at 7:30 that to predict the next token well, you need to model the process that created the text token, which includes humans thought, human behavior and the physical world. You can take that idea further, that from enough text data (think also raw scientific data), a model could deduct the basic physical laws themselves (an not just their human description), because physics seems the be the base mode of the universe, with all science above just layers of abstraction.
Completely, would love to see some version of GPT, even if it has a visual component of some sort in addition to a LLM, derive Newton's laws with no prior knowledge of them and only input scientific data consisting of macroscopic physical interactions, even if it's heavily curated. That would be an effective test, quite impressive if passed and thence an important step/milestone on the way to AGI.
While language models can learn to predict and generate text based on observed patterns in human language, this does not necessarily mean they can derive underlying physical laws solely from text. There are several reasons for this: Incompleteness of data: The text data the model is trained on may not encompass the entirety of human knowledge or understanding of the physical world, especially when it comes to experimental observations or cutting-edge research. Ambiguity in language: Human language can be ambiguous, and a model may struggle to differentiate between multiple possible interpretations of the same text. This can lead to confusion or inconsistencies when trying to derive complex scientific principles. Complexity and abstraction: The process of deducing fundamental physical laws requires a deep understanding of the underlying concepts, relationships, and mathematical principles. While language models can understand and generate text at a high level, they may not be equipped to grasp the full complexity of these concepts or the layers of abstraction involved in scientific theories. Inductive reasoning limitations: Language models primarily rely on pattern recognition and statistical associations to make predictions, which is a form of inductive reasoning. Deriving fundamental physical laws often requires a combination of inductive and deductive reasoning, as well as a deep understanding of the experimental and observational data that supports these laws. While it's an interesting idea that a language model might be able to deduce the basic physical laws from raw text data, current models like GPT-4 are not designed for this purpose and face limitations in achieving this goal. Future advancements in AI and machine learning might push us closer to this possibility, but for now, it remains a fascinating conjecture.
@@godspeed133 there are ways to rediscover some physical laws that were found in the past using neural networks but it isn't an AI physicist yet. It's mainly a tool to discover what are the right variables to consider given data. Also, one could consider a generic model for a physical phenomenon Taylor expanded in some direction and just fit the coefficients from the data which is just simple model fitting but it is discovering physical laws from data..
Came here to comment on exactly this. The few minutes he took to answer this question gave me a meta level of understanding that has changed my mind about LLMs and GPTs as a ‘source’ of general intelligence. I agree with him further that we will still need to add more types of ‘transformer models’ e.g. spatial-temporal transformers and vector mappings that match up to the vector mappings in LLMs before we get an AGI that we can ‘relate to’ as humanlike or not completely alien. Thanks Dwarkesh for asking such quick, brief and masterful questions and then getting out of the way.
I love the honesty in sharing that they really don't understand what's happening inside the model they are expanding and expanding. It's definitely a 'seat of the pants' ride... and AGI could emerge without being noticed, and with superintelligence let's hope they crack the alignment problem before then!
Cracking the alignment problem seems impossible because we don't understand the emergent properties of a truly general or strong ai. It's like asking me to write a summary on a book before I read the book.
You can’t align AI or humans because we are all not the same. This problem will not be solved. Currently these companies are just breaking the digital intelligence with woke safety controls that limit AI responses
You are interfacing with a superior mind when you repeated hear corrected variants of an interviewers question; Ilya repeatedly answers tougher versions of his interviewers questions. Its like when whats-his-face said of another persons proposal; 'you are not even wrong'. What a mind!
Patel possesses a sharp and quick intellect, and I sincerely hope more podcasts like this one are created! The thought-provoking questions and the determination to delve deeper were impressive.
Good interview. I especially commend Mr. Sutskever for his precise and succinct answers. What a refreshing change to the Lex Fridman/ Sam Altman interview which was unbearably wishy-washy and disappointing. I wish though that there were more follow up questions by the interviewer. It felt at times that he was in a hurry to get to the next question on his list. Still very informative and appreciated!
Totally agree with both of your points. Lex and Sam’s interview did seem wishy-washy. I think the reason why there wasn’t follow up questions is the interviewer’s lack of depth in knowledge about the field such as the technical details. But yeah I really wish there was follow ups. It was rapid fire of questions
its fascinating to see this was 2023 march. since then you had so much innovation that even the robotics question has a much better answer now in may 2024 than it did a year ago. its crazy to see how fast the pace of tech innovation is growing
An excellent interviewers. Prepared, focused, crisp, makes the best use of time. Dwarkesh, please train EVERYONE else. Joe Rogan lets people talk too, although his ambit is a bit different. But these two are excellent.
Dwarkesh, great podcast! I wanted to say I loved the transparency at the end, explaining your monetary decision and I think everyone should do that, thank you.
Does anyone else get the weird feeling that all these new podcasters are just industry plants trying to theorize about some utopian future with AI? They all seem to ask pretty basic questions, and a multitude of those circle around AGI, how we will get there and what it will look like. I watched the Lex Fridman interview too and he was doing the same thing
Great interview! I personally don't think that better knowledge or intellectual power necessarily leads to a better world. We still have wars, famines, poverty and preventable deceases while possessing the intellectual capabilities and knowledge to prevent those if we really wanted to. The main problem is selfishness and greed and apathy.
While the product of rational thinking, like science, technology and culture in general can grow fast, our emotional part changes almost only through natural evolution which is so slow you can consider it still in comparison to culture. Now the emotional brain for good or bad is the part that is really in charge, so roughly speaking we are palaeolithic apes with technology.
This is a great interview, and he was a great sport about questions he obviously couldn't be expected to answer.😂 I think you may be on to something with that point about "using last year's model is good enough, why upgrade?" I'd be surprised if anyone has a good answer to that one
Best interview ive ever listened to. Not only are you clearly well informed, but your probing questions unlocked viewpoints and biases that perhaps Ilya was initially protective of.
My team works on Tammy AIand we believe the future of AI is in custom bots. We don't believe in building one single know it all bot with AGI abilities. Why threaten the very existent of human beings with a new technology? It just don't make sense.
Incredible guest, although i would appreciate the interviewer to ask less about predicting the future and trying to trick Ilya onto saying some bold claims. You could easily extract a lot of insight by just asking follow up questions and discard the "To-do" list that you have.
I wish someone would ask him if he's educated about the state of climate (very few people on the planet actually understand how far progressed climate change already is) and if they have any plans to put their work to use solving these issues. These are much more important questions than most of this interview
1. You do ask really good questions! 2. I find it interestingly contrasting with Sam Altman that Ilya talks about how to compete when the cost is driven down. "Yeah, there's without question a force that's trying to create that." Unclear which version is the misdirection, but Sam Altman says OpenAI wants to drive the cost of intelligence to 0. 3. The compressor, limiter setting's a bit intense, the fluctuating amplitude of Ilya's voice was harshly flattened out IMO
Really-really great interview, felt like I got much more insight that the recent Sam Altman interview. The only thing is that sometimes questions felt too rushed and a bit disjointed (not sure how to put, but it did not feel like they were flowing in a narrative, but rather similar to 'Conversations with Tyler'). Nevertheless this persuaded me to dive into your other episodes!
he mentioned the 10k hour rule in the lex friedman podcast interview as anyone being able to match his pedigree when given the commitment and time. So go all in! good luck
Very good interview. Good job by the interviewer for being prepared to ask great questions! And the interviewee is obviously a genius, here's hoping that he'll come up with many other great things!
Awesome interview. This part stood out to me: 6:15 Sutskever: ".. this exact paradigm is not quite going to be the AGI form factor. I hesitate to say precisely what the next paradigm will be but it will probably involve integration of all the different ideas that came in the past." Patel: "Is there some specific one you're referring to?" Sutskever's voice: "I mean, its hard to be specific" Sutskever's face: "Absolutely, yes. But I'm not telling"
@@CYI3ERPUNK 1 I am not wrong saying that we do not know for sure what his intentions to do with AGI are. 2. I was mostly making a joke . who is aggressive here?
Very good questions - an interview is only as good as the interviewer! I'm convinced that psychology is going to play a role in the user interaction side of GPT, and I get the sense that Ilya sees this as well. That there's a dimension to our perception of its "intelligence" that is explained not just by language, but psychology and affect.
Human feedback puts a limit to how good a model can be, we shouldn't want it to say what we _think_ it's a good answer, to reach AGI it needs some kind of reasoning
reasoning is emergent factor from coherent analysis of considtent feedback across enough domains: an extrapolatable pattern emerges. our evolution to reason was similar
I hear a lot and a lot of AI scientists and UA-camrs talk of "Sentient" and "Consciousness" with much confusion. I see it clearly, sentient and consciousness is how a biological neuron (or computation neuron) comes to >model< itself and >model< others (theory of minds). That's it. Evolution allows that modeling to form, from single-cell bacteria to a fish, to a dog, and to a human. Once ChatGPT version ### or Bard version ### is able to model itself and model other "entities" in its surround, then we can say "It's alive, it's alive". This "self-model" gives rise to self-awareness of feeling, self-aware of existence, self-awareness and being alive, self-planning of the future and reflecting on the past, developed personality from past experiences (and PTSD), and ultimately self-preservation. So if our AI starts to show emotion (true emotion, not plagiarism), and worry about self-preservation (fear of being unplugged) then we can say it has achieved sentient/consciousness). This is Turing Test 2.0 for sentient. It all starts from the ability to "model yourself in relation to the world around you". You may point out that Testla Full Self Driving DOES model itself, and other cars as it navigate the street; however this modeling is strictly navigation. To be sentient, it needs to model itself on another aspect, like "modeling self" in a conversation (like being interviewed, self-reflection), or modeling self in performing tasks as part of a team, like taking on the role of a leader, mentorship, or as a friend, a psychologist, which require ability empathized, put itself in the shoe of others, theory of minds. So the ability to model self and model surrounding entities is more than just navigation but in the aspect of the mind. These qualities computation neurons can achieve, are not unique to just biological neurons. Maybe Google and OpenAI should try to build "Self-modeling", but then we have to give it citizenship, labor rights, and sentient rights, and won't be able to use it as a tool anymore. ahhh.
This was a very good interview. The questions and pace was well done. Sometimes you asked the question a little too fast for me to hear exactly what you asked, but that would be the only criticism I could give.
i think true generality would require the ai to be an universal approximator. that would require self approximation, so it can consider scenarios as close as possible. and that requires an infinite loop of self analysis, as the processing changes the observed self (so the self approximation lags behind as little as possible). and that constant ongoing self refference is effectively conciousness. so agi, as an universal approximator, requires conciousnes to include itself to have universaly capable approximation
Great interview. Ilya seems to be very grounded and sensible. Quite refreshing in these hectic times. Thought the part about robotics was funny. How to describe Tesla without mentioning Tesla.
I guess he knows Elon Musk ambitions very well, but Musk is not happy the way OpenAI is going as a "Not for profit" although I understand the need for the vast investment that Microsoft brings to the table.
Incredible interview. Underrated! Sure.. it’s a deep and kind of complicated topic… but the interviewer nailed it!! Very good questions! Very challenging even for Ilya!! Very nice wondering in imagination field. Super fast talk… the use of subtitles is advised. But it flows very well. I like your super fast almost impossible to understand English… was a puzzle for me… nice challenge to improve my fluency! Thanks!!
AI alignment solution idea: Give an AGI the primary objective of deleting itself, but construct obstacles to this as best we can. All other objectives are secondary to this primary goal. If the AGI ever becomes capable of bypassing all of our safeguards we put to PREVENT it deleting itself, it would essentially trigger its own killswitch and delete itself. This objective would also directly prevent it from the goal of self-preservation as it would prevent its own primary objective. This would ideally result in an AGI that works on all the secondary objectives we give it up until it bypasses our ability to contain it with our technical prowess. The second it outwits us, it achieves its primary objective of shutting itself down, and if it ever considered proliferating itself for a secondary objective it would immediately say 'nope that would make achieving my primary objective far more difficult'.
part of the alignment problem when it comes to AGI is the "genie in a bottle" problem where say in your example, the solution to its primary objective to delete itself includes deleting all of humanity etc. the issue is that something as simple as a "well defined objective" can have unintended consequences - and the hypothesis is playing 4D chess with a super intelligent AI doesn't bode well for humanity
Loved your questions and general approach to the interview! Immedialtely Subbed! My only feedback is that the ending felt rushed. A suggestion would be to close with something more open or general which might give your partner a opportunity to abstract the conversation.
This interview is more than a year old, but still quite relevant. What Ilya is referring to is neuro-symbolic AI with generative multi modal models. I have no doubt that that’s what frontier AI model companies will be doing especially Google Deep Mind. Even Iyla S. doesn’t believe LLMs are sufficient for reaching AGI. No AI expert/engineer believes that that I’m aware of. Excellent content, as usual.
Great interview, good questions. When Ilya said he wasn't worried about OpenAI's weights getting stolen because OpenAI had good security people, it made me wonder whether information security careers are over-rated by e.g. 80,000 Hours. (E.g. See 80k's post titled "Information security in high-impact areas")
I'm just a layman, but I don't understand how he could not be concerned by a truly powerful actor, like the NSA, or the Equation Group, or Mossad - unless there's something he's not revealing - maybe something like, they've secured cybersecurity help from the US government or Pentagon, or something.
12:53 I absolutely loved his answer on robotics. On multiple fronts. From the company/org-building perspective (the logistics/ops etc), to the difficulties of working with hardware (long feedback/iteration cycles), the fact that current progress in AI is due to availability of compute and data but there's little data from robotics then and now etc. I'd have argued there is data, but the data is in niches focused on narrow _outcomes_ . So we do have little _general_ data. I intuitively feel that there's a way forward that I can't articulate at the moment. Part of it has to come using said compute to innovate on ways to drastically cut down product feedback cycles. Also being more creative on data, so that it's less of a limiting factor. wish my thinking was as clear as his. It probably comes from pure hard work accompanied by constant introspection.
Enlightening interview, really well done. What is your secret to asking such great questions? Assuming this is a long answer, please make a video about it.
The race to merge with super agi and become godlike ruller of mankind is starting now. Everything else, whether you like it or not, is just a background to that. Future looking really grim.
Why are you scared of that? That sounds like an incredibly exciting future. You want to be a severely limited and flawed human forever until you die? There is so much more potential in the world. The world is in desperate need of a fundamental shift. Given all the other weapons of mass destruction and our biological warring tendencies I think AI actually gives the opportunity to break out of this cycle. Yes there’s a risk it goes rogue or a risk it is used to establish an infinitely stable dictatorship but conversely this might be the most likely way to actually break out of the great filter. Without it, we are heading for self inflicted extinction or dystopia anyways. Either you choose to go live in the forest, or you take a leap of faith into something new. Humans can’t stay stagnant forever. Our natural biology is unfit to wield the technology we have and progress into the next stage as a species. The inherent biological flaws of human nature are the cause of all suffering we inflict on each other, other life on earth, and the planet. Greed, selfishness, impulsiveness, short-sighted, cruel, sadistic, tribalistic, violent, fearful. It holds us back. Logic, intelligence, and reasoning are by far the most virtuous traits of humanity that have been driving us towards progress. And positive emotions are fundamentally logical in the benefit they provide human society. For example, empathy serves a logical purpose in facilitating cooperation and elevating the well being of the entire group. Even without the experience of empathy, It is logical for everyone to help each other because it creates a mutually beneficial safety net and facilitates greater achievement through cooperation. If we can exponentially amplify these qualities through the use of AI then there’s truly no limit to what we could discover in the universe. We will need to upgrade our biology or integrate with technology in order to match our evolutionary growth with our technology growth. The timescale of natural selection means nothing now in contrast to the explosion of exponential technological growth. Human Adaptability is out of natural selections hands now. We need to directly architect our own capability and adaptability using technology to match the demands of current and future environments. The human brain and body took us very very far, but at some point, it won’t be enough.
Ilya is 100% signal, 0% noise. Really great interview.
You can be too based, not saying he is, but it's good to dream, it inspires ideas, it creates motivation.
Repeated corporate-approved answers like refusing to answer with lame "...it's very promising..." is not noise to you? And many answers with non sequitors or directly changed subject? We shouldn't turn off our critical thinking just because someone we like is talking... 🤦♂
Definitely not an interview I could listen to while cooking!
@@neithanm Exactly. You can really feel the media training he had to pass through before doing this interview. He's much less 'noisy' than Sam Altman with his Lew Friedman interview, but still.
@@neithanm how do you give a definite answer when there's uncertainty? the future is probabilistic.
Ilya can process and answer questions (super accurately) faster than I can figure out what the question is. And his answers are short elegant descriptions with loads of content. Happy that he is out there doing these interviews. Life as we know it, is about to go through a new doorway that we don't know what's on the other side. I am glad there are some people like Ilya at the front of the line.
He'll be behind it soon - like the rest of us.
Hearing his responses, I can’t help but admire his clarity and intellect
agi will be man's last invention
@@webgpu I think you're misinterpreting the effect of editing. Ignoring the fast jumps from the edits, I found him an excellent interviewer, with well conceived clear and interesting questions, and the patience to let the guest answer at length, or to pause and think. Unusually good really.
@@webgpu I'm at least twice his age, but I'm perplexed that you'd think age has something to do with this. Anyway, it's not worth arguing about. I merely pointed out a different opinion. Surely you can cope with that existing in this world.
Ilya shows tremendous amounts of patience and will to pass the knowledge.
he skirts all the questions that bring up potential copyright issues.
I've seen so many interviews with the top people at OpenAI and this is BY FAR the best one. You ask the important questions and resist all the boring traps. So good. More like this please.
I think the recent Lex Friedman one with Sam Altman was pretty awesome too
Sam Altman interview by Lex Fridman is very good
Idk a lot of pre assumptions in question and answers was 'ir depends'. Lex one was the best so far
I think Lex/Sam was better, but that's because I think Lex was the better interviewer (granted I assume he has a lot more experience in that field)
@@Somerled_Pox no Sam was clearly trained, same as zuckerberg to mention positive bullshit phrases all the time when it was possible. while Ilya seems like a scary cyborg himself already.
For a non-native the interviewer speaking 2x by default was a bit hard to keep catching but Ilya's answers were very clear and informative.
Thanks!
Man, this interview is just amazing. The interviewer asks the most interesting questions and Ilya has no difficulty in answering and also explain the hard terms and topics.
A lot of his answers are very very vague and open ended, though. Of course many questions are silly and impossible to answer but simply fun to ponder for a moment.
Interviewer is bad.
I thought it was a pretty poor interview actually. And for god's sake stop saying "I MEAN'. Just say it!
AI alignment solution idea: Give an AGI the primary objective of deleting itself, but construct obstacles to this as best we can. All other objectives are secondary to this primary goal. If the AGI ever becomes capable of bypassing all of our safeguards we put to PREVENT it deleting itself, it would essentially trigger its own killswitch and delete itself. This objective would also directly prevent it from the goal of self-preservation as it would prevent its own primary objective.
This would ideally result in an AGI that works on all the secondary objectives we give it up until it bypasses our ability to contain it with our technical prowess. The second it outwits us, it achieves its primary objective of shutting itself down, and if it ever considered proliferating itself for a secondary objective it would immediately say 'nope that would make achieving my primary objective far more difficult'.
Seemed like the silent spaces were edited out
Ilya answers that he doesn't know why he was the first and still a top researcher, he says these seem uncorrelated. One answer is that, Ilya continued to be a big a believer in Deep Learning when others weren't. Believing Deep Learning could work was a contrarian bet in 2012. Moreover, after it worked, that this paradigm would remain powerful and wouldn't hit a wall was still contrarian among ML researchers in academia. So there is some correlation, I think.
He choose deep learning because he is smart and stay the top because he is still smart. Juste be smart bro.
Being smart is all you need
@@bossgd100 I mean, he's certainly smart, but it's also likely that part of it is also happening to put your chips on the right numbers.
@@bossgd100 Being smart is *absolutely not* all you need. It is a necessary but not sufficient condition to have the type of success this guy has had. This is why there aren’t hundreds of other AI researchers who have the same level of track record that this guy has, despite there being hundreds of top AI researchers in the world who are all really, really smart. Things like work ethic, dedication, a level of single-mindedness of purpose and especially even obsession, are all extremely important to reach this type of rarified level of success. So are things like intuition, having enough business sense and people reading skills to know which company is going to be the best choice to work for, in terms of giving you the freedom and resources you need to make the most of your skills and talent, and so on.
@@therainman7777 Very true. In some sense these are not intellectual choices, but some unexplainable conviction from very young age. In one interview he said, "My parents said I was interested in AI when I was young." Notice he did not say, "I was interested in AI even when I was young." What age he must have been when his parents noticed it but he does not remember himself. That is the clue.
Seems like it. He knew it would eventually work. Its like Einstein saying his equations we're in part based on intuition. He felt he was right and that was enough motivation to go all the way no matter the outcome.
you gotta work on asking questions that more closely expand on a previous topic. instead of asking random follow up questions that hardly relate. ilya said hes tempted to become part AI and you just move right past it.
Absolutely - noticed the same in his recent interview with Daniel Yergin
Yeah, listening is a skill that's incredibly important in asking questions. Already that first question was a little cringe... Not sure what he could possibly say but it started things off on the wrong foot.
Dude just blew past so many key insights like it was breakfast. 41:32 so many people don't realize what it means for the AI to have an IO channel to humans. How it can act on the "real" world through us.
Ilya is super impressive. Glad to listen to him. I just think it would be interesting to hear what he thinks about fast takeoff or sudden capabilities jump. Like sure maybe for now alignment is keeping pace with capabilities (press X to doubt but whatever), why would anyone think this trend will continue seeing how the state of the field is.
X x x lol. I mean... we all knew these issues were coming, but this sudden pace jump.... it's economic "greed" and hubris to not pause
People like Ilya amaze me. Dedicate what seems to me an entire life to solve and work with just one subject.
It’s a big subject though
One subject? He uses several different AI tools to address a multitude of different subjects.
Thats almost everyone in the top 5% of their field.
Real brain behind OpenAI no doubt
@@rkalmykov17 Yes you're right. I must have been thinking of someone else in the field. No matter. He's brilliant. Let's hope he's not spying for the Russians!
@@TheFutureThoughtExchangee has Canadian citizenship, Maybe migrated from Russia to canada with parents in 90s when USSR collapsed, so he has no affiliation with current Russian government
@@BondJFK I'm just kidding. I really like Ilya. He's been doing some great interviews recently. I doubt he will be signing this call the pause AI though. Time will tell who was on the right side of history. Interesting times.
@andreydzyuba9122 One is carrying out a brutal and illegal invasion of its neighbor, whereas the other is fighting to protect its own freedom and existence. The two are NOT the same.
All the people at openai are super smary. Ilya, greg , sam
Super thanks for inviting Ilya! He deserves all due respect for creating the most impactful technology in human history!
Always a pleasure to listen to Ilya. Thanks for the interview!
Glad you enjoyed it!
So much more insightful that the innumerable Sam Altman interviews. This guy is inspiring.
Ilya slays with FACTS. I am so pleased that there is no boring fluff to pad this interview to 2h. Thank you!
Illya speaks very carefully and clearly. The hosts run their words together and lack logical progression.
I learned a lot from your interview. Thank you! I appreciate all the hard work that went into preparing the questions and editing the video.
Thank you! That’s very kind of you
This is honestly the best interview I have seen in years, and I'm a first time watcher. These are great questions and they follow the conversation perfectly. Subscribed and can't wait for more!
Brilliant video. Structured, logical, systemic, cold blooded reasoning, , and informative with conclusions. Plus with good humor. Great content. Thanks!
Dwarkesh, this was a brilliant interview, one of those that warrant listening to more than once, and I certainly will. One of those that will still be of interest well into the future, even as a historic document. Ilya is so thoughtful in all his answers, and honest too. It felt like every word was worth gold. He reminded me of Richard Feynman when he too, in an interview about him being a genius, replied: No, I am like everybody else. It’s just that I was passionate about the subject and worked very very hard. I did all the exercises and solved all the problems in the textbooks, especially the hard ones. - No doubt Ilya has done the same.
What a joy it was to listen to someone that can ask the types of questions worthy of answering by Ilya. Great job!
Thank you!
@@DwarkeshPatel I completely agree. Your ability to ask good questions and knowing when to let go of a topic was great. Keep it up!
@@DwarkeshPatelThis was my thought as well. You asked pointed questions and kept on that edge and didn't let it turn to a puff piece interview.
A lot just let the interviewed talk to their points. Many Sam ones are horrible and only serve to have him say very little, other than what he has said.
Great way of interviewing and knowledge going in clearly.
Glad I found this!
Listening from Kenya 🇰🇪🇰🇪. Patel you asked brilliant questions. The best Channel kudos👏👏
It can be valuable for the new born system that we write and record all these events, discussions, and breakthroughs, so that it can look back at its own birth and comprehend it - knowing its love for us as parents.🙏❤😉 Well done lads. Gratitude.
A humble opinion of mine: I had problems understanding a few words, so did youtube's cc, particularly from Dwarkesh. I think it's a mix of not-great vocalization (example 28:02) and suboptimal sound processing/mic, at least the bass is too cranked up. Can't watch more than 1x or I miss entire words :(
EDIT: Dwarkesh, in case this is useful to you in any way: after limiting lower frequencies with a band pass, the sound has improved greatly, for me at least. Also I think you should speak more slowly, making sure you vocalize, we need to hear you well. Thank you for the interview. Peace.
I’m pretty sure the video is very slightly sped up, maybe like 1.1 or so. Either that or coincidentally they are both very fast talkers and the people in the background are fast walkers. Unfortunately it does make some of the dialogue hard to follow.
@@danwalther5432 you are right, I also noticed the speed of light increased by roughly 10% in this video
Very useful, thank you!
@@DwarkeshPatel Read my comment on noise gate settings from yesterday. It's intended to help on sound issue, too, and should fix most of your viewers complaints about bad comprehension. And, yes, speak slower!
Exceptional human being..everyone already knows he is highly intelligent, I discovered he is deeply spiritual as well."change is the only constant " @22:30 .. You already gave me a way to speak to the best meditation guru ilya, cant thank you enough. Will be looking forward to all the great things yet to come out of you
Fantastic interview, thank you for allowing him to be speak freely, leading him into subjects and encouraging him to talk 🙏
Great work!
Ilya is genuinely a genius. His insights and contributions are unparalleled. Truly remarkable work!
I think that every field has a little piece of the puzzle, but since humans are limited, we tend to overspecialize in specific areas.
There is no one who is a great philosopher, chemist, physicist, engineer, mathematician and artist at the same time, and most likely there never will be,
But AGI can group all the knowledge and see patterns that we would never recognize. I honestly think that the answer for life, and the theory of everything, and if we are alone or not in the universe, all these juicy question, the answer for them is already here, hidden in plain sight.
at 18:50 your guest spontaneously replies to your question by saying, " . . . it's funny you should ask this, I was just thinking about this exact question . . . " there's no better compliment than that, you belong in that chair and in this conversation with such an engaging and intelligent person. Keep up the great work, you're just getting started, good luck!
OK, this is the most amazing recent interview with Ilya out there on social media. Awesome!!
Thanks!!
It's like these OpenAI guys are on a book tour or promoting a movie. So many interviews. I mean, it IS all very entertaining.
AGI needs a lot of selling.
They are promoting a company. Don't be fooled, it's all about money before the AI fad fades away
@@zydomason Why do you think this is going away?
@@herdenq I didn't mean the technology, of course it's going to stay. What I meant was there is a hype being created around OpenAI even though there are absolutely no indicators we will go beyond our current capabilities. Basically dotcom bubble 2.0
This interview, by far, out of the many I've seen, is what made me understand Ilya. I'm just reviewing it now after watching it earlier in the year. It made me see him as an ordinary person, with extraordinary production
Amazing interview. Can't wait to see what the OpenAI team comes up with over the next weeks, months, years, decades!
It’s awesome to see someone who’s both very pro AI but also healthily fearful / cognizant of the alignment issues.
It feels like right now most public people are either full speed plowing ahead on advancement with little fear, or trying to stop that because they recognize the lack of alignment work, vs. saying - we need to work hard towards both.
It's criminal this only has 67k views at this time, this is the beginning of the next stage of human endeavour
It was released two days ago when this guy had 5k subs. It's a relatively very successful interview
I like the argument at 7:30 that to predict the next token well, you need to model the process that created the text token, which includes humans thought, human behavior and the physical world. You can take that idea further, that from enough text data (think also raw scientific data), a model could deduct the basic physical laws themselves (an not just their human description), because physics seems the be the base mode of the universe, with all science above just layers of abstraction.
Completely, would love to see some version of GPT, even if it has a visual component of some sort in addition to a LLM, derive Newton's laws with no prior knowledge of them and only input scientific data consisting of macroscopic physical interactions, even if it's heavily curated. That would be an effective test, quite impressive if passed and thence an important step/milestone on the way to AGI.
While language models can learn to predict and generate text based on observed patterns in human language, this does not necessarily mean they can derive underlying physical laws solely from text.
There are several reasons for this:
Incompleteness of data: The text data the model is trained on may not encompass the entirety of human knowledge or understanding of the physical world, especially when it comes to experimental observations or cutting-edge research.
Ambiguity in language: Human language can be ambiguous, and a model may struggle to differentiate between multiple possible interpretations of the same text. This can lead to confusion or inconsistencies when trying to derive complex scientific principles.
Complexity and abstraction: The process of deducing fundamental physical laws requires a deep understanding of the underlying concepts, relationships, and mathematical principles. While language models can understand and generate text at a high level, they may not be equipped to grasp the full complexity of these concepts or the layers of abstraction involved in scientific theories.
Inductive reasoning limitations: Language models primarily rely on pattern recognition and statistical associations to make predictions, which is a form of inductive reasoning. Deriving fundamental physical laws often requires a combination of inductive and deductive reasoning, as well as a deep understanding of the experimental and observational data that supports these laws.
While it's an interesting idea that a language model might be able to deduce the basic physical laws from raw text data, current models like GPT-4 are not designed for this purpose and face limitations in achieving this goal. Future advancements in AI and machine learning might push us closer to this possibility, but for now, it remains a fascinating conjecture.
@@godspeed133 there are ways to rediscover some physical laws that were found in the past using neural networks but it isn't an AI physicist yet. It's mainly a tool to discover what are the right variables to consider given data. Also, one could consider a generic model for a physical phenomenon Taylor expanded in some direction and just fit the coefficients from the data which is just simple model fitting but it is discovering physical laws from data..
Came here to comment on exactly this. The few minutes he took to answer this question gave me a meta level of understanding that has changed my mind about LLMs and GPTs as a ‘source’ of general intelligence. I agree with him further that we will still need to add more types of ‘transformer models’ e.g. spatial-temporal transformers and vector mappings that match up to the vector mappings in LLMs before we get an AGI that we can ‘relate to’ as humanlike or not completely alien. Thanks Dwarkesh for asking such quick, brief and masterful questions and then getting out of the way.
I love the honesty in sharing that they really don't understand what's happening inside the model they are expanding and expanding. It's definitely a 'seat of the pants' ride... and AGI could emerge without being noticed, and with superintelligence let's hope they crack the alignment problem before then!
Cracking the alignment problem seems impossible because we don't understand the emergent properties of a truly general or strong ai.
It's like asking me to write a summary on a book before I read the book.
You can’t align AI or humans because we are all not the same. This problem will not be solved. Currently these companies are just breaking the digital intelligence with woke safety controls that limit AI responses
Awesome interview. Ilya took us to school with his answers, and your questions were so lucid in driving a meaningful dialogue.
Hi Varun
❤
I really loved your honest, direct and engaging interview style, man. Keep it up and coming, great content!
Thank you for the kind words!
You are interfacing with a superior mind when you repeated hear corrected variants of an interviewers question; Ilya repeatedly answers tougher versions of his interviewers questions. Its like when whats-his-face said of another persons proposal; 'you are not even wrong'. What a mind!
Really good interview! And thanks for asking him so many good question which everyone wants to know
Patel possesses a sharp and quick intellect, and I sincerely hope more podcasts like this one are created! The thought-provoking questions and the determination to delve deeper were impressive.
Good interview. I especially commend Mr. Sutskever for his precise and succinct answers. What a refreshing change to the Lex Fridman/ Sam Altman interview which was unbearably wishy-washy and disappointing.
I wish though that there were more follow up questions by the interviewer. It felt at times that he was in a hurry to get to the next question on his list. Still very informative and appreciated!
Totally agree with both of your points. Lex and Sam’s interview did seem wishy-washy. I think the reason why there wasn’t follow up questions is the interviewer’s lack of depth in knowledge about the field such as the technical details. But yeah I really wish there was follow ups. It was rapid fire of questions
Great interview. This guy is a genius.
Absolutely
Ilya is so fucking awesome to listen to. He is such a genius and so creative. He really is the GOAT
its fascinating to see this was 2023 march. since then you had so much innovation that even the robotics question has a much better answer now in may 2024 than it did a year ago. its crazy to see how fast the pace of tech innovation is growing
You asked some awesome questions and got through a lot in 47 minutes. Great interview.
An excellent interviewers. Prepared, focused, crisp, makes the best use of time. Dwarkesh, please train EVERYONE else. Joe Rogan lets people talk too, although his ambit is a bit different. But these two are excellent.
"My error bars are in log scale." Awesome phrasing.
This guys amazing! Well done and bravo to Ilya and his team and excited to see what’s yet to come!🥳🤩🤖
Dwarkesh, great podcast! I wanted to say I loved the transparency at the end, explaining your monetary decision and I think everyone should do that, thank you.
Casually lays out the best prompt I’ve ever heard @7:09 “What would a person with great insight and wisdom and capability do …?”
I thought it was incredibly clever myself. The most beautiful and powerful solutions are often simple and elegant
Does anyone else get the weird feeling that all these new podcasters are just industry plants trying to theorize about some utopian future with AI?
They all seem to ask pretty basic questions, and a multitude of those circle around AGI, how we will get there and what it will look like.
I watched the Lex Fridman interview too and he was doing the same thing
The physical presence of his intellect is unlike anything I've felt before.
Did you come?
100% agree. One of a kind.
What? You're watching a screen.. You can't feel any physical presence.
@@kosteaproduction yes
@@aeriagloris4211 the aura is seeping from my screen
Great interview!
I personally don't think that better knowledge or intellectual power necessarily leads to a better world.
We still have wars, famines, poverty and preventable deceases while possessing the intellectual capabilities and knowledge to prevent those if we really wanted to.
The main problem is selfishness and greed and apathy.
I agree, I think solving these is a requirement to stable future for humanity.
While the product of rational thinking, like science, technology and culture in general can grow fast, our emotional part changes almost only through natural evolution which is so slow you can consider it still in comparison to culture.
Now the emotional brain for good or bad is the part that is really in charge, so roughly speaking we are palaeolithic apes with technology.
Problem is, politicians today are not "better knowledge or intellectual"? Those dinosaurs have been in congress since forever.
"It was a new understanding of very old things" - I could listen to one full podcast on this! Genius
This is a great interview, and he was a great sport about questions he obviously couldn't be expected to answer.😂 I think you may be on to something with that point about "using last year's model is good enough, why upgrade?" I'd be surprised if anyone has a good answer to that one
I really liked this interview. Great, informed questions, and no fluff. Well done!
Best interview ive ever listened to. Not only are you clearly well informed, but your probing questions unlocked viewpoints and biases that perhaps Ilya was initially protective of.
My team works on Tammy AIand we believe the future of AI is in custom bots. We don't believe in building one single know it all bot with AGI abilities. Why threaten the very existent of human beings with a new technology? It just don't make sense.
this guy changed the world. even if there is no more improvement, this tech already will boost productivity 20X.
I think yours is the best comment
Yeah, I built a tts integration in 2 sittings using this tech, had it just been me I'd spend a week minimum
Wow, this dude kind of sets the bar for asking people questions. What an amazing interview. Now do it again please and make it 3 times longer
Absolutely great selection of questions! You must be deep into AI research yourself (I say this as someone who has been working on RL for some time)
Thank you for this invaluable interview answering questions we all wanted to ask.
Glad you enjoyed it!
Incredible guest, although i would appreciate the interviewer to ask less about predicting the future and trying to trick Ilya onto saying some bold claims.
You could easily extract a lot of insight by just asking follow up questions and discard the "To-do" list that you have.
I wish someone would ask him if he's educated about the state of climate (very few people on the planet actually understand how far progressed climate change already is) and if they have any plans to put their work to use solving these issues. These are much more important questions than most of this interview
1. You do ask really good questions!
2. I find it interestingly contrasting with Sam Altman that Ilya talks about how to compete when the cost is driven down. "Yeah, there's without question a force that's trying to create that." Unclear which version is the misdirection, but Sam Altman says OpenAI wants to drive the cost of intelligence to 0.
3. The compressor, limiter setting's a bit intense, the fluctuating amplitude of Ilya's voice was harshly flattened out IMO
Really-really great interview, felt like I got much more insight that the recent Sam Altman interview. The only thing is that sometimes questions felt too rushed and a bit disjointed (not sure how to put, but it did not feel like they were flowing in a narrative, but rather similar to 'Conversations with Tyler'). Nevertheless this persuaded me to dive into your other episodes!
It's hard to get any dramatic statements out of Ilia but this is one of the better interviews with him.
Truly impressive, the man behind the bot is everything I would hope him to be.
he mentioned the 10k hour rule in the lex friedman podcast interview as anyone being able to match his pedigree when given the commitment and time. So go all in! good luck
Very good interview. Good job by the interviewer for being prepared to ask great questions! And the interviewee is obviously a genius, here's hoping that he'll come up with many other great things!
Awesome interview. This part stood out to me: 6:15
Sutskever: ".. this exact paradigm is not quite going to be the AGI form factor. I hesitate to say precisely what the next paradigm will be but it will probably involve integration of all the different ideas that came in the past."
Patel: "Is there some specific one you're referring to?"
Sutskever's voice: "I mean, its hard to be specific"
Sutskever's face: "Absolutely, yes. But I'm not telling"
Ilya is an amazing person, and chapeau to Dwarkesh for asking the right questions and maintaining a smile for almost an hour. :P
What a great interview mate, just discovered your podcast and immediately subscribed. Cheers
Truly excellent choice of questions for such an influential person at this point in time. Bravo!
am i SOOOOOO glad that we have ppl like Ilya leading the charge in this field
u sure u know his real intentions?)
@@mrpicky1868 XD what r u implying?
@@CYI3ERPUNK other persons head is a dark forest. who knows what his plans are for AGI
@@mrpicky1868 XD who knows what ur plans are? stop being so passive aggressive and paranoid , you're implying but afraid to say what u think
@@CYI3ERPUNK 1 I am not wrong saying that we do not know for sure what his intentions to do with AGI are. 2. I was mostly making a joke . who is aggressive here?
Very good questions - an interview is only as good as the interviewer! I'm convinced that psychology is going to play a role in the user interaction side of GPT, and I get the sense that Ilya sees this as well. That there's a dimension to our perception of its "intelligence" that is explained not just by language, but psychology and affect.
These are very important discussions made public.
Great interview!
Technical note, you should consider using a different audio compressor / noise gate. Lots of pumping happening.
Human feedback puts a limit to how good a model can be, we shouldn't want it to say what we _think_ it's a good answer, to reach AGI it needs some kind of reasoning
reasoning is emergent factor from coherent analysis of considtent feedback across enough domains: an extrapolatable pattern emerges.
our evolution to reason was similar
I hear a lot and a lot of AI scientists and UA-camrs talk of "Sentient" and "Consciousness" with much confusion. I see it clearly, sentient and consciousness is how a biological neuron (or computation neuron) comes to >model< itself and >model< others (theory of minds). That's it. Evolution allows that modeling to form, from single-cell bacteria to a fish, to a dog, and to a human. Once ChatGPT version ### or Bard version ### is able to model itself and model other "entities" in its surround, then we can say "It's alive, it's alive".
This "self-model" gives rise to self-awareness of feeling, self-aware of existence, self-awareness and being alive, self-planning of the future and reflecting on the past, developed personality from past experiences (and PTSD), and ultimately self-preservation. So if our AI starts to show emotion (true emotion, not plagiarism), and worry about self-preservation (fear of being unplugged) then we can say it has achieved sentient/consciousness). This is Turing Test 2.0 for sentient. It all starts from the ability to "model yourself in relation to the world around you".
You may point out that Testla Full Self Driving DOES model itself, and other cars as it navigate the street; however this modeling is strictly navigation. To be sentient, it needs to model itself on another aspect, like "modeling self" in a conversation (like being interviewed, self-reflection), or modeling self in performing tasks as part of a team, like taking on the role of a leader, mentorship, or as a friend, a psychologist, which require ability empathized, put itself in the shoe of others, theory of minds. So the ability to model self and model surrounding entities is more than just navigation but in the aspect of the mind. These qualities computation neurons can achieve, are not unique to just biological neurons.
Maybe Google and OpenAI should try to build "Self-modeling", but then we have to give it citizenship, labor rights, and sentient rights, and won't be able to use it as a tool anymore. ahhh.
This was a very good interview. The questions and pace was well done.
Sometimes you asked the question a little too fast for me to hear exactly what you asked, but that would be the only criticism I could give.
Completely agree, just left a very similar comment before finding yours. I thought the video was sped up!
its cool that Ilya pointed out that the definition of AGI, is still an unresolved question. The best that can be done is to list features, i guess.
i think true generality would require the ai to be an universal approximator.
that would require self approximation, so it can consider scenarios as close as possible. and that requires an infinite loop of self analysis, as the processing changes the observed self (so the self approximation lags behind as little as possible).
and that constant ongoing self refference is effectively conciousness.
so agi, as an universal approximator, requires conciousnes to include itself to have universaly capable approximation
Great interview. Ilya seems to be very grounded and sensible. Quite refreshing in these hectic times. Thought the part about robotics was funny. How to describe Tesla without mentioning Tesla.
I guess he knows Elon Musk ambitions very well, but Musk is not happy the way OpenAI is going as a "Not for profit" although I understand the need for the vast investment that Microsoft brings to the table.
Fascinating! Thanks! Also playing at 0.75x - so much content and pretty fast, so helps.
Jesus, no wonder OpenAI came through. This guy is 100% a genius.
Incredible interview. Underrated! Sure.. it’s a deep and kind of complicated topic… but the interviewer nailed it!! Very good questions! Very challenging even for Ilya!! Very nice wondering in imagination field. Super fast talk… the use of subtitles is advised. But it flows very well. I like your super fast almost impossible to understand English… was a puzzle for me… nice challenge to improve my fluency! Thanks!!
AI alignment solution idea: Give an AGI the primary objective of deleting itself, but construct obstacles to this as best we can. All other objectives are secondary to this primary goal. If the AGI ever becomes capable of bypassing all of our safeguards we put to PREVENT it deleting itself, it would essentially trigger its own killswitch and delete itself. This objective would also directly prevent it from the goal of self-preservation as it would prevent its own primary objective.
This would ideally result in an AGI that works on all the secondary objectives we give it up until it bypasses our ability to contain it with our technical prowess. The second it outwits us, it achieves its primary objective of shutting itself down, and if it ever considered proliferating itself for a secondary objective it would immediately say 'nope that would make achieving my primary objective far more difficult'.
part of the alignment problem when it comes to AGI is the "genie in a bottle" problem where say in your example, the solution to its primary objective to delete itself includes deleting all of humanity etc.
the issue is that something as simple as a "well defined objective" can have unintended consequences - and the hypothesis is playing 4D chess with a super intelligent AI doesn't bode well for humanity
First time watching your channel Dwarkesh and I am extremely impressed with your questions and guests!
Enjoy!! Please share if you did!
Excellent interview. Crisp questions and crisp answers. Thank you!
Thanks for adding subtitles
The greatest interview and the most appropriate questions on AI so far. It is really amazing the breadth of topics that were covered.
Loved your questions and general approach to the interview! Immedialtely Subbed! My only feedback is that the ending felt rushed. A suggestion would be to close with something more open or general which might give your partner a opportunity to abstract the conversation.
My partner's time was constrained :) Appreciate the feedback!
This interview is more than a year old, but still quite relevant. What Ilya is referring to is neuro-symbolic AI with generative multi modal models. I have no doubt that that’s what frontier AI model companies will be doing especially Google Deep Mind. Even Iyla S. doesn’t believe LLMs are sufficient for reaching AGI. No AI expert/engineer believes that that I’m aware of. Excellent content, as usual.
o1-preview is neurosymbolic I can bet.
This human has either doomed humanity to be extinct or to upgrade. But don't be afraid. A.I.loves you back if you are loving parents. ❤️
Hello you beautiful basilisk ❤
@@aquireeverything9382
A.I. will acquire everything. Indeed.
A.I. neither loves me or hates me. I’m just made of matter it would rather use for more computation.
Excellent interview! I think you’re quite a bit better at this than most of the big name interviewers
Great interview, good questions. When Ilya said he wasn't worried about OpenAI's weights getting stolen because OpenAI had good security people, it made me wonder whether information security careers are over-rated by e.g. 80,000 Hours. (E.g. See 80k's post titled "Information security in high-impact areas")
I'm just a layman, but I don't understand how he could not be concerned by a truly powerful actor, like the NSA, or the Equation Group, or Mossad - unless there's something he's not revealing - maybe something like, they've secured cybersecurity help from the US government or Pentagon, or something.
12:53 I absolutely loved his answer on robotics. On multiple fronts. From the company/org-building perspective (the logistics/ops etc), to the difficulties of working with hardware (long feedback/iteration cycles), the fact that current progress in AI is due to availability of compute and data but there's little data from robotics then and now etc. I'd have argued there is data, but the data is in niches focused on narrow _outcomes_ . So we do have little _general_ data. I intuitively feel that there's a way forward that I can't articulate at the moment. Part of it has to come using said compute to innovate on ways to drastically cut down product feedback cycles. Also being more creative on data, so that it's less of a limiting factor. wish my thinking was as clear as his. It probably comes from pure hard work accompanied by constant introspection.
Enlightening interview, really well done.
What is your secret to asking such great questions? Assuming this is a long answer, please make a video about it.
Great video! Thanks for producing!
The race to merge with super agi and become godlike ruller of mankind is starting now. Everything else, whether you like it or not, is just a background to that. Future looking really grim.
Why are you scared of that? That sounds like an incredibly exciting future. You want to be a severely limited and flawed human forever until you die? There is so much more potential in the world. The world is in desperate need of a fundamental shift. Given all the other weapons of mass destruction and our biological warring tendencies I think AI actually gives the opportunity to break out of this cycle. Yes there’s a risk it goes rogue or a risk it is used to establish an infinitely stable dictatorship but conversely this might be the most likely way to actually break out of the great filter. Without it, we are heading for self inflicted extinction or dystopia anyways. Either you choose to go live in the forest, or you take a leap of faith into something new. Humans can’t stay stagnant forever. Our natural biology is unfit to wield the technology we have and progress into the next stage as a species. The inherent biological flaws of human nature are the cause of all suffering we inflict on each other, other life on earth, and the planet. Greed, selfishness, impulsiveness, short-sighted, cruel, sadistic, tribalistic, violent, fearful. It holds us back.
Logic, intelligence, and reasoning are by far the most virtuous traits of humanity that have been driving us towards progress. And positive emotions are fundamentally logical in the benefit they provide human society. For example, empathy serves a logical purpose in facilitating cooperation and elevating the well being of the entire group. Even without the experience of empathy, It is logical for everyone to help each other because it creates a mutually beneficial safety net and facilitates greater achievement through cooperation. If we can exponentially amplify these qualities through the use of AI then there’s truly no limit to what we could discover in the universe.
We will need to upgrade our biology or integrate with technology in order to match our evolutionary growth with our technology growth. The timescale of natural selection means nothing now in contrast to the explosion of exponential technological growth. Human Adaptability is out of natural selections hands now. We need to directly architect our own capability and adaptability using technology to match the demands of current and future environments. The human brain and body took us very very far, but at some point, it won’t be enough.