Aza Raskin on interspecies communication, empathy, and AI’s next frontier

Hi, so the conversation you're about to hear as a conversation RAN had on stage, the Master's Scale Summit with Azeraskin, who is the CEO and co-founder of the Earth Species Project. If the Earth Species Project was successful, we might have this intro in, you know, "Bloogle Whale" or "Dolphin" or something else, but we'll show you part of the reason why this conversation was so magical and why we want to share it with you. Read and I have known Azer for a long time, and I think you can hear it in the podcast episode that we are just so delighted to hear everything from crows to beluga whales to elephants to dolphins. So I'm excited for everyone to listen. So here is our episode with Azeraskin. Welcome and it's awesome to be doing this at the Master's Scale Summit of having our possible podcast, and I can't think of literally a better guest and subject than Azer and the Earth Species Project. We won't need an AI to understand this conversation. Tell us about the Earth Species Project. What's what's what is it? The Earth Species Project, we are sort of the frontier of interspecies understanding. The question is, can we use frontier AI to understand the other languages and cultures of Earth? And what most people don't know, for instance, is that whales and dolphins, they have cultures, they have languages, they have distinct dialects, they've been passing down their culture for 34 million years. So as we can tell, so that's 85 times longer than the human species has existed. And in the last 20 years, science has progressed a lot in understanding these incredible other beings, that they have cultures that's now fully documented, but almost no progress has been made. In fact, no progress has been made on, say, the communication of Orcas. And just one, I think, really interesting side tidbit is Orcas have fashion fads. I don't know if you know this. A couple, yeah, I think it was in the 90s, first documented that an Orca killed a salmon and started wearing it as a hat. Orcas started following suit. And then for like a season or two, there was a whole salmon thing that was going on. I remember like salmon's in this season, that's like the next month, it's going to be aqua, we're ready. And we will have salmon hats for each, underneath your chairs. Yeah, so that's what we're up about. And it's sort of like there are two portions of this project. There's the science portion. And this is like build the tools. Right now, there's a scientific revolution obviously happening in every human domain, but that hasn't been brought to the non-human domain yet. And so there's a power, a scientific revolution. And that's what we're building these large-scale foundation models, natural M things, like that to do that. But then the even bigger goal is sort of create a Copernican revolution, which is to say that, you know, when we learn the language or the culture of another human group, like it takes that group from being something that's like out there and it brings them into your heart a little bit. And it changes how you act. And I think the same thing for animals. As we learn that animals actually have rich language and deep culture. It takes them from like nature out there and it pulls them into our heart and it changes who we are. So when I hear a dog bark, I hear a dog bark. And when I hear, you know, you go to the zoo, you hear a lion roar, but you guys are decoding animal communication. Can you just give us some examples of that animal communication and what you are seeing with these animals? Yeah, all right. So let's play a guessing game. Does anyone know which animal makes these sounds? Not bad. It is a kind of whale. Not an orca. Good guess though. Beluga. Yes. Nice work. So these are a couple of beluga communicating while hunting. And what I love about this is that most people like we sort of know humpback whale song, but this sounds to me like an alien modem. And in fact, all that digital high-pitched noise, that wasn't digital stuff. That was them communicating. And so it seems like they may actually communicate in modem like packets. They include their name. Have you done a bod-rate? That's a great question. And did you guys, I'm sure you played a lot with dial-up? Oh, yeah, of course. Yeah. Back in Bulletin Board days, we left our walkers outside. My, if I ever do an untalent show, is the ability to do like, which is how I'm going to communicate with AI, really. Exactly. Or how am I going to communicate with AI? That's right. But what's interesting about this, right? So here's an interesting puzzle. Orcas can communicate up to 150 kilohertz. Okay, we can hear up to 20 kilohertz, but in water, high frequencies don't go very far. So why are they communicating at such high pitch? And the answer is we don't know yet. But the theory is, is that because they are often communicating in close call, like intimate calls, and they're moving around each other, they're often touching each other. This is, I think you and I were talking about backstage, but we're doing work with University of Leone on these incredible crows. And they have their own unique culture where they do collective child rearing. So they sort of form a little commune or a couple or something. It's not a murder of crows. That's going to over great these days. They have their own dialect. They'll take outside adults in, teach them their new dialect. Language and their crows will start participating in the communal child rearing thing. And we, so we work with the University, they have little backpacks on the crows. And that means we get to record what they're saying all the time. And one of the things that we have been finding is around 70% of the calls that they make are actually quiet intimate calls. Calls they make to each other. And when we brought these to the scientists and biologists that we're working with, they were like, we haven't seen these calls before. It appears that something like, and again, this is all preliminary. So, you know, science, but 70% of the calls that crows make, science hasn't really touched or been aware of. And so, we've been trying to understand the language of another species by only listening to its chowts. They'd be like trying to understand humans by only listening to Twitter. They're only listening to Jim Kramer being really loud. And the same thing, like the same tools that we're now building for crows, to your point about scale. These methods can scale. Most people say like, well, you're starting with just a couple species, right? Like, actually, no. The way AI works is learning one about one species teaches you about all species. If I learn in human language, but teach an AI German and French and Japanese, it actually can learn very quickly Esperanto and Aramaic because each language has a kind of pattern that matches. And the same thing is true of the animal kingdom. And so, we're building these tools that work across the entire tree of life. And we're heading into these what we think are going to be sort of alpha-fold moments. We're going to be able to show that many species have names or many species have compositional language or many species have abstract thinking going from abstract to general or abstract to specific. And that's sort of like where we are. Yeah. And one of the things that I learned from early days when we were doing the, when you kicked off the Earth species project and it's like, well, what I year in, is you had discovered this, these trainers who had trained these two dolphins to do an original aerobatic trick. Yes. So, describe this a little bit because, by the way, this begins to give you a lens of that there are definite communication and thought that we very much undercount as human beings, because we are not yet in dialogue with them. Part of what you were just saying is, if we can get in dialogue, there is all kinds of other things that come. So, talk a little bit about the dolphins. Yeah. So, this was a 1994 University of Hawaii study. And in this study, they were teaching dolphins gestures. And the first gesture they taught the dolphin was do something you've never done before, which is to say innovate. I think it's a gesture read is trying to do it here. It was Jeff's yoga pose. And, interestingly enough, the dolphins can understand the concept of innovate, of do something new. And already, they require an immense amount of cognitive capacity because that requires remembering everything you've done before, understanding the concept of negation, not one of those things, and then invent whole cloth, some new behavior, some new trick. That's already cool. But then the researchers had this brilliant idea, which is, we'll teach them a second gesture, which is do something together. And then they would say to the dolphins, do something you've never done before together, innovate together. And the dolphins would go down, exchange sonic information, come up, and do the same thing they had never done before at the same time. I mean, I couldn't even do that with a human. I went under the water. It's like, you and me. Like, you would be a disaster. We should try this. Yeah, this is actually going to be like the ice breaking game for the rest of the scale next year. Yes, exactly. So I think it's so interesting that you can again, these species are so different. You're learning from crows, which are tiny to dolphins that are in the water, too. I feel like we've all seen that like beautiful heartbreaking meme of the elephant to like every year returns to the spot that their mother died. And that teaches us that there's so much more going on than we think. So when it comes to elephants, like, what have you learned about both communication, but then also like social group dynamics within the elephant, within the crew? Yeah. So I don't know how many people know this, but elephants like they can communicate infersonically. So sub like our ability to hear they will shake the ground and actually transmit sound through the ground and get picked up by other elephants through their feet. So there's like, there's a lot going on with elephants. One of our partners Joyce Poole runs the elephant listening project pointed out that when elephants communicate, they communicate on top of each other, they overlap. So you know, with humans, like we do turn taking, I speak, you speak, I speak, except for if you're a New Yorker, I guess. I did not bribe him in New York on that. Constantly in New York, San Francisco bad. But elephants do the opposite. It seems like roughly 90% of the time when one communicates the other starts to communicate. And so Joyce has this hypothesis, which is that maybe there's a kind of communal grammar that is in order to say a sentence actually requires all of us. And we don't know if this is true yet, but I love it as a thought experiment. Because if it is true, that means there is a kind of harmonization happening, not in music, but in meaning. And that's a way that the more of them that speak, maybe the greater coherence they have. And that's a lesson humanity needs to learn. Well, actually speaking of kind of humanity lessons, one of the things when we opened yesterday with Feifei, she was like, okay, there's more to the world than language and language rules. And so spatial intelligence. But here's another thing of what do you think we might imagine about how we're learning about language and human language by also exploring the languages of all these other species. I've never asked you this question before. But I know you're philosophical. So what are some like glimmers of possibility there? Well, I think this one of the elephants is one is that maybe you don't need to do sequential turn-taking to come up with meaning. Maybe there's a way. It's a little bit like music, right? When human beings play music, there's an emergence that happens and a way of creating meaning that is more communal. I think that is one. I think the real answer is we just don't know. This podcast is supported by Google. Hey, everyone. David here, one of the product leads for Google Gemini. If you dream it and describe it, VO3 and Gemini can help you bring it to life as a video. Now with incredible sound effects, background noise, and even dialogue. Try it with a Google AI pro plan or get the highest access with the ultra plan. Sign up at Gemini.google to get started and show us what you create. Often people ask us like, what is the first thing you want to say to an animal? I'm like, actually, that's ordinary, but I'm not sure if I can say it right now. I'm not sure if I can say it, but I'm not sure if I can say it, but I'm not sure if I can say it, but I'm not sure if I can say it, but I'm not sure if I can say it, but I'm not sure if I can say it, but I'm not sure. We don't need the wrong way to think about it. We're very careful. We don't say, break the interspecies communication barrier. We say, break the interspecies understanding barrier, because I think it's when we listen that we change. It's not when we speak. So that's the real rule. And just to give you a couple examples, these are dolphins looking in mirrors. Do people know about the mirror test? Mirror test. How do you tell whether something is self-aware, like a being? Well, one way of doing it is you take an animal, you paint a dot on them in a place that they can't see, you show the mirror, and if they start trying to get the dot off, well, that means they've done something very important. They've associated that thing in that mirror, that image, that is me, and a number of different animals pass elephants, for instance, do. But researchers thought for the longest time they didn't, and that's because they were using small mirrors. It's like, so human. And so these are, you can see what happens when dolphins look at themselves in a mirror, and they have what is maybe the most universal mammalian experience, which is you look in a mirror and you check out your abs. And so the realization here is like, what might animals communicate about? Well, they're clearly having the experience of self-awareness. So that's one of the most profound things, and actually dolphins have names they call each other by, even in the third person, elephants have names, that was discovered by Joyce Fuller partners in the last two years. Belugas do, they also include clan identity. So like identities, just parent parents, actually spend the first couple weeks of the chick's life, like leaned over whispering a unique name into each of their chick's ears, which they'll then use, again, for the rest of the life. Like, what's more core to identity than a name? So they may be, if they're communicating, they may be commuting out one of the most profound things that we have, and just give one other example. These are our lemurs, taking hits off of centipedes. So they bite down on centipedes, they enter these trans-like states, they get very cuddly. This is perhaps where Burning Man originally came from, and it turns out a lot of animals do this. Dolphins will intentionally inflate pufferfish, and pass them around in a circle to get high off of their venom, which is the original puff puff pass. And recently has discovered that chimps, so chimps will communicate by banging on tree trunks and strangler fig trees, and they'll use this to communicate for kilometers, and actually the human beings have drum languages, speaking of like what might we learn, they have drum languages where they can communicate over many kilometers just by drumming and say something very specific. They recently discovered that the name of the research just escapes me, but that chimpanzees will go inside of hollowed out boughbob trees, like the males will, and they'll start drumming, and this was confusing, why are they doing that? Well, it's confusing because, well, they go in, and it seems that they're doing some kind of like trance ritual, they're like they're hooding, they're panting, we obviously don't know really what's going on, but it's this very fascinating behavior. But one of the things actually also that I remember from an early one, and I think I'll get this right, we also had an ability to have a kind of almost like the mathematical proof that we were having a conversation with corvids. Say a little bit about what that discovery was, I know we didn't know what was being said, but there was a conversation happening. Yeah, so this was not with corvids, this is with another songwriter, Zebra Finch, and we're just at the place now where we can start investigating conversation, where we can model how like a bird responds to another bird, replace the bird with an AI, and start to see how do those conversations shift. And, you know, the interesting thing, and this is sort of like the plot twist of the whole project, is that we are able to communicate fluently, have a fluent conversation before we are able to understand what we're saying. Yes. It's sort of like the ability. Anyway, there's a certain amount of human beings that also have that, but yes. It's true. I don't know if you feel this way, but sometimes I feel like human beings, I'm like, I think you're sort of like an LLM with your temperature turned up too high. It's exactly, it's like, well, the temperature switch doesn't work. Yeah, but the thing that we're really discovering, this goes back to the crows, and also the belugas, right? It's like 70% of crow communication, just science has never heard, beluga communication is sort of the same thing. That is, when we're working with some of the best researchers, like Dr. Valera Vagara, who did the work to discover that belugas seem to have names, she has tags on whales, or whales, if you will. That's good. I've totally never used that one before. But she can't use a lot of her data because there are multiple whales that are overlapping at the same time, and she can't figure out who's saying what. And so she's only able, just to throw away 97% of her data, right? And this will be, you should really perk up, because this is the most vocal underwater species with what appears to be the largest vocabulary, and we're only able to look at 3% of the data as science, like the ocean is 5% explored, beluga is 3% explored. So why we think animal language processing is the next frontier, and we view this moment philosophically, sort of like the invention of the Hubble telescope. Where, you know, in 1995, astronomers pointed the Hubble telescope at an empty patch in sky, and there's a big scientific gamble, of course, what they discovered was not nothing, what they discovered was the most number of galaxies that we had ever seen in one spot before. And I think this is the general principle. When we have a scientific revolution, we get a new tool that can see it greater depths and greater perceptive abilities than ever before. Science then pointed at a space we thought was empty, and what we discover is not nothing, but we discover is everything. So I can imagine two criticisms coming sort of from opposite ends of the spectrum. And one could be, like, why are you messing with animals? Why are you trying to understand them? Why are you communicating maybe you're introducing some negative thing by sort of being there? And on the other end, I can imagine someone saying, sort of like, so what? Like, why? Like, okay, who cares? We know about animal communication. Like, we have so many problems. Like, why are we even worried about that? So what do you do when people say those two things? Before, is there a response? I think my response to the first one is woof. Yes, exactly. Come back to this a little bit. So, like, encryption, and we'll tell you what it is later. If we don't go to a little provocative for just one second, please. The way we treat animals is the way AI will treat us. So we need to increase our sphere of care as fast as we possibly can. And I do think there are these moments in history where we get a shift in perspective, and that changes more than just policy. It changes deeply our identity as a species, right? So Jane Goodall is like a hero, I think, of mine. And we had her on possible. Awesome. And she just passed away. And she was able to show that chimpanzees use tools. And that really changed who we thought we were as a species. We were always man humans, the tool makers. And now what really feels like uniquely identifies us, it's where humans, the language users, like we're building our digital gods in our image, coming out of language. And when in the history we've really encountered and taken the species from out there and into our hearts, I think of it as when songs of the humpback whale came out. And this is Roger and Katie Payne. It's this album where we got to hear as a Western society for the first time, the culture of another species. And it goes on Voyager 1. After on the golden record, it gets played in front of the UN General Assembly, becomes a multi-platinum album, I think, like three times platinum. You came there anchor for a star check one. Exactly. Go back in time, save the whales. So it changes culture and it's the reason why we have minkey whales and humpbacks today. We go to the moon, get those incredible images, earth rise, blue marble. And that doses all of humanity with the overview effect. And we get the EPA, modern environmental movement, NOAA. That was in the Nixon era. So there are these moments where a shift in perspective changes us. And I think that's what we're really after. And I just want to get back to what we're messing with animals. We have to be really careful and we have to think about the responsibility of what making new technology does. So you know, whenever we invent a new technology, we invent a new class of responsibility or uncover it, right? We didn't need the right to be forgotten until the internet can remember us forever. We didn't have privacy written into U.S. law didn't exist until that need wasn't there until Kodak started to produce the mass-produced camera. And then Brandeis, one of our most brilliant legal minds, created this idea of the right to privacy. So the same thing for communicating with animals, give one example, is Katie Payne, one of the people that discovered, actually, that elephants communicating with sonically, did songs of humpback whale. At some point, she played back the call of a matriarch for an elephant tribe that had already passed away. And all of the elephants came running over to find her and they spent the next couple of days in distress. Because they're like, where is she? Where is she? Humpback whales, for whatever reason, the Australian humpback whales are the K-pop singers? Of the Demon Hunters as well. I see a franchise for that. It's a new whale business model. And their song will go viral. And sometimes, if it's a particularly good song, much of the world population will start singing it within a couple generations. And if we're not careful, and we just take a synthetic whale and put in the ocean and have it sing, we might create, you know, viral memes, whale QAnon. We don't want to be like the Sora 2 of the underwater world. And so there is real responsibility that comes with the new power. And that's why we really focus less on communication and much more on understanding. Amazing. Well, we got to two of the 30 questions that we had. But as you can all see, what is an Earth species project doing is magical. Thank you. Thank you. Thank you. Also around the pause, it's just a deep thank you to you, Reed. Yeah. Because Reed, we've known each other since I was at Mozilla for many years, you were the first person I came to talk to about Earth species. Most people at that point, this is before Transformers anything else, would have just sort of like been like that's nice, pat me on the head. And Reed has supported from the very beginning. It's the reason I were able to do what we do. So I just want to say thank you. And please give him a round of applause. Big thanks to Dan Neelen, Eve Troll, Brian Pugh, Aaron Bostonelli, Jeff Berman, and the frickin awesome teams at WeightWatt and Earth species project. Hi folks, Paige Bailey here from the Google Deep Mind Debril team. For our developers out there, we know there's a constant trade-off between model intelligence, speed, and cost. Gemini 2.5 Flash aims right at that challenge. It's got the speed you expect from Flash, but with upgraded reasoning power. And crucially, we've added controls like setting thinking budgets, so you can decide how much reasoning to apply optimizing for latency and costs. So try out Gemini 2.5 Flash at aisstudio.google.com and let us know what you vote.