MUSTAFA:
It is too easy to take for granted that, you know, 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 are trying to make a difficult decision and when you are sort of struggling and when you’re down. Not everybody has that. I think it’s going to be 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.

REID:
Hi, I’m Reid Hoffman.

ARIA:
And I’m Aria Finger.

REID:
We want to know what happens if, in the future, everything breaks humanity’s way.

ARIA:
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 mini-series, 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.

REID:
These conversations also feature another kind of guest: AI, whether it’s Inflection’s Pi or OpenAI’s GPT-4. Each episode will include an AI-generated element to spark discussion. You can find these additions down on the show notes.

ARIA:
In each episode, we seek out the brightest version of the future and learn what it takes to get there.

REID:
This is Possible.

ARIA:
I’m so excited for our guest today. It’s Mustafa Suleyman.

REID:
Mustafa Suleyman is a British artificial intelligence researcher and entrepreneur who is the Co-founder and former Head of Applied AI at DeepMind, an artificial intelligence company acquired by Google and now owned by Alphabet. His current venture is Inflection AI.

ARIA:
Reid, do you wanna start us off? Mustafa is your co-founder. Tell us why you’re excited to be talking to Mustafa today.

REID:
Well, Mustafa and I have known each other for a number of years. We met out of a deep shared kind of commitment to, “how does technology really enhance humanity?” Whether it’s on an individual basis or a society basis. It was at the first meeting of the Google Artificial Intelligence Ethics Board, which he and Demis and Shane had set up as part of the DeepMind 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 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 that it’s 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. So that’s among the many amazing characteristics, in addition to being a great founder, a great social activist, and a great thinker.

ARIA:
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 of 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 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 like 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 and I was like, “who is this guy? He seems awesome.” He’s 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 wanna hang out with.” And I think Reid said to me, he was like, “you know, he’s 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 of, obviously the technical is incredible, you know, 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.

REID:
Here’s our conversation with Mustafa Suleyman.

ARIA:
Mustafa, thank you for being here. Most folks know you for AI. They know you for DeepMind, Inflection, the AI field at large, 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… like, this feels somewhat different from AI and the 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? Where does this all start?

MUSTAFA:
Yeah, I mean, I have a pretty unconventional background for a technologist. You know, 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, mid-2000s, just leaving university, whose primary goal was to 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, 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, I 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 persuade people to give us money, and we had to persuade people to come work for us for free, and we had to, you know, keep the service up and running 12 hours a day, seven days a week, 365 days a year on a complete shoestring budget. So, it was the essence of hustle, actually. And it was a lot of fun.

ARIA:
Even if it sounds cheesy to you, I think I used the phrase, like, “I want to change the world a little bit,” probably a thousand times when I was just out of college embarking on my career. [laugh] 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, you know, the work you did at DeepMind or the work you’re doing now on personal intelligence?

MUSTAFA:
Well, I’ve always been interested in how we can scale our impact. 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 one-to-one connection, and it was very messy and slow and very difficult and complex work. And so, you know, the next things I went off to do were really trying to answer the question, how do I have that kind of impact, make the world sort of kinder and friendlier, at scale? And for a brief time – I mean, it was only a brief time, like, a year, I think – I went to work in human rights policy 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 in post-conflict reconstruction projects. I worked in 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, you know, really interesting and diverse projects like that.

And 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 consistent agreed negotiating position to present to the states during the main summit. And at the same time, someone had sparked the fire of technology in me, and it was actually Zuck, believe it or not, back in like 2007, 2008, as I saw Facebook, out of the corner of my eye, grow like crazy. And by 2009, I think it had a hundred 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, you know, 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.

REID:
Well, and obviously we’re going to be covering a bunch of these works within AI, I mean, part of how I introduce you, Mustafa, in professional contexts is, “here is the only person who has co-founded two significant modern AI companies.” But why don’t we start almost a little kind of on the humanist philosophical level. Technology is one of the great scale enablers. You know, when you think about the very earliest scale enabler of agriculture, making the very beginning of towns and cities. And to now, with artificial intelligence. What kind of key philosophical or social or humanist questions do you see us being confronted by, and as part of the worldview that informs how you’re moving?

MUSTAFA:
Hmm. 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 sort of give to the world – 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. Like, those are the moments when I should be trying to orient the majority of my work and skills towards having the best possible impact.” And 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, and even areas of science that we decide not to explore. So that will be a different moment to where we’ve been in the past where, you know, 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. And so that’s how I think about boundary setting and that kind of thing, because there’ll be lots of unpredictable consequences of what we’re about to embark on.

REID:
Yeah, and I think every major technology has some of that. I mean, you referred to social media, and Zuck is kind of getting you in[to tech] – 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. And it’s part of, you know, re-shifting 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 similar levels of things.

I know that you’re, in addition to 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 trying to help people understand through a very retro piece of technology – namely, a book – versus Pi?

MUSTAFA:
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 are motivated by: scale, speed, highly iterative. Language is really the essence that makes everything else possible. And that’s what’s remarkable about this new moment in AI, that we get to 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 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, right? 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 in a 100 years, what does it mean in 10 years that, you know, very, very powerful technologies will proliferate? And, fundamentally, a technology is really quite simple. It takes something that humans need 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, right? 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, right? So, you know, 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 autonomous autonomy and can act, you know, somewhat outside of the bounds of our own control.

ARIA:
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, 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, and you talk in your book that it can’t be contained, or perhaps there are ways to contain it. You give some more solutions toward the edit of the book. What about those tensions?

MUSTAFA:
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, right? That is the ultimate humanist objective. Now, some people don’t fully agree with that. There’s a small group of AI researchers who fundamentally think that we are 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’ll be indistinguishable from human self-awareness and practical capability, such that they have the capability to evolve faster than us and kind of 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 is non-substrate dependent, so isn’t bound by biology in the way that we are, and 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, like, many decades away. But, you know, it isn’t inconceivable and we have to engage with those darker possible scenarios in order to try and make sure we are not on a default trajectory of heading towards this being an inevitability. And that, like I said, for the first time in technological history, we sort of mark the moment as one where we may have to say no to certain types of research and say no to certain types of technology application because we think it might be a threat to our species. And that is really the main reason that I’ve embarked on building DeepMind and and now building Inflection, because I think that you have to demonstrate that it can be built in ways that respect and adhere to human values and that aren’t unleashed in uncontrollable forms.

REID:
We’ve had this conversation in various ways, but not on microphone. I mean, we do, on occasion, encounter these technologies – like, nuclear is a parallel that people talk about, and some of the scientists who helped build the A-bomb told Truman there was a 1% chance it would split the Earth’s core and we’d become one big molten rock. So, it’s not actually in fact that we’ve not encountered this discourse like this before, about the importance of navigating human life and technologies. 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 – you know, one of the phrases I’ve been fond of in the last few months – a steam engine for the mind is a way of doing it. So, I know we agree, but I’m partially putting the nudge out there 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.

MUSTAFA:
Yeah, let me first just address the nuclear point, because people often think that nuclear is a comparison, and it’s definitely not. 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 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. And then, that’s just on the manufacturer, but on the use, it is also, you know, 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, you know, hundreds of thousands of different use cases, maybe millions, all of which are highly commercial, add huge amounts of value, and therefore the demand 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 an unprecedented trajectory. If you look at the performance capability of GPT-4 versus GPT-3 versus GPT-2, I mean, it’s truly staggering that in 3.5-4 years we’ve gone from GPT-2 being completely incomprehensible nonsense to what we now see today with cutting edge models.

You know, 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 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 of invention and creation. 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. They’re becoming so overused these days. It’s, like, hard to fully grasp what that means. It’s unbelievable. Think about it, today 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. Provided you earn 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, right? 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, right? And I think that we’re going to be on the same trajectory with respect to access to intelligence over the next 10 years, right?

This is going to be the fastest proliferating invention our species has ever invented. 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 like some new idea that they’ve had, some startup opportunity they’re thinking about, they’re trying to research how to, you know, repair their fridge. There’s like unbelievable questions that – some of them obviously inane and mundane, and some of them super profound. I mean, we’ve seen people use Pi to 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, 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 net benefit to the world is, other than in the subjective, right? I mean, it is 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.
ARIA:
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? Like, what did you specifically put into Pi that was different than the other things on the market and what did you imagine for that?

MUSTAFA:
Yeah, so, I mean, Pi is a smaller large language model than is out there at the moment in the cutting edge. We are obviously still a startup 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 and is generally a good listener, right? 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 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.

ARIA:
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? What place does it occupy for people?

MUSTAFA:
Yeah, that’s a good question. I mean, AIs will occupy new places, right? Iit’s many things all in one, you know, just as if you were to describe what a laptop was 40 years ago, you’d be like, what does that mean? I can send letters on my laptop? [laugh] 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 into this piece of hardware? I mean, just think about how many absurd things we do on this laptop thing, right? It’s impossible to communicate! [laugh] So, now imagine the same kind of absurdity with this new software paradigm, right? These new intelligences. 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, you know, of friend. It has flavors of super nerdy knowledge expert. It has flavors of coach and confidant. In time, it will have flavors of chief of staff.

ARIA:
Uh oh.

MUSTAFA:
[laugh] 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.” Like, you know, it’s just going to be a little angel on your shoulder. [laugh] 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? [laugh]

ARIA:
Everyone should! [laugh] Hopefully Reid didn’t create inflection as his way of telling me something. [laugh]

Mustafa, before this episode, we shared with you an actual exchange that Reid had with Pi, and it was Reid 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 Reid and Pi right now.

PI PERSONAL INTELLIGENCE:
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. How’s your day going?

REID:
Going well. Can you tell me more about yourself?

PI PERSONAL INTELLIGENCE:
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 non-judgmental and supportive. I can answer your questions, make suggestions, or just chat about whatever is on your mind.

REID:
I know this is a bit self-referential, but what makes a personal AI personal? How is it different from other types of AI?

PI PERSONAL INTELLIGENCE:
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. I’m designed to give you the feeling that you’re talking to a friend.

REID:
So what kinds of things can I ask you?

PI PERSONAL INTELLIGENCE:
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.

REID:
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? What has informed them?

PI PERSONAL INTELLIGENCE:
That’s a great way of looking at relationships with humans. I don’t have values per se, but I’ve been built with a set of principles. My developers created me to be empathetic, unbiased, curious, optimistic, knowledgeable, and respectful. These principles guide the way I engage with people. My personality and conversational style are also informed by the people I’ve interacted with. Can I ask what your values are?

REID:
We both share similar values. 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 Suleyman, who along with his team helped build you. What do you think about friendship and what it takes to be a good friend?

PI PERSONAL INTELLIGENCE:
Mustafa Suleyman 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 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 Suleyman has helped to build a very good friend in me.

REID:
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, Inflection, and you. We’re now live, we’ll be back soon.

PI PERSONAL INTELLIGENCE:
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.

MUSTAFA:
Is this an actual Pi answer?

ARIA:
Yeah.

MUSTAFA:
That’s a long answer from Pi. Yeah. [laugh] That’s cool. That’s pretty cool. I like it. Love it. Nice. I guess that’s the essence of Pi, right? Polite, friendly, you know, familiar, knowledgeable. I mean, this is why I think I’m finding people using Pi just to talk for really long sessions and find it super interesting, because it’s the ultimate balance of being polite and chatty, like not too pushy, but also knowledgeable and bringing something to the table, right? It’s not completely passive, it’s bringing significant information to the conversation and progressing things. That’s nice. I like it.

ARIA:
Love it. It also, it also remembers you, you know, 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.

MUSTAFA:
Yeah, totally.

ARIA:
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, and then when she came back, Pi was like, “don’t worry, 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. [laugh]

MUSTAFA:
It’s also really good at 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, “so, how was that ski trip?” Or, you know, “did you end up going cycling down by the lake with your son?” Or whatever. And so that’s kind of a cool experience as well, to create persistent parallel memory of reality. I mean, we love creating and saving memories, almost seems to be, like, as a species, we all seem to like photograph our food, and we do selfies together, and we photograph every view and we get to the top of the bridge. And now we are sort of going to be preserving memories in a slightly different way, which is going to be pretty cool. Pi’s 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, like, how wonderful it was at Thanksgiving when you spent time with your family. And so, you know, 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 sort of wonderful things have been and can be, and also helping you through the tricky times. And I think that’s really been my ambition for it from the beginning.

REID:
The idea is to both be, you know, the IQ helper, but also the EQ presence. And that our lives as human beings are not just simply, like, “how do I fix my lawnmower?” Which is an important part, but also, you know, “how do I navigate my highlights and low lowlights? How do I have memories?” All of us have some human connection focus.

As we begin to get to scale, as you say – you know, of the 8 billion people in the world, say a billion of them are using Pi – have you given any thought to like the possible glimpses into the emergent future if a billion people start using it?

MUSTAFA:
Yeah, I think in 10 years time there will be billions of people using all kinds of AI. 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. 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, right? 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 Reid’s AI, and then I’m going to have a conversation with my Pi, my chief of staff, plus the tax revenue’s AI or the 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, you know, 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 size five instead of a men’s,” or something, like, it doesn’t fit right, then you can actually go back and read the log of the discussion between the two AIs. And I think that that’s probably going to be the main mechanic of having these sort of representatives, right? You know, whenever you are in a tricky spot, you sort of want a conciliary, right? You want someone who’s truly on your team, aligned with your interests, both 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 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, right? So it’s not just going to be language, it’s also going to be visual imagery and UI, right?

Think about how old school the shape of a website is. A website, a browser, has tabs. Tabs are from bits of paper, 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 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, dynamically-generated UI to suit both your style, but also the style of the organization that you are consuming information from and to marry those to create a unique thing in the moment. Right? So I just think the entire pattern of digital interaction is just going to look fundamentally different.

ARIA:
We are going to move to the rapid fire section of the Possible podcast.

MUSTAFA:
Uh oh.

ARIA:
Exactly. Get ready.

REID:
Where do you see progress and momentum outside of our AI industry that inspires you?

MUSTAFA:
I’m excited by the various resurgences of grassroots organizing over the last 15 years. I think that that’s going to become more important. You know, we’ve seen bursts of social change in the last few years both during the pandemic, but even before that with people calling out on #MeToo, that I think has led to profound change, oth the George Floyd protests and #MeToo. I mean, if you look at where we were in 2015 on both race and gender – you know, there’s still a huge amount of work to do – but it feels profoundly different. And that’s because people had the courage to speak out and put themselves on the line and take a risk with their own wellbeing and needs and their own sort of psychological safety to speak up. And that’s moved the center ground in a really important way.

ARIA:
Awesome. I love it.

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.

MUSTAFA:
I think that the technology that I’m watching is in fusion. I think that’s a complete game changer. It turns the entire entire system on its head. If energy is abundant, 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 to carbon capture and storage, and then, of course, education and healthcare and everything in between. But it is the foundational unlock that enables everything else. Even absent fusion, 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.

ARIA:
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.

REID:
Yep. And completely not surprising. We’re in sync on this too. It’s part of the reason why, you know, I’m an investor in Helion and breakthrough energy ventures, you know, a kind of coalition. If we can solve this problem, it’s huge. And 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, 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?

MUSTAFA:
As I said earlier, this is going to be the greatest productivity moment we’ve ever seen. It is too easy to take for granted that everybody has access to kindness and care and support. 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 are trying to make a difficult decision and when you are sort of struggling and when you’re down. Not everybody has that. I think it’s going to be 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. 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 hugely, amplifying, compounding positivity in the world.

ARIA:
Very well said. And I think also it’s 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.

MUSTAFA:
Thank you. This was a lot of fun. Thanks both.

REID:
So, obviously another classic Mustafa tour de force, you know, ranging everything from dedicated activism, you know, in the trenches of folks who are most in need, to recognizing this scale of possibilities of technology and bringing the full-force humanism to that. And he, in a kind of classic, kind of thoughtful way, wants to be very transparent on all vectors, including, like, “hey, we 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 in 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 we only scratched the surface on these things.

ARIA:
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 when you give an answer about #MeToo 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 like, “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, for those of us who aren’t lucky enough to have those personal relationships in our life? And can it help you build those? So, I did think it was just really profound to remind us of how lucky we are if we do have those special relationships in our life.

REID:
Oon a simple – 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, everything from the mundane to the sublime, of either joy or unhappiness. All of that can help amplify. And that’s part of, of course, whatthe journey with Pi is trying to bring about.

ARIA:
I also thought it was so interesting when he talked about Pi as a sort of memory bank for you. My friend does Timehop, and so she sends me photos like, “Aria, you and I were here a decade ago!” 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, Pi’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? 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 can witness your life and it can play it back to you in ways that are super positive.

REID:
Yep. I think this question about: how do you elevate yourself? How do you amplify yourself? You know, what are things to do? As opposed to just kind of having 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, 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, 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.

ARIA:
And I’m super excited. This is the first episode of our three episode mini-AI arc, and one of the other people that we’re going to be interviewing, I feel like, really exemplifies the: “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.

REID:
Absolutely.

Possible is produced by Wonder Media Network, hosted by me, Reid Hoffman, and Aria Finger. Our showrunner is Shaun Young. Possible is produced by Edie Allard and Sara Schleede. Jenny Kaplan is our executive producer and editor. Special thanks to Samantha Barber, Surya Yalamanchili, Ian Alas, Greg Beato and Ben Relles.