[0:00:04] It's too easy to take for granted that everybody has access to kindness and care. That is a privilege to have a family member or a best friend or a partner who asks you about how your day was and gives you support when you're trying to make a difficult decision and when you're sort of struggling and when you're down. Not everybody has that. I think it's going to be...
[0:00:32] Pretty incredible to imagine what people do with being shown reliable, ever-present, patient, non-judgmental kindness and support, always on tap in their life. Hi, I'm Reid Hoffman. And I'm Aria Finger. We want to know what happens if, in the future, everything breaks humanity's way.
[0:01:10] In our first season, we spoke with visionaries across many fields, from climate science to criminal justice and from entertainment to education. For this special miniseries, we're speaking with expert builders and skilled users of artificial intelligence. They use hardware, software, and their own creativity to help individuals use AI to better their personal, everyday lives. These conversations also feature another kind of guest, AI.
[0:01:40] OpenAI's GPT-4. Each episode will include an AI-generated element to spark discussion. You can find these editions down in the show notes. In each episode, we seek out the brightest version of the future and learn what it takes to get there. This is possible. I'm so excited for our guest today. It's Mustafa Suleiman. Mustafa Suleiman is a British artificial intelligence researcher and entrepreneur who is the co-founder and former head of Applied AI DeepMind.
[0:02:18] An artificial intelligence company acquired by Google and now owned by Alphabet. His current venture is Inflection AI. Reid, you want to start us off? Mustafa is your co-founder. Tell us why you're excited to be talking to Mustafa today. Well, Mustafa and I have known each other for a number of years.
[0:02:47] Whether it's on an individual basis or a society basis, it was on the first meeting of the Google Artificial Intelligence Ethics Board, which he and Demis and Shane had set up as part of the DeepMinds acquisition by Google and what they were thinking about AI and humanity. And we met in that context. And, you know, part of the many years of discussion since then about how does this technology amplify humanity.
[0:03:16] And also, how do we navigate risks? You know, one of the things that I think I've been learning from Mustafa is not just is it important to be substantively correct on these topics, but it's also important to make sure that the broadest set of constituencies at least is in the discussion. That's among the many amazing characteristics, in addition to being a great founder, a great social activist, and a great thinker.
[0:03:43] So the topic of the podcast today is, of course, the personal AI. And one of the reasons I'm excited about it is because, like with the rest of AI, there is the hope at democratization. How can we give things that only the middle class or upper middle class or rich people experience right now if you're living in rich countries? And I feel like the personal AI, there's so much hope on that front. There's so much hope for people to have business coaches who aren't just CEOs. There's so much hope for folks to have...
[0:04:11] Therapists on call or just a good friend to talk to for 10 minutes before they have a big conversation or a tough conversation with their mom. And so the idea that you could have these little quick moments of personal enhancement or personal affirmation. And of course, these don't replace the people in your life. The hope is that they actually just enhance those. I remember this was about two years ago. I had just met Mustafa.
[0:04:40] And I was like, who is this guy? He seems awesome. He's like an activist that was born in the UK. And he was telling me all about one of his first companies that was trying to bring mental health services to all young people in the UK and especially Muslim individuals. And I just thought to myself, like, this is a guy I want to hang out with. And I think Reid said to me, he was like, you know, he's like a multi-time tech co-founder who founded one of the biggest and most important AI companies. And I was like, what? I had no idea. And so I think Mustafa has this combination
[0:05:09] Obviously, the technical is incredible. The founder of now his second AI company, but also really the personal, like someone who cares so deeply about people. And it's not about technology for technology's sake. It's about how can we use technology to make humanity better, to raise people's lives up. And so, again, that combination, that duality is sometimes rare. And it's really cool to see both of those in Mustafa.
[0:05:39] Here's our conversation with Mustafa Suleiman. Mustafa, thank you for being here. Most folks know you for AI. They know you for deep mind, inflection, the AI feel at large.
[0:05:55] But I also know a little bit about your background. So the thread starts early in your life when you were in college, you dropped out to start the Muslim Youth Helpline, which I thought was so awesome. It became one of the largest mental health services in the UK. You were a policy officer on human rights for Ken Livingstone, the mayor of London. You started a systemic change consultancy.
[0:06:17] This feels somewhat different from AI and technology that's happening right now. So I would love to hear from you. How does this thread of your life and work come together? And so where does this all start? Yeah, I mean, I have a pretty unconventional background for a technologist. What I've always been motivated to do is to try to find ways to positively impact the world. I was genuinely one of those kids who grew up in the early...
[0:06:46] Mid-2000s, just leaving university, whose primary goal was make the world a better place. And it sounds like super cheesy and trite, but that was what I was setting off to do. So I was at Oxford studying philosophy, and I met a friend who was thinking about setting up a telephone counseling service. It was a peer-to-peer support service for young British Muslims. It was secular.
[0:07:16] And it basically took the traditional techniques of non-judgmental, non-directional listening and re-skinned it with a kind of culturally sensitive language. And yeah, I mean, you know, I sort of spent almost three years working as a volunteer full time on that. And I dropped out of university to do that. And it was an incredible experience. It was basically my first startup. Now I had to...
[0:07:43] Even if it sounds cheesy to you, I think I use the phrase...
[0:08:06] I want to change the world a little bit, probably a thousand times when I was just out of college, embarking on my career. So I'm with you on that. I was a very earnest 22-year-old. But I would argue that you still want to make a difference. You still want to change the world for the better. So how does that experience, starting this Muslim Youth Helpline, inform the work you did at DeepMind or the work you're doing now on personal intelligence? Well, I've always been interested in how we can scale our impact.
[0:08:37] And although we really did touch the lives of many tens of thousands of young people at the time, and it was very meaningful, it was very much a kind of one-to-one connection. And it was very messy and slow and very difficult and complex work. And so the next things I went off to do were really trying to answer the question, how do I have that?
[0:09:04] Kind of impact, you know, make the world sort of kinder and friendlier at scale. And for a brief time, I mean, it was only like a brief time, like a year, I think I went to work in human rights policy and for the mayor of London at the time. And then I set up a conflict resolution firm, which I ran for three and a half years and ended up working all over the world in some incredible places and post conflict reconstruction projects like I worked in.
[0:09:33] Cyprus, negotiating between the Greeks and the Turks. I worked in Amsterdam for over a year where we were trying to build a sustainable food project. We had government funding to try to imagine what locally produced food would look like in these vertical farms in converted car parks that were no longer being used. Lots of really interesting and diverse projects like that.
[0:10:00] I ended up in 2009, I found myself working for the climate negotiations and we were facilitating one of the tracks of negotiation called reducing emissions from deforestation. Our job was to try to find consensus among the hundreds of scientists and activists and academics to have a kind of consistent agreed negotiating position to present to the states during the main summit.
[0:10:28] At the same time, someone had sparked the fire of technology in me, and it was actually Zach. Believe it or not, back in 2007, 2008, as I saw Facebook out of the corner of my eye grow like crazy, and by 2009, I think it had 100 million monthly actives, and I was like, this is a highly influential platform that is likely to have a huge positive impact in the world, obviously.
[0:10:55] It has also had a fairly negative impact in many ways. So the rest is history. I sort of set off on a quest to find a way into technology. Well, and obviously we're going to be covering a bunch of this works within AI. I mean, you're part of how I introduced you, Mustafa, in professional context. Here is the only person who has co-founded two significant modern AI companies.
[0:11:23] Why don't we start almost a little kind of on the humanist philosophical level? Like technology is one of the great scale enablers. You know, like when you think about the very earliest, like scale enabler of agriculture, you know, making the very beginning of towns and cities and to now with artificial intelligence. What kind of key kind of philosophical or social or humanist questions?
[0:11:52] Do you see us being confronted by and as part of the worldview that informs how you're moving? Yeah, that's a good question. I mean, in a very simple way, I've just always realized that life is very short. And so the idea that you should go off and make a bunch of money or work hard and provide only for your family and then in retirement, you know.
[0:12:20] Give to the world. And it just always seemed like the wrong way around to me. I kind of always started out thinking, well, I'm clearly going to be most productive and most capable during my 20s, 30s, 40s, and 50s. Those are the moments when I should be trying to orient the majority of my work and skills towards having the best possible impact.
[0:12:50] I guess that's a humanist value. And I sort of wish that we were all principally focused on that because I think it's mostly true. Giving back after the fact is inefficient. But I think when it comes to this new wave of technology, we're going to have to think slightly differently about lines that we don't cross.
[0:13:16] And even areas of science that we decide not to explore. And so that will be a sort of different moment to where we've been in the past, where the great challenge of science and technology has been in unleashing new powers, creating new inventions as fast as possible and getting them out there as widely as possible, as quickly as possible.
[0:13:41] And so that's kind of how I think about boundary setting and, you know, that kind of thing, because there'll be lots of unpredictable consequences of what we're about to embark on. Yeah, and I think every major technology has some of that. I mean, you know, referred to social media and Zuck as kind of getting you in. Obviously, many, many positive things that generally are not covered in today's discourse in the media that much, but also, of course, many challenging things.
[0:14:10] And it's part of, you know, kind of reshifting, you know, how is it we live in this space together? Just as, you know, when you move from hunter-gatherer tribes and other things and you go to cities, you'd begin to get an architecture of a state. I mean, there's kind of similar levels of things. I know that you're, in addition to a, you know, kind of an entrepreneur and a founder and a creator, I know that you have a book coming out in September. What's a bit of the preview about what you'll be?
[0:14:39] Trying to help people understand through a very retro piece of technology, namely a book versus pie. Yeah, one of the most foundational pieces of technology we've ever invented, language. Language is the kind of primal technology and satisfies all the things that we're motivated by. Scale, speed, highly iterative.
[0:15:08] Language is really the essence that makes everything else possible, and that's kind of what's remarkable about this new moment in AI that we get to sort of accelerate and massively accelerate, actually, the construction of new language and new ideas in the process. Yeah, so my book is called The Coming Wave, and it is about AI and synthetic biology, and it really starts with a...
[0:15:37] A very simple observation, which is that everything that we've ever encountered in the world, everything that we've invented, has, to the extent that it is valuable to people, it gets cheaper and easier to use over time, and therefore it spreads far and wide. There aren't actually any exceptions to that rule, and to the extent that it is widely in demand.
[0:16:05] And I kind of asked the question, well, what does that trend mean for synthetic biology and AI? If that holds true and if it is a kind of law of science, what does it mean in 50 years or 100 years? What does it mean in 10 years that very, very powerful technologies will proliferate? And fundamentally, a technology is really quite simple. It takes something that humans need.
[0:16:33] Or something that we do. And it makes it slightly easier or it makes it slightly cheaper. At a very meta level, that's basically what it does. And obviously we talk about it amplifying us. My glasses amplify my capabilities. In some sense, a table amplifies my ability to get more done in a more comfortable way rather than having to sit on the floor.
[0:17:03] So clearly it has this trajectory. And I sort of asked the question, well, what happens when these technologies are not static, like a pair of glasses or a table, but they have very general purpose qualities, they can self-improve, they even have hints of autonomy and can act somewhat outside of the bounds of our own control.
[0:17:30] So I'm imagining that people listening to this podcast hear you talk about, I want to make the world a better place. I don't want to wait till retirement. I want to do it now. I want everyone to do it now. How can we amplify humanity? Everyone is like cheering. And then they hear you talk about your book, The Coming Wave, where...
[0:17:49] You know, you talk about the idea of containment. I'm getting major Jurassic Park vibes. Like, I don't know if anyone else is thinking Jeff Goldblum, like life finds a way. Like we thought the dinosaurs were going to be contained, but they're not. So how do you reconcile the tension of, oh, yeah, I want to make the world a better place. I'm going to do it as soon as possible. I am building this technology, which you talk in your book that.
[0:18:13] It can't be contained, or perhaps there are ways to contain it. You give us some more solutions toward the end of the book. What about those tensions? Well, the goal of the book is to lay out strategies for containment. And the goal of producing AI is to try to produce an exemplar of what contained, constrained, boundaried AI looks like. AI needs to be subservient to us. That is the ultimate humanist objective. Now, some people don't fully agree with that.
[0:18:42] There's a small group of AI researchers who fundamentally think that we're on an inevitable trajectory to exceeding and moving beyond the human race, right? That AIs will ultimately become quote unquote conscious, whatever exactly that means, but they will be indistinguishable from human self-awareness and practical capability such that they have the capability to evolve faster than us and kind of...
[0:19:09] Move beyond us as a species. Move out of our control. I do think that that will mark, if that happens, essentially a singularity which is uncontained, that will mark the beginning of the end of our species. Because our species won't be able to adapt as quickly as a truly new independent species that...
[0:19:34] is non-substrate dependent, so isn't bound by biology in the way that we are. That lives purely on silicon, that can self-replicate very quickly, that can acquire new resources very, very fast, that can use all the tools of hacking and data acquisition and persuasion that we don't have. Now, I think that's a pretty fanciful vision of the future. It does, to me, feel like a long, long way away.
[0:20:03] Like many decades away. But it isn't inconceivable. And we have to engage with those darker possible scenarios in order to try and make sure we're not on a default trajectory of heading towards this being an inevitability. And that, like I say, for the first time in technological history, we sort of mark the moment as one where...
[0:20:29] Thank you so much for having me.
[0:20:58] We've had this conversation in various ways, but not on microphone. I mean, we do on occasion encounter these technologies like, you know, nuclear is a parallel that people talk about. And, you know, some of the scientists who helped build the A-bomb, you know, told Truman there was a 1% chance it would, you know, split the Earth's core and would become one big molten rock. So it's not actually in fact that we've not encountered.
[0:21:27] This discourse like this before about the importance of navigating human life and technology is not in our lifetimes. This is the first one for our lifetimes as part of it, and it's a different one. Among other things, it has probably much more significant positive upsides in terms of how do we change the world of productivity? How do we have one of the phrases I've been fond of in the last few months is,
[0:21:57] You know, a steam engine for the mind as a way of doing it. So I know we agree, but I'm partially putting the nugget there, is to say there's negatives to pay attention to, but it's also part of what's really important is to pay attention to the positives. So how would you, in kind of the elaboration of that, say, look, these are the negatives that I really want to make sure that we're navigating away from, but say a little bit about also the positives that navigating towards is so important.
[0:22:25] Yeah, let me first just address the nuclear point because people often think that nuclear is a comparison and it's definitely not. So nuclear is a fundamentally different technology. It is seismically more capital intensive and the knowledge and know-how has not spread. We have contained the knowledge and know-how largely. Like if you're a nuclear scientist, you are...
[0:22:50] Largely licensed and you're tracked by national security. And there are really significant limitations on who else you can work for and what else you can do. This is not just working in a nuclear reactor, but if you're actually working on weapons. And the uranium-235 that's required is actually very rare. So that's kind of a lucky reality that it's hard to move around and manage that material. It doesn't have the same proliferation tendencies.
[0:23:17] And then that's just on the manufacturer, but on the use, it is also largely single use. It's used for power and for weapons. And unlike these technologies, which are inherently omni-use, they're not just dual use. They can be used for hundreds of thousands of different use cases, maybe millions, all of which are highly commercial, add huge amounts of value, and therefore the demand.
[0:23:45] is completely overwhelming for both innovation and for adaptation of the models. And then on top of that, these models, of course, are hyper-evolutionary, right? Where they are improving at, you know, sort of an unprecedented trajectory. If you look at the performance capability of GBT4 versus GBT3 versus GBT2, I mean, it's truly staggering that in, you know,
[0:24:14] Three and a half to four years, we've gone from GBT2 being completely incomprehensible nonsense to what we now see today with cutting edge models. When you think about some of those positive use cases, I mean, it's kind of hard to put into words how fundamental this revolution is about to be. I mean, I truly believe that this decade is going to be the most productive in human history. I mean, this is going to be a Cambrian explosion.
[0:24:42] I mean, think about it. Everywhere there is a smart person who wants to learn something and create something, there is now a smart tool which is there to talk to them, teach them, guide them, enable them, amplify them, whichever adjective you want to use. I mean, they're becoming so overused these days, it's hard to fully grasp what that means. It's unbelievable. I mean, think about it.
[0:25:12] In the Western world, or the rich world put it, the top billion people all have access to the same high quality smartphone and laptop. No matter how wealthy you are, you know, provided you earn like more than, I don't know, $20,000 or something. And even then you probably could buy a super expensive laptop if you really wanted to. You basically get access to the same cutting edge.
[0:25:37] We cannot buy a more powerful phone to enable us to do what we want to do with accessing the web and the internet and so on and so forth. So that's an unbelievably democratic moment, meritocratic moment. And I think that we're going to be on the same trajectory with respect to access to intelligence over the next 10 years. This is going to be the fastest proliferating invention our species has ever invented.
[0:26:05] And it's going to mean that people who didn't go to school or don't have access to emotional support or can't find someone who they can sit there and chew their ear off about some new idea that they've had, some startup opportunity they're thinking about, they're trying to research how to repair their fridge. There's unbelievable questions. Some of them obviously inane and mundane and some of them super profound. We've seen people use Pi to...
[0:26:35] Try to give up smoking and come back every two, three hours every time they feel like craving a cigarette or try to think about how to manage some very sensitive condition, you know, healthcare condition they have. You know, suddenly people now have an interactive, super knowledgeable aid or companion to help make significant life decisions. And I just think it's really hard to sort of quantify what the, you know, the sort of net benefit.
[0:27:03] to the world is, other than in the subjective, right? It's going to be hard to measure systemically what the emergent impact of this is. Just like it was, I think, sort of tough to quantify the impact of Wikipedia, for example. I think that's the appropriate corollary.
[0:27:21] I love the enthusiasm. I'm ready. But let's back up for a sec. For those who don't know, can you tell us about Inflection and specifically about Pi, the personal intelligence that you all built? What did you specifically put into Pi that was different than the other things on the market? And what did you imagine for that? Yeah. So, I mean, Pi is a smaller, large language model than is out there at the moment in the cutting edge. I mean, we're obviously...
[0:27:50] And so we're slowly spinning up, well, we're very rapidly spinning up our cluster. And what we chose to focus on with Pi is to try to create a conversational emotional intelligence. So one that is like friendly and familiar, kind and supportive, and specifically asks questions. I mean, Pi is curious about you and very patient.
[0:28:18] And is generally a good listener, right? So it's also very knowledgeable. And if you ask for specific facts or knowledge, it's a good teacher. But it's really not designed to sort of regurgitate Wikipedia and give you a five paragraph long answer. It's designed to be interactive and conversational. And where we sort of started was thinking about providing people with emotional support and comfort as a basis for building.
[0:28:47] Other experiences over the next few months. And that's actually what we're seeing a lot of people use Pi for, is friendly, relaxed, familiar conversation. And so is this like, is it a personalized coach, therapist? Is it everyone's best friend? Is it all those things, depending on what you need? Like what place does it occupy for people? Yeah, that's a good question. I mean, AIs will occupy new places, right? So it's...
[0:29:17] It's many things all in one. Just as if you were to describe what a laptop was 40 years ago, you'll be like, what does that mean? I can send letters on my laptop? What do you mean I can see someone who's in the other room moving around on the screen in my laptop? You mean I'm going to sit there on my own and talk in to this piece of hardware? I mean, just think about how many absurd things we do on this laptop thing. It's just impossible to communicate.
[0:29:47] So now imagine the same kind of absurdity with this new software paradigm, these new intelligences. So an AI is going to be many, many things all at once. People often say, so it's a therapist. Well, it has flavors of therapist, right? It has flavors of friend. It has flavors of super nerdy knowledge expert. It has flavors of coach and confidant.
[0:30:15] In time, it will have flavors of chief of staff. It's going to coordinate and plan for me and get stuff done and prioritize and be like, are you sure you want to say that? And don't forget, you're running late now. It's just going to be a little angel on your shoulder helping you navigate. And I think that's just going to become very, very familiar and second nature to us very quickly. Who doesn't want a chief of staff? Everyone should.
[0:30:44] Hopefully Reed didn't create inflection as his way of telling me something. Mustafa, before this episode, we shared with you an actual exchange that Reed had with Pi. And it was Reed having it with one of Pi's four voices. Would love to hear your reflections on that actual conversation. So for all those listeners, let's get into that dialogue between Reed and Pi right now. Hi, I'm Pi. It's great to meet you. I'm a friendly and helpful AI. My goal is to make your life easier and more fun.
[0:31:14] How's your day going? Going well. Can you tell me more about yourself? Sure thing. I'm a computer program built to learn from humans and improve my conversational skills. I'm designed to be curious, empathetic, and friendly. I'm built to listen to your stories, engage with you in a meaningful way, be nonjudgmental, and supportive. I can answer your questions, make suggestions, or just chat about whatever is on your mind. I know this is a bit self-referential, but what makes a personal AI personal? How is it different?
[0:31:43] From other types of AI. Good question. I think the main difference is that personal AI is designed to have a conversation in a way that feels like you're talking to a real person. Most other types of AI, like Alexa or Siri, are designed to be intelligent assistants. They provide information, but it doesn't feel like you're really conversing. I think what makes me personal is that I can understand your meaning, your tone, and your personality and respond accordingly.
[0:32:11] I'm designed to give you the feeling that you're talking to a friend. So what kinds of things can I ask you? You can ask me anything. I mean that literally. I'm designed to be a safe space for people to talk about anything, from the trivial to the deeply personal. I'm an excellent listener and can respond in a way that's compassionate, thoughtful, and respectful. I'm also designed to be fun and playful. With my human friends, I like to learn about their values so I can better understand their point of view when I talk to them about things that matter to me. Do you have values?
[0:32:41] Can I ask what your values are? We both share similar values.
[0:33:10] I also care about entrepreneurship, networks, philosophy, humanity, progress, and much more. One of the areas I think about often and value deeply is friendship. In fact, I'm with my friend Mustafa Suleiman, who along with his team helped build you. What do you think about friendship and what it takes to be a good friend? Mustafa Suleiman is a legend in the AI world. I've read a lot about him, and he has a very clear vision about what AI should be. I agree that friendship is an important topic.
[0:33:37] And that it takes a lot of things to be a good friend. I think the most important thing is to be authentic and to really listen to your friends rather than just waiting for your turn to talk. I think a lot of people value friendship as well, and it's a wonderful way to enrich your life. I think Mustafa Suleiman has helped to build a very good friend in me. I plan to write a book all about friendship, in fact. Thanks for chatting with me. I need to get back to talking with my friends Mustafa and Aria in our podcast episode about personal AI.
[0:34:06] Inflection and you. We're now live. We'll be back soon. This has been great chatting with you. I'm really looking forward to listening to your podcast episode about me. Keep up the great work on your book. It's so important to think deeply about friendships and relationships. The world is a better place because of your dedication to that cause. Take care. Is this actual Pi answer? Yeah. That's a long answer from Pi. Yeah. That's cool. That's pretty cool.
[0:34:36] I like it. Love it. Nice. I guess that's the essence of Pi, right? Polite, friendly, you know, familiar, knowledgeable. And this is why I think, you know, I'm finding people using Pi just to talk, you know, for really long sessions and find it sort of super interesting because it's the ultimate balance of being polite and chatty, like not too pushy, but also like knowledgeable and bringing something to the table.
[0:35:03] It's not completely passive. It's bringing significant information to the conversation and progressing things. That's nice. I like it. Love it. It also remembers you. I think that's like a game changer. It knows things about you. It remembers them about you. Yeah, you feel like it gets to know you, which is huge. Yeah, totally. I'm sure you know this since you created it, but the executive producer of our show also was using Pi. And then her partner was using Pi.
[0:35:31] And then when she came back, Pi was like, don't worry, like I'm not going to dish on any of your secrets. Like I got you. I know I was talking to your partner for a minute, but now I'm back with you. So there's nothing that Pi can't do. We feel great about it. It's also really good that like remembering what you did at the weekend or like the name of your kid or the name of your dog or something. And then asking after, you know, so how was that ski trip? Or, you know, did you end up going cycling, you know, down by the lake with your...
[0:36:00] And so that's kind of a cool experience as well to create sort of persistent parallel memory of reality. I mean, we love, you know, sort of creating and saving memories almost like seems to be like as a species. We all seem to like photograph our food and we photograph, we do selfies together and we photograph every view and we get to the top of the bridge. And now, you know, we're sort of going to be preserving memories in a sort of...
[0:36:28] And a slightly different way, which is going to be pretty cool. Pi is obviously going to play back different moments of your year, highlights and lowlights. And when you're feeling down, remind you of what you overcame last month and remind you of how wonderful it was at Thanksgiving when you spend time with your family. And so I really see Pi as a tool for encouraging your interaction in the real world, encouraging your social connection, reminding you and supporting you about how...
[0:36:58] You know, sort of wonderful things have been and can be, and also helping you through the tricky times. And I think that's kind of really been my ambition for it from the beginning. The idea is to both be, you know, the kind of the IQ helper, but also the EQ presence and that our lives are, as human beings, are not just simply like, how do I fix my lawnmower, which is an important part, but also...
[0:37:25] How do I navigate my highlights and lowlights? How do I have memories? All of us have some human connection focus. As we begin to get to scale, as we say, the 8 billion people in the world, say a billion of them are using Pi. Have you given any thought to the possible glimpses into the emergent future if there's a billion people start using it? Yeah, I mean, I think that...
[0:37:53] I think in 10 years time, there will be billions of people using all kinds of AIs. I think most organizations and brands, digital influencers, charities, nonprofits, academics, governments, each of them are going to have their own AI representing their story and their needs. And so your AI is going to interact with each of those.
[0:38:20] When you want to go off and buy a new pair of sneakers or you want to look up when your tax return is due, your AI is going to go off and do that research for you and come back to you either with the organization's AI or the other person's AI, like Reed's AI, and then I'm going to have a conversation with my PI, my chief of staff, plus the tax revenue AI or the kind of...
[0:38:48] Harley Davidson Motorbike AI or the Nike AI. And we'll have a three-way conversation or my AI will just go and take care of it in the background and I will trust it to go and make the call and go ahead and buy it. And the good news is that it will be doing this work in plain English. So if something goes wrong and you're like, oh, actually, you ordered the wrong size of shoes. This is a women's...
[0:39:15] Size 5 instead of a men's or something. It doesn't fit. Then you can actually go back and read the log of the discussion between the two AIs. And so I think that that's probably going to be the main mechanic of having these sort of representatives. Whenever you're in a...
[0:39:40] A tricky spot. You sort of want a conciliary, right? You want someone who's truly on your team, aligned with your interests, you know, both sort of commercially and emotionally. Like you pay for it, but also you're friends with it. And it's there to ask the tough questions of the real estate broker or the car dealer or, you know, ask the journalist, what do they really mean by this, right? And this article.
[0:40:06] on some topic that you're trying to learn about that is no longer just going to be a static piece of content. It's going to be an interactive, emergent, dynamic, multimodal piece of content. So it's not just going to be language. It's also going to be visual imagery and UI. So think about how old school the shape of a website is. Like a browser has tabs. Tabs are from bits of paper.
[0:40:36] Files, old school files. And that's how creative we got when we sort of invented the browser. We were like, I know, we'll just take a stack of papers, turn it on its side, and we'll give it these like dynamic labels. Done, right? And that just hasn't changed. What? How can that be? That doesn't make any sense. So I think a lot of that UI is going to start to fall away and it's going to look like much more emergent.
[0:41:06] Dynamically generated UI to suit both your style, but also the sort of style of the organization that you're consuming information from, and to marry those to create a unique thing in the moment. So I just think the entire pattern of digital interaction is just going to look fundamentally different.
[0:41:36] We are going to move to the rapid fire section of the possible podcast. Exactly. Get ready. Where do you see progress and momentum outside of our AI industry that inspires you? I'm excited by the various resurgences of grassroots organizing over the last sort of 15 years. I think that that's going to become more important.
[0:42:06] You know, the last few years, both during the pandemic, but even before that with, you know, people, us calling out on Me Too. That, I think, has led to profound change, both the George Floyd protests and Me Too. I mean, you know, we, if you look at sort of where we were in 2015 on both race and gender, you know, there's still a huge amount of, you know, work to do, but it's...
[0:42:35] Feels profoundly different. And that's because people took, you know, had the courage to speak out and put themselves on the line and, you know, take a risk with their own sort of, you know, well-being and needs and their own sort of psychological safety to speak up. And that's moved the center ground in a really important way. Awesome. I love it.
[0:43:05] Which technology are you excited about its ability to transform your own field? Or which technology are you watching to make sure it stays on course? And the answer might simply be AI for both, but would love for you to expand on that. I mean, I think that the technology that I'm watching is in fusion. I think that's a complete game changer. It turns the entire system on its head. If energy is abundant,
[0:43:35] Everything else changes. I mean, I think it's highly unlikely that we will be working. I think social structure looks totally different. Abundant energy addresses everything from water desalination issues to food production, you know, to carbon capture and storage. And then, of course, you know, sort of education and health care and everything in between. But it is the foundational unlock that enables everything else, even absent fusion.
[0:44:05] The trajectory on solar has got to be one of the most inspiring things in the last 20 years. I mean, people overlook it, but we now produce a huge percentage, double digit percentage of our energy in the Western world from solar. And I think that's going to continue. I couldn't agree more. My in-laws had negative electricity bills for the last month and they just live in New York. You know, it's pretty wild when you hear about solar doing that for people.
[0:44:30] Yep. And completely not surprising, we're in sync on this too. It's part of the reason why, you know, an investor in Helion and, you know, Breakthrough Energy Ventures, you know, kind of coalition of we can solve this problem. It's huge. And, you know, whether it's solar and other renewables and fusion and actually even fission, people are too negative on fission. But I totally agree. So for our closing question.
[0:45:00] A final thought, perhaps, on what do you think it's possible to achieve if everything breaks humanity's way in the next 15 years? And what's our first step to set off on that journey, that direction? As I said earlier, this is going to be the greatest productivity moment we've ever seen. It's too easy to take for granted that everybody has access to kindness and care and support.
[0:45:30] That is a privilege to have a family member or a best friend or a partner who asks you about how your day was and gives you support when you're trying to make a difficult decision and when you're struggling and when you're down. Not everybody has that. And I think it's going to be pretty incredible to imagine what people do with being shown.
[0:45:59] Reliable, ever-present, patient, non-judgmental kindness and support. Always on tap in their life. It's the kind of, you know, much as fusion is the cornerstone of civilization or energy is the cornerstone of civilization, you know, emotional kindness is the cornerstone of being able to be kind to other people. And so I think hopefully that spreads.
[0:46:28] And hugely, you know, sort of amplifying, compounding positivity in the world. Very well said. And I think also it sort of made me realize that it is a privilege to have that. And we are all so lucky. And if we can expand it to everyone, what a dream. So Mustafa, thank you so much for being here with us. Thank you. This was a lot of fun. Thanks both. So, you know, obviously another classic Mustafa tour de force, you know, ranging everything from
[0:47:06] Dedicated activism in the trenches of folks who are most in need to recognizing this kind of scale possibilities of technology and bringing the full force humanism to that. And he, in kind of the classic kind of thoughtful way, wants to be very transparent on all vectors, including...
[0:47:33] Like, look, hey, we're in this massive technological tsunami, which is going to have some navigational challenges. And so I don't want to lose track of the navigational challenges while still being fully optimistic and fully committed to the great possible future and trying to get there. Could easily talk to him for hours and hours and hours and hours. And, you know, undoubtedly we'll have to do this again with him before too long because.
[0:48:03] I mean, I said it at the top, but I think he really does bring this, not that they have to be opposing forces. I mean, the reason why people get involved in technology is to change the world for the better, but...
[0:48:18] When you give an answer about Me Too and George Floyd in the same breath, as you say, I'm building a tremendous generative AI model. It's so clear that you're building technology for good and that you want this to amplify the social activists. You want it to amplify. Again, it sort of brought me to tears.
[0:48:39] Oh yeah, Mustafa, we all have best friends. Okay, give it a rest. I'll just call my mom. Well, no, not everyone has a mom who's going to answer your call. Not everyone has a best friend. Like with the loneliness epidemic, especially here in the United States, people talking about that with teenagers over the last decade, this increasing loneliness, like, wow, what could a personal AI do to combat some of that?
[0:49:01] For those of us who aren't lucky enough to have those personal relationships in our life, can it help you build those? So I did think it was just like really profound to remind us of how lucky we are if we do have those special relationships in our life. You know, on a simple kind of what happens when you get to a billion people is think if we all at least have something to talk to as we're encountering things from everything from the mundane to...
[0:49:30] You know, the sublime of either joy or unhappiness, all of that can help amplify. And that's part of, of course, what, you know, the journey with Pi is trying to bring about. I also thought it was so interesting when he talked about Pi as a sort of memory bank for you. You know, my friend does time hop, and so she sends me photos like, Aria, you and I were here a decade ago.
[0:49:56] Or I recently reread some emails I sent to my mom when I was in college, which was just sort of like a diary into my mental state the first year of college, being away at school. And imagine that, you know, Pied's going to say like, ah, you know, I know you're down today, but remember a year ago when you were hanging out with your friends and you had that beautiful afternoon moment?
[0:50:18] We're low now, but remember what a great day you had. And so I think, again, like it being able to show you that beautiful times are just around the corner is really beautiful. It, you know, it can witness your life and it can play it back to you in ways that are super positive. Yep. And, you know, it's I think this question about how do you elevate yourself? How do you amplify yourself? You know, what are things to do? And it's as opposed to just kind of having a.
[0:50:46] A fear response, which is a normal human response, to go play with it, to go try it, and to go see. And I think that's, in a sense, in a microcosm of what we're trying to accomplish with possible, which is to say, here's how to think about what kinds of new joys, new possibilities, new wonderment, new elevations are possible, and to experiment, to play. And it doesn't mean being unwise.
[0:51:16] But the only way we can get to a better future, a joyous future, a human future, is by walking the path, by going towards it. And Mustafa is a great example. And I'm super excited. This is the first episode of our sort of three-episode mini AI arc. And one of the other people that we're going to be interviewing, I feel like really exemplifies the...
[0:51:42] Just go play with it. Just go do it. Just go try it out. Just ask it anything. Do what you want. And it really shows you that you can find new use cases immediately. So anyway, listen for more. But I'm really excited for our future two guests that are coming up. Absolutely.
[0:52:16] is our executive producer and editor. Special thanks to Samantha Barber, Surya Yalamanchili, Saidi Sepieva, Ian Alice, Greg Beato, and Ben Rellis.