#250 Pedro Domingos on the Real Path to AGI

31:00My bet is at the end of the day, you will need to combine these and not just ideas from two of them, but from all of them, right? How long that will take to happen is an open question. Now, you mentioned that I say there is necessarily one master algorithm. What I mean by that is not that you're going to have this. A lot of people do this, like I have a symbolic subroutine that I call when I have a symbolic problem, like, you know, retrieval augmented generation. For those of you who are familiar with it, is it really a symbolic subroutine in a neural system? Right. That is not that's a very shallow combination of the two.

31:37And, you know, I spent some time in the book explaining why that is not the answer. I really think, UD, the deep unification of the two, and a good analogy to this, reasoning by analogy, is electromagnetism. Maxwell didn't say the physical world is a program that has the electricity subroutine that he calls some of the time, and other times he calls the magnetism subroutine. That's not it at all. What he actually showed brilliantly is that they are the same force. It's a unification, not a combination. And I believe this is actually what's needed. And when AI is mature, as in any mature science, this is what we'll have.

32:18Now, there's an important point here, which is that algorithm, there's no single form that it has to take. And again, a good analogy here is Turing machines. Alan Turing discovered or invented the concept of a Turing machine, which is a machine that can do anything, which at the time was a very strange idea. like a sewing machine would sew and a typewriter would type. But what's a machine that both sews and types? But a computer, this is the essence of a computer. It's not a machine that does anything. But there are hundreds or probably even thousands of different things that are what are called Turing complete or Turing equivalent.

32:53His actual machine nobody uses. We have a von Neumann computer. What you have in your cell phone and we all have are von Neumann computers, which is an architecture. So the real Turing machine is the von Neumann architecture, but conceptually they're the same. And there will continue to be new ones. So my point is that we need to arrive at the first form of the master algorithm. In some ways, it doesn't even matter what it is. Then there will be many variations that are good for different things. But in machine learning, if you will, it's inductive reasoning. We don't even have what they have for deductive reasoning, which is the concept of a Turing machine.

33:26And that's the first thing we need to get to. Yeah. And to get to these, you know, these reasoning models, whether they're unified or whether they're, you know, two systems, one calling the other, as they get more powerful, I mean, do you think that they will help solve this problem of how to unify the values? various tribes. I mean, do you have any confidence in reasoning models' ability to advance scientific research? Yeah. So there are many paths to the master algorithm, and people are following them. You could start with any two of these things, combine them, unify them, and then bring in the third.

34:20That's largely what I've done over the last 20 years. And so there are many ways to get there. The one that is most popular right now is to start with a connectionist system, with a deep learning system, and then bolt on symbolic reasoning capabilities. The way that has been done mostly so far is very shallow, and I don't think it's going to survive the test of time. But I do think, in very concrete ways, that there is a deep combination of them. But to take maybe a better example, think of transformers. A transformer is a type of neural network. A transformer is a neural network, but it's much more powerful than a multilay perceptron, which was the architecture that preceded it, in a way that there's many attempts to understand what it's doing.

35:05But I would say that at least my best understanding of what it's doing is it actually has some of the capabilities that symbolic AI has that previous neural networks didn't have. So in a way, a transformer is, in a way that we don't fully understand yet, a combination of connectionist and symbolic features. And that's what makes it so powerful. In fact, the closest thing to the mass rocket that we have today is a transformer. And if you think about it, right, a lot of people were skeptical when the book came out because it was before transformers came out. But transformers today are one algorithm, right, that does all of these things that you see in the media every day.

35:41It's remarkable. Yeah. Yeah. But on the idea of using AI to advance research, you know, basically transformer-based systems, do you see promise in that? or do you think that the AI itself won't be able to advance thought, that we're still going to need human intuition and creative thought and all of that? I mean, at some point, it seems like these models are becoming powerful enough and there's enough human knowledge encoded out there that they can ingest that there will be some creativity. So I think you have a number of questions there. So let me try this one part at a time. Creativity, right?

36:48People used to think that creativity was something that computers would never have. you know about more of x paradox which is this notion that the things that we think are easy for you are hard and vice versa yeah because the things that are easy for humans are easy for humans because evolution spent 500 million years evolving us and i have this slide that i've used in various talks that is more of x paradox easy versus hard and one of the lines that i have in there is easy creativity hard reliability and i would say to people reliability is hard creativity is easy and be like, what are you talking about?

37:26You're smoking. And even a couple of years ago, right? And these days I just rest my case. And I like, like creativity is like, well, you know, use whatever Dali or Midjourney, like generate videos, generate poems, generate music, just like, you know, whatever. Creativity is easy. Reliability, no one knows how to make an LLM reliable today. And that's the problem, right? So I don't think creativity, intuition, are there's nothing magical about them right we i mean i used to be a musician and write songs and people you know have this notion like writing a song is like some kind of magical inspiration that comes to you from the muse right it's not right in fact anybody can write an okay song writing a hit is really hard right but you know like if you spend you know whatever a year learning to play guitar piano and start playing you will write okay songs right so creativity is not magic So I don't think there's anything that humans do that AI can't.

38:20And most people in the community, this is what they believe. It might be very hard. It might take a long time. But unless you have some mystical belief about what goes on in the brain, it's a bunch of atoms. And in fact, if you believe in reductionism, then the mass problem exists because it's the one running in your head and mind right now. So I think at that level, there's no doubt that we'll get there. Now, where are we in terms of AI being able to, for example, do real scientific discovery? Could an AI, for example, come up with general relativity or solve the problem of unifying it with the standard model?

38:54And the answer to that is today we're nowhere close to that. And this is very interesting. In fact, people have remarked on this, that the application of AI across the sciences is progressing very rapidly. Right. Physics, economics, biology, et cetera. They're full of AI these days, but it's AI that does lower level stuff. It doesn't do that really creative, you know, things like people, Newton and Einstein and whatnot did. Right. And it's interesting because the LLMs have a bigger knowledge base. They've read every paper that's ever been written. So come on, like, where are the discoveries? Right.

39:29A human being with that knowledge base would be doing amazing things every day now. So clearly something is still missing. And our job, the researchers, is to discover that thing that's still missing. And I have this long running argument going on with Jan LeCun because Jan, he thinks back propagation is the master algorithm. He thinks machine learning will evolve and blah, blah, blah. But at the end of the day, the solution is still going to be gradient descent. right he's like a fundamentalist connectionist in that regard and you know and i asked him this question that he has no answer to he's like okay how did einstein come up with general relativity by gradient descent right now you're like there's no answer to that question so clearly something is missing and do you think what's missing is in one of these schools that you've defined Exactly.

40:25That is the right question. So for example, Douglas Hofstadter in his book that I mentioned, general relativity is one of these examples of something that was discovered by analogy. So clearly reasoning by analogy is important. And again, it's very interesting that Jeff Hinton, who's really the godfather of deep learning, he says, he's been saying this forever, that neural networks are better than symbolic AI because they reason by analogy. but Jeff where is the reasoning by analogy explain to me where it is now I I mean I can tell you like where I think the reasoning by analogy is happening in neural networks and and and at the end of the day we're going to have a single algorithm that in a way looks like a neural network but does reasoning by analogy and the symbolic reasoning and you know we could get into the weeds there but this is I think you know where the solution is are there are there is there active research on that in building reasoning by analogy into these systems?

41:29So, for example, there's a longstanding active area of research on what is called automated discovery. And it started in the 70s with people like Pat Langley doing theses where they show, like, look, this system rediscovers Kepler's laws or Boyle's law or simple laws like that. And actually, again, recently that's picked up and people have all this work on, you know, discovering differential equations and discovering, you know, how different systems work using AI. It's still, I think, at the level of Kepler, not at the level of Newton. The level of Newton requires, I think, some of this reasoning by analogy.

42:03And again, there are people like, you know, in psychology and cognitive science who have looked at, you know, how that might work, right? So there's a lot written on this. I don't think anyone has solved it. I also think that, disappointingly, in mainstream AI research, there's a lot of stuff going on, but not this. There aren't a lot of people going explicitly, like, how can we get a neural network to reason by analogy and therefore do scientific discovery? Which to me is a scandal, right? Like, you know, some fraction of people should be doing this instead of doing more tweaks on LLMs. yeah and also the evolutionary uh uh tribe or or school that sounds very promising I mean you don't have to start with random strings you can start with a system that's already very advanced and uh and go from there I mean are there people doing that I mean there are and and and what you make a very good point is like, you probably don't want to start with random strings.

43:07And fortunately, a lot of machine learning people have this machine learning and the connectionists and evolutionaries are both great examples of this. They have this fundamentalist machine learning attitude in that like, no, we want to learn everything from scratch. If you put in knowledge, you're cheating. The Bayjans and the symbolists don't have that problem at all. On the contrary, the Bayjans are all about priors, which is literally putting in your prior knowledge. And the symbolists are all about combining learning with knowledge-based AI, because that's what symbolic AI traditionally was.

43:37You start with the knowledge base and then you refine it, which I think is an excellent idea. So why throw that away? And you could think of what LMs are doing today is they are acquiring a knowledge base from texts. That's what they're doing in a very convoluted way. And then they're flexible about how they reason with that knowledge base. It's very opaque. And then clearly what's missing is the ability to reason on top of that text, which really is what things like Awan and so on and DeepSeek are trying to do. So there is a way to look at all of this and say, yeah, in one way or another, things are moving in the right direction and we will eventually get there.

44:14Yeah.

44:19It just seems that there is enough research out there now. and in all these different disciplines or schools that these AI reasoning models could or should be able to go across all of it and find analogies or go across all of it and find opportunities for evolutionary algorithms to advance what's already there? I mean,

45:10it's interesting because, I mean, so to take the evolutionaries to begin with, they are at this point the tribe that is most distant from the others. The other four, I mean, the support vector machine people, I mean, like all these people, they're mixing it up at this point. The evolutionaries, there's very little, but there are a few things like, you know, generative adversarial networks, as you very sharply pointed out last time, there is a flavor of co-evolution to that. So there is one path by which they could come in. There's also this whole area of multi-agent systems. So there's a type of reinforcement learning that is very close to ideas from generic algorithms.

45:51OpenAI at one point, before the whole chat GPT thing, they had these papers showing like, look, there are a lot of problems for which surprisingly, if instead of reinforcement learning, we just use a simple generic algorithm, it actually gets to the solution faster. So a lot of this is happening. Now, unfortunately, the problem is that like there's more ai research than ever before today like by an order of magnitude or two but most of it is along a very narrow front and you know one common view i would even say probably the prevailing view is that like you know we're just going to keep pounding on this and you know we're going to do so many things eventually we'll solve the problem i'm not so sure because you know the saying that you know nine women can't make a baby in one month right you don't get you You don't get, for example, to general activity by having a thousand random physicists just do what they do for a century.

46:43It doesn't work that way. One way that I often put this is like solving AI is not a sprint. It's a marathon. I really think somebody needs to go really deep. And there are people trying to do that, but I think not enough. And then once we do that, we will see how all these people are just spinning their wheels. and all of that unfortunately is going to go in the garbage can of history and like I know which part of those I want to be in you know I've been talking to people about quantum computing in the past week which is advancing and you know and the timeline is looking more not realistic I should say it's looking shorter to getting to a practical quantum computer.

47:35Do you think that will advance any of this? I mean, if you can, and not for just large problems that classical computing struggles with, but presumably with quantum computers, you'll begin to understand where quantum physics and Newtonian physics meet. Like what happens at the quantum level when a virus attaches to a protein? You know, right now it's very Newtonian the way people think about it. You know, it's shapes.

48:30But anyway, do you track what's happening in quantum and think about how that may address some of these issues? So the promise of quantum computing is that it can solve problems exponentially faster than classical computing. And if that ever comes to pass, boy, can we ever use it in AI. We will be the biggest consumers of quantum computing in the universe. No questions asked. So that is the number one promise of quantum computing. Now, having said that, there are a lot of caveats here. One of them is that if you talk to the people who are kind of like serious and knowledgeable of quantum computing as opposed to the people hyping various things, they will tell you that it's exceedingly unlikely at this point that there will ever be a general purpose quantum computer in the same way that there are general purpose classical computers right that's just not it doesn't look like it's going to happen there may be quantum computers for specific problems that are there that is the hope right and if those problems are are important enough then great and and those some of those problems are in ai so in fact there's one type of quantum computing that is about finding global optimal by tunneling out of the of the local ones right there's this company called d-wave that you know uh claims to have done this and yeah we could totally use that So that's the promise.

49:52Now, will that ever happen? And how soon will that be? We are still, you know, I roughly track, not very closely, what's happening in quantum computing, just out of curiosity more than anything else. I mean, honestly, I think it's a hard bet to lay because quantum computing is such a hard problem. And there's this whole notion like, well, there's all this computing in superposition and that's where the magic happens. But the error correction really kills you. The whole thing is so fragile. And making it robust is so expensive. You need a thousand qubits to actually have a robust one. And it goes on from there.

50:32You need super, super low temperature. Right. And and, you know, the first real, you know, quantum computer that does something useful is years away, maybe decades away. Right. And you could also make arguments, which some people do, about why this is never going to happen, that all of quantum computing is really a misunderstanding and a misconception and they're just deluding themselves. Or that it's a very nice idea at a theoretical level, which is where it began, but it'll never be practical. So, you know, as an AI person, my attitude to this is I wish them great success. I don't think we in AI, we will depend on that success.

51:08We, you know, in a way, AI is about a different path to doing exponentially faster computation. It's about being smart about your classical algorithms. And in fact, one suspicion that some people have, in fact, Demis Asavis was talking about it the other day, is that maybe, and I think this is a real possibility, we in AI, or in computer science more general, will discover algorithms that are smart enough that actually you don't need the quantum computing anymore, that exponential gain, we can already have it in other ways that, you know, just use classical computers. So we'll see. It's an interesting space.

51:42But I think, you know, for the most, and there's, I mean, there's a lot of, you know, there's papers on like quantum machine learning and blah, blah, blah. I should say there's another way intriguing one in which quantum computing might be relevant to AI, which is there are ideas. This is often how fields of research wind up having an impact is not in what they were trying to do because it failed, but because people came up with ideas that were then useful somewhere else. Right. And it could be that it could very well be that quantum computing comes up with ideas that in the end will actually be useful in machine learning.

52:16You know, there's almost nothing in this world that isn't potentially useful in machine learning. So I could see that happening with quantum computing. But so far, on either that front or the front of practical computers, I haven't really seen anything that I think people in AI need to be paying attention to. Having said that, I've heard rumors that this is what Ulysses Skiver is doing in his new company, is quantum computing for AI. So we'll see. yeah uh you know since you wrote the book uh ai has advanced dramatically and and very quickly and seems to not be slowing down there was talk like two or three months ago that everything was slowing down and you know i don't see it slowing down so um do you do you have in your head kind of a timeline for how close we're getting to a master algorithm uh great question and and you know my answer is we could be almost there or it could be very far away nobody really knows anybody who gives you a precise prediction is making it up or diluting themselves now uh here's why technology progresses in S curves, right?

53:35It's something slow progress and then fast progress, and then it slows down again and it plateaus, right? The early part of an S curve looks mathematically like it is an exponential. But then what people forget is that, you know, the slowdown is coming, right? You have an initial phase of increasing returns, the exciting one, and then a phase of decreasing returns, which is where most technologies are stuck for most of their existence, like cars and planes and TVs and whatnot. not. And now the thing that has happened in the AI in the last 10 years, clearly we've been on that upward curve, right?

54:07And every few years people go like, oh, things are slowing down now. In fact, I forget if you mentioned this already, but like I had this conversation with the illicit skiver at iClear in 2017, where he said like, oh, you know, deep learning is slowing down. Like, you know, there's no more progress. It's like, well, not so fast. Like, you know, new things will come up right and you know a month later the transformers paper came out so you know and there are people today i could actually argue both sides of this so like you could look at today and say like well but things are slowing down in particular the way you see things slowing down is that it takes exponentially more resources to produce the same amount of progress in fact the folks that you know open the end topic is like yeah yeah that's the way it is right you know like you gave me a billion dollars and now you have to give me 10 billion and tomorrow it'll be a trillion you know pony up which to me is alarming.

54:56But by that standard, things are slowing down. But again, that's normal. The question is, will there be a new idea that gives us another boost? For example, LXNet was one such idea. It was really just doing things on GPUs, but that's fair game. And transformers are another idea. And in many ways, you could say there hasn't been a big idea. And GANs maybe were such an idea. Again, it plateaued. But some people say there hasn't been any major progress in AI since Transformers came out, which is almost 10 years ago, or Attention, right? Attention is 10 years old now, right? So who knows, right? It really depends.

55:33This is not preordained, right? It's not like we're on some deterministic curve. It's like, you know, we, the researchers have to, you know, come up with ideas. And if we do, you know, the exponential will keep going until finally it saturates somewhere. The question is, is it about to saturate? Then will it saturate for a year or 10 or 100? My hope, right, uh uh is that no we we if i had to lay down my bets right i said like no we are not about to plateau we are going you know like this fast progress of the last 10 years is going to look slow compared to the next 10 right get your hand around that but this is not going to happen by magic it's really going to require uh major new ideas of which you know with all due respect to the guys doing o1 and deep seek those are not those are tweaks they're nice they're perfectly good work, but those are not the thing that's going to give us, you know, the next phase of the exponential.

56:26Yeah.

56:31Although, you know, and granted, Sam Altman is, you know, he's got a proprietary model that he's trying to protect and generate excitement around. But when you listen to him, it does feel like this reasoning is going to continue to develop. When you listen to Jeff Hinton, I mean, he already thinks these generative pre-trained models are conscious at some level. I mean, it's incredible the stuff he interpolates or sees in these models. And then you have, you know, Rich Sutton's team. I mean, Ilya's team. And all of these people are starting to jump ahead to ASI, to artificial superintelligence.

57:38intelligence, like don't even worry about general intelligence. Well, but I don't know. I'm a journalist and I've, I've seen what the field has done since I started paying attention in 2017 when I first met you. And you know, if we get to super intelligence, then basically we have the master algorithm, right? Well, once we have the master algorithm, we have AGI. I mean, by definition, we have AGI. Otherwise, it's not the master algorithm. And again, as people never tire of pointing out, once we have that, then just scaling it up, we can have 10 or 1 ,000 times or a million times the intelligence of a human being.

58:25Right. So that part is easy. The hard part is getting to that master algorithm. Right. But, you know, like you mentioned people like Sam Altman, Jeff, and et cetera. It's interesting because these are all different cases. Right. and they have different reasons to say what they do. And it's good to understand a little bit of that. Sam Altman is a very smart guy, but he's a VC. He's a great hustler. He's great at raising capital and sniffing out opportunities. He's really good at that, right? And persuading people and so on. But his technical knowledge, I think he himself would admit it, is not very deep, right?

59:02And I remember him saying like in an interview with Hoffman like years ago, Like, yeah, transformers aren't going to do it. He doesn't say that now because it would be inconvenient. But he didn't know what would, he said, like, something else is going to have to come along, but he didn't know what it was. And I think that's true, right? It's just not what you will hear him say today, right? I think it's easy for a lot of people, and I think he partly has fallen into that. You see what these algorithms are doing, and you get very excited and go like, oh, superintelligence is almost here. You have to remember that when you see, I mean, like, this is a lesson that you have to learn in machine learning.

59:36is like the algorithm always seems to be doing a lot more than it really is. You find examples where it's amazing. You're like, wow, you know, super intelligence. But then there's other examples where, you know, it does stupid things that a child wouldn't, right? And that problem is still with us. So you have to not get too carried away with that. There's also the sales pitch, but ignoring that, right? Now, Jeff Hinton, right, is a guy who is perennially over-optimistic about what neural networks can do. and you know more power to him because that's what kept him going for 40 years right whereas others gave up but he has always i think underestimate he has this notion that the like let me put it this way jeff does believe in the mass algorithm as does rich satin in fact i asked a bunch of people at the time i was writing the book and those were the two strongest yeses i got were you know jeff and rich of course they're different for them but i think jeff believes in a no, like what Jeff thinks of as the master algorithm and how the brain works is too simple.

1:00:37He actually, I think Jeff massively underestimates the true complexity of the human brain or even a simplified algorithm that would do the same. So every decade, and in fact, this is a well-known joke in the field, and he will say that himself, you know, in a self-deprecating way. He always thinks that like, yeah, I've just figured out how the brain works. We're on the verge of whatever consciousness, et cetera, et cetera. So in a way, his latest things about like chat GPT being blah, blah, are totally consistent with Jeff, but unfortunately, they're totally consistent with him seeing more than is there and underestimating the, you know, the length and the difficulty of the path to human level intelligence.

1:01:18And Rich Sutton in his own way with reinforcement learning instead of new networks, is also prone to that. I think he's, you know, at this, you know, like Jeff in some ways to oversimplify has been very successful and Rich hasn't. So in a way, Rich has learned the famous, you know, bitter lesson. And I think he's a little bit more, a little wiser now, if you will. But at the end of the day, I mean, I think, you know, let me put it this way. The founding fathers of AI were crazy, right? In the 1950s, we're saying we'll have human level intelligence in 10 years. They're crazy. They were madmen. But thank God for those madmen because they started the field, right?

1:01:59So in a way, it's good to have those madmen and those are optimistic people. But if you were an investor deciding what to invest in, I would take what people like Jeff and Rich and Sam said with a large grain of salt. Yeah. Yeah. Although, you know, you have models now that, well, and they're not single models, they're assemblies of models, but that can handle audio, you know, or speech. they can handle text, they can handle imagery, they can handle whatever other modalities are out there, and they can generate answers in all these different modalities. I mean, that's certainly more general than things were in the 90s or the early 2000s.

1:02:58So there is, I mean, AGI, it's not going to be a moment. There's a spectrum. And it seems to me we're in the early part of that spectrum where, yeah, you have these models that are multimodal and now have a certain amount of reasoning. so if you look at it that way there's a progression and we're moving along that spectrum Absolutely, so let me even make that point more strongly before making some caveats these aren't even assemblies of models you have some of these transformer based models that one model the same model does audio and video and text and speech and all of that, it's remarkable right this really is the closest to the mass rock and we've ever gotten and is a completely different place from where we were in the 90s like it just doesn't compare yeah i mean in the 90s i was doing a phd thesis on the sets with you know 500 examples trying to learn to do medical diagnosis and you know create assignment and things like that we're in a completely different place now right having said all that and you very you know you made the key point there which is agi is not a point is like AGI is, you know, human intelligence is a whole bunch of different abilities.

1:04:18In some of them, computers are already way better than humans. Right. They can add to numbers a lot faster than you like, and they can play chess better. But in other ways, they're still far behind. So there's not going to be a point at which you reach AGI or a point at which you reach super intelligence. You need to think at a final, like, in each of these dimensions, how far along are we? And in some of those dimensions, we are far along. However, and, you know, this is the big caveat, is like, where's my house bot? My hotspot is nowhere in sight. We're nowhere near AGI. You could say, oh, you've reached AGI once you beat humans at every single one of those things.

1:04:54Well, to take the twin to the spectrum, we're nowhere near there on producing Einstein's. We have these systems that know more than anybody ever could, but no Einstein. Or, you know, a MAID, right? MAID is an incredibly sophisticated system that no AI can mimic right now. making the beds, you know, you know, loading the dishwasher. Like we don't have that. We don't have that. And we do not have, I mean, I talk to robotics people, not, I'm not a boss, but like, so, you know, what's happening. I'm like, no one has a, you know, despite what you might hear in the media, you know, like we do not currently have a path to having a house bought in your house anytime in the next, whatever, five or 10 years.

1:05:35Right. Right. Okay. Okay, well, we're up to over an hour. Let's leave it here. And I really enjoyed this series, and I'm hoping listeners have learned a lot. Are you working on a new book? So I do have a couple of books that I want to write and that I'm making notes towards, right? But I did recently publish a book, 2040, A Silicon Valley Satire. The main focus of my work right now is doing research. So I do want to, you know, I have something that I think will make a big difference. And I want to get that ready and release it and then see where that goes. And then after that, I probably will be writing my next book.

1:06:18Create an oasis with Thuma, a modern design company that specializes in furniture and home goods. By stripping away everything but the essential, Thuma makes elevated beds with premium materials and intentional details. I'm in the process of reorganizing my house and I'm giving Thuma a serious look for help in renovating and redesigning. Thuma combines the perfect balance of form, craftsmanship, and functionality. With over 17 ,000 five-star reviews, the Thuma Bed Collection is proof that simplicity is the truest form of sophistication. Using the technique of Japanese joinery, pieces are crafted from solid wood and precision cut for a silent, stable foundation.

1:07:16With clean lines, subtle curves, and minimalist style, the Thuma Bed Collection is available in four signature finishes to match any design aesthetic. Headboard upgrades are available for customization as desired. To get$100 toward your first bed purchase, go to Thuma, that's T-H-U-M-A dot C-O slash IonAI. IonAI all run together, E-Y-E-O-N-A-I. So for$100 off your first purchase, go to thuma.co slash ionai. That's T-H-U-M-A dot C-O slash ionai to receive$100 off your first bed purchase.