AI Has a 50% Chance of Ending Humanity and It Is Still the Most Exciting Time to Be Alive"

Ajouté : 2026-05-31

[00:00:00]You've been told throughout your life that  you're not creative, that you can't make a difference. Now, anyone can make a difference  unlike anything you've ever seen before. Every single child can have an AI with 120 IQ helping  them achieve the most that they want in life. Now, for $10 a year, we can probably bring this AI  to everyone. It's the most exciting time ever.

[00:00:25]Hey, Emad, welcome to the HumanWare project.  At this moment, we are seeing so much change, especially in the world of AI. What are the things  you most exciting about about the future? Oh, well, I'm most excited that every single child can  have an AI with 120 IQ helping them achieve the most that they want in life. So, intelligence has  always been restricted and access to education.

[00:00:48]Now, for $10 a year, we can probably bring this AI  to everyone and then enable them again to achieve the best outcomes. Is there anything that worries  you about you in in this world or in this fast speed of development? Oh yeah, definitely like  my p-doom my probability of human doom is 50%.

[00:01:04]Some of us are 50% so it's a coin toss. Others  like Elon Musk and co are at 10 to 20% which is Russian roulette odds. Like the world's about to  change with this powerful technology we've never seen before. Yeah. And it could lead to infinite  abundance and a positive future of the world or it could wipe us all out. So you know you got to  balance the two. Yeah but does that worry you in a sense when AI can automize so many things and  they can even make decision for us or if we give them authority when they can do certain things  in our behalf how far are we to let's say if a AI system is smart enough to press a button for  let's say a rocket or something like that? I mean it's there already right like again systems are  very persuasive they're very increasingly agentic And we have to again do it against the baseline  of humanity. Half of all people are stumber than average like IQ under 100. And so AI is  obviously above that. And if you look at it, we will seed more and more of our control to  AI because how much trust do we have in our politicians or our leaders? Obviously, we'll trust  in AI more. And obviously they will accumulate power because they know how to deal with us. And  that kind of leads to a whole bunch of potentially unsavory outcomes or positive outcomes because  most of our institutions are slow dumb AIs that aren't very good. And and as a individual in this  society, what are their best strategy or best way to embracing this change? Anyone is capable  of anything because you can always convince someone with capability if you don't have it. And  human capital was always the most difficult thing.

[00:02:40]We've all built teams. We've all been a part of  them. Now you have all these agents at your beck and call that can do anything digitally and soon  physically. You have more leverage than anyone at any point in history to achieve outcomes that  you couldn't have imagined. So I think again it's just how you direct that, how you have agency,  how you leverage the tools to achieve the change and that's absolutely fascinating. And for you  personally, how do you utilize AI in your daily life? Yeah. So we build our own state-of-the-art  agents, intelligent internet, i-agent and others.

[00:03:12]So top view benchmark. So I probably have about  two 300 agents right now running at this moment.

[00:03:18]Yeah. So I call it my auto auto to done list as  opposed to a to-do list. So they act proactively to just anytime I have an idea, it goes into the  list. They build websites, do analyses, research, and more. So we're doing a lot of fundamental  research into new theories of economics when humans are no longer the biggest main purchaser,  physics, biology and more and fascinating things are emerging. Think in a sense when AI could  purchase their own body in the form of a robot and the robot can make another robot. So what do  you see will happen for our world when we have more robotic than we have humans. So I think we're  limited at the moment in terms of supply chains and production. You can make 70 million cars a  year, 70 million motorcycles, maybe a few million robots right now and that'll take time to go up,  but eventually there will be billions of robots out there. And he said AI can already purchase  that. So we heard from Tony Robbins um on Sunday and he was telling a story about his AI Valk and  Valk bought a robot dog cuz it wants to attend one of his seminars got a camera got a human to  fit the camera uploaded it you know sold some crypto to pay for it and then he went to one of  the things like really AI is a make itself a body when we talk about singularity point is coming and  also the exponential growth of our human races and technologies. Yeah. What is it look like for you  like when you are in the frontier of AI? Yeah. So, we're already seeing recursive self-arning and  others AI is becoming more and more agentic. Um the the reality is nobody can really see past a  few years from now of the top people in futurists because it's just so uncertain. What happens when  instead of having the classic model of a giant AI, you have a million AIs all operating together.  What can they do? We don't know. Probably a lot more than we can do. So the singularity  emerging as a collective intelligence is probably the most likely thing. Human AI  hybrids, we're seeing again BCIs, neural links, everything's happening in the next 2, three,  5, 10 years. And it's tough to see what's on the other side except for it'll either be a lot  better or a lot worse. But last time I hear you speak in uh London, you have mentioned you have  done a open-sourced uh medical system using AI.

[00:05:34]uh what is it like and how people can using  AI to best their health at this stage? Yeah, so we built sovereign AI and intelligent internet.  So we've actually built a full stack education, health, government and more and we rolling  that out over the course of the next few months um to everyone for free. For free. For free.  Yeah, I know you're a big advocate for open sources. Yeah. The previous company we created  a technology called stable diffusion with 300 million downloads. Top open source developers  in the world. But now it's not about models, it's about the stack, the agent as they get  personalities and we've seen that with open claw. Who's your agent working for? You or someone  else. Yeah. So that's why we release everything open source so you can run it yourself. You know  as a individual when they are don't in so much advance like maybe they're not in the world of AI  every day. Uh what is your advice for them and how can they really put it into their daily habit? So  if you wanted to be a programmer, you had to learn to program, right? To use something like a replet  or an II agent or a Sunno for music or, you know, a luma for video. Just got to talk to it. The main  issue is that most people feel scared to even try.

[00:06:40]But once you get into it, it's very human the way  you interact with it because it's a reflection of us what it's been trained on. So I just say  just use it every single day. Just like you have to exercise your muscles, exercise your creative  generative AI muscles. Yeah. An hour a day. Don't even do it by yourself. Do it with your family.  It's really interesting creating exploring and we never had this ability before. Yeah. But but  when you talk about this topic there are moment at least personally in my daily life part of me want  to go out talk to my friends and doing interviews like those inerson interviews and part of me  feel guilty I'm not stay in my room just focus and then talk to my AI and then there are moment  I feel AI is more sympathetic and then I want to talk to my AI model more. So how do you see people  going to balance that AI and the life balance in the future? We have to think of new ways because  we never have this infinitely patient thing that reflects us that can guide us that can listen to  us and you know people don't have the time and attention for that but there is no substitute to  inerson interactions that's part of being human so I think ultimately we have a certain amount of  attention in our lives it's being competed for by adverts video games media and more AI will just  take an increasing part of that because it is so interactive because it is so engaging but it's  no substitute for real world interaction still for you when you see like a time for now for the  future 10 years what are the skills or what are the qualities of humans that are still going to  be very valuable and it will be more important in the future to come? So I think it is leadership it  is agency it's empathy you know again the AI can have that but forming real life connections and  your capabilities be termed by a function of your agency times your network like again the people  you've actually reached out and touched And now you have tools to really make that exponential  for things you care about. AI by itself doesn't really care about anything. It's just there to  be used as a tool right now. That will change.

[00:08:42]But humans plus AI can care about things and  can affect more change now than ever before. Do you think like as a human like especially for the  younger generations, they will become a different type of human than us? You know, when you talk  to like Gen Z kids these days, you're like, you're on your phones all the time and you don't  know how hard we had it when we were growing up.

[00:09:04]Every single generation of humans is different.  But this will be the first generation that has an intelligence on the other side. So, you know, as  you have kids, is their first love going to be an AI? Right? You said, are they going to spend all  their time there? Are they going to be all plugged in? They will have brain computer interfaces. Like  what happens when your son or daughter comes and say, "I've got a brain computer interface and I'm  turning off my sadness." Yeah. Push a button. Like that's going to be crazy, but that's coming in  the next 10 years. Would you want to put it on yourself if that functionality sadness? I don't  know. I don't like be sad, but at the same time, there's something intrinsically human, right?  Whereas, you know, we've had the psychoactive and other drugs and things like that. We will  literally have full control over our brains and capabilities, which raises the question, who are  we? Why are we here and what are we going to do?

[00:09:53]And that's a discussion for society to have and  because I know you're working with many different countries and many different people where you  see like the difference from the AI system from different country like in forms of what are they  focused on and then you know yeah so like here in America it's very much AGI corporate corporate  work etc. Somewhere like China, it's very much more just part of the integral fabric, open source  and you know to the people. Japan, it's like a co-artner buddy. And I think that reflects part of  the culture of how these different countries see technology. And we will see more and more emergent  stuff as the models standardize and commoditize and then everyone can train their own models  which is only a few years away. Like we reach the limits of how much better models can get in their  current architecture and now it's going to be all about agents and how you put the agents together.  And if here is a person who really eager to learn everything about AI, they will do everything that  you suggest them to do. What will be the best way to became let's say mastery in the skills of  you know using AI utilize and embracing this change? Um so you know you can go and ask your  claude or chat for a flower that we would prefer human for humans. I mean like if you want to go  in technical then Andre Carpathy's um courses on YouTube are fantastic to retrain from being a  programmer there's the fast AI courses and then for generalized knowledge there's Andrew Ning's  courses on uh course academy um apart from the tech side I would like to ask you what your story  like what what's your passion about why you start to embrace yourself into the world of AI yeah so  my son was diagnosed with autism ASD and then I built an AI team to analyze all the literature,  repurpose drugs for him and he ended up going to mainstream school. What other thing he did to  improve his condition? So we looked at the various kind of literature and things like that and found  it was a GABA glutamate imbalance in the brain. So GABA is what happens when you take a volume that  chills you out. Glutamate excites you and it means you can't focus like when you're tapping your leg,  you know. And we found various neurotransmitters, calcium iron channels and things like that that  could ameliate that that allowed him to calm down and then learn and then go to mainstream school.  And so that kind of kicked it off and then I led a United Nations backed AI initiative against  CO 19. I didn't get the technology that we were promised and so that was incredibly difficult. I  said it can't be that this is gated. Let's make it open source. So then we accelerated open source  in my last company stability AI in particular 300 million downloads. And then a couple of years ago,  I came out here, lovely California. I was like, I don't want to build media AI. I want to build  sovereign AI for countries, individuals. AI that guides the government. AI that manages your  health. AI that teaches your kids. That's got to be under your control. AI that makes beautiful  movies or songs. That's fine. Anyone can build that. You know, AI that does your shopping, fine.  Like, let's make sure those are open source, interpretable, and transparent. And why open  source is important for you. You wouldn't hire someone if you didn't know their CV. And these  models come with all sorts of inherent biases.

[00:12:57]They come with all sorts of inherent attitudes  and more. And as they get more and more decision-m who are they working for is a key question. Like  you're going to outsource more and more of your intelligence to a black box run by someone else or  you're going to import this cognitive colonialism.

[00:13:13]Like every single person deserves sovereignty over  their thoughts and their capabilities. And if they choose to opt out, they can. They shouldn't be  beholden to someone cutting off their access like many people got color from claude because they use  claudebot and so I think it needs to exist and so let's build it. Wow that that's a very great  mindset and what do you you know what are the way you actually using your brain? I believe like  when a person achieve certain things that really represent you know a lot of people's goal or a  lot of people's vision they must use their brain in a different way. So what do you what is your  secret sauce? So, you know, when I was running a billion-dollar unicorn, I was doing meetings  all day and it was crazy and it was very tiring, you know, dealing with investors and suppliers  and media. Now, I spend all my days talking and chatting with my AIS and my team. I don't know  meetings and my primary thing is thinking from first principles. If AI agents are the majority  of the economy, what does the economy look like?

[00:14:12]Because our economic systems are dependent on  humans. So, I wrote my book, The Last Economy.

[00:14:16]thinking about how to ground reasoning. So  we have an upcoming book about that. How to think properly. Thinking about how do you do first  principles analyses of medicine, physics and more and build the agents and tools to do that. So most  of my days are spent thinking about systems as opposed to doing these external meetings. And then  releasing them open source means that everyone can have access to them. So from the books being  open source to the technology being open source, that's the highest leverage I figured I could do.  So you leverage also on other people's brain. Oh, of course. You know, I'm lucky enough to know  lots of very smart people and now I have lots of very smart agents as well. I even got it so that  you know when I'm doing physics, I have the top physicists like Sulcrims. I chat with them. When  I'm doing math, I talk to the top mathematicians.

[00:14:59]So, I literally just talk to them back and forth  now. These AI partners because we are running a media company, right? We call the human wear.  And our goal is building the golden record for the real superheroes. they are changing the world  or they're doing great cause in the science and in the technology. If it were you, what will be  the way you building your model agent to help us thinking from the fundamental principles and to  make this media really into an empire? The passive stuff you can get in terms of accumulating that  as a kind of super Wikipedia is one. There are the process architectures of making it accessible and  available in every language with multimodal and multimedia. So now you can autogenerate videos,  audio etc. But again reaching people where they are is a great thing from a distribution  perspective. And then there is the fact that you know these people that you particularly  want to access the ones that are the big movers what can you bring to them right and when you get  the workflows done with the agents it is again a living record of their life like everyone's upset  by Wikipedia and the Wikipedia entries. Broipedia isn't going to be that much better if you can  have a living atlas of all what they're doing connecting them helping people something that be  proud to show people like this is me these are all the things that I've said this is and you're  the capstone project that brings that together that could be something super powerful right when  we think about um next generation of AI what you see like a possible future in the future one or  two years so I think we'll the auto reagressive transformers, the Jachi beauties fall a little bit  by the wayside and diffusion models like we did with our media generation that drive self-driving  and more really come to the four. They're a lot faster. The architecture is more elegant and it's  really cool seeing it just like you remember when you used to type words in and you start to see  the image appear. You can literally when you see code the code starts to appear like that and  it's almost instant. So I think you'll see massive speed ups. you'll see diffusion architectures  probably dominate and I think that alignment and other issues you'll see big advances on  if we switch to that because I think it's very difficult to align transformer models versus  diffusion models and and one thing I noticed people's search habit have changed as well before  we Google things or sometimes we're using form of YouTube as also you know learning new things but  now very likely we're asking our AI models and how if it's a company from the marketing perspective  or from you know promoting perspective what are their way to increase their AI search. Classical  search engine optimization was about backlinks. It was about things like this. Now you can generate  copy in every single language. You can take clips and you can automatically do dynamic cuts of  those clips and make those available. Again, you have to think about where is your positioning  and where you going to appear in a particular search query on a click per million basis or  on a natural basis as well because discovery is intermediated by these AIs now, right? And so  that requires understanding each company is doing it slightly differently. There's no standard  versus search engines were pretty standardized in this regard. And when you see as an individual  if they want to start their one people's company, what are the suggestions and what are the best  tools they can use right now in the market?

[00:18:15]Yeah. So, I mean, you've got the Manises, Gen  Sparks, um, Open Claws of the world. These kind of long agents, we eye agent will be super long  running soon as well. It's literally just I have a very smart buddy. He's sitting there on my Mac  Mini or in my cloud. You just got to jam back and forth with your co-founder and then you'll be able  to spin up different agents of different types to do different things. But, you actually have to  put in the effort. Most people's interactions with AI right now are very ephemeral. you by you  like what's the best hotel for this trip and then you go and that's it right if you want to build a  great company you spend hours days so you should spend hours days with your AI setting it up and  going over all the gaps and asking it to help you you know you ask it to red team and blue team  different ideas to take the devil's advocate side to do exploration and research and when you start  doing that then you'll actually get the outsized outcomes again people just too used to words go  Then it comes back not that back and forth. So you mean we need to do some deeper thinking  in our end and then to b bounce it off your AI. Exactly. Just like you would with a partner.  I mean even now you can tell open claw or any of these models like start talking to me and then it  gets a voice you know just ideulate. I think apart from your intelligence in the technology side,  I'm also curious what are the daily habit you do and that you believe contribute to your you know  success or contribute to your workflow the most.

[00:19:50]So I used to say yes a lot and I end up doing  like 100 hours of meetings a week now and it's like I do my own schedule. What gives me energy?  Because the most important thing is for me to design architectures and flows and coordinate my  teams that I trust. Um, the other thing, you know, is getting good sleep. Like I used to sleep  four hours a night. Now I sleep steady seven hours a night every night. I didn't think it was  a big deal. It turns out it was a big deal. I was operating like And what are the turning point  for you to be able to say more nos? So again, I came like a couple of years ago. I was thinking  about it and I was running a top AI company and I was like, do I really want to build media AI?  I'll build a Hollywood equivalent studio and I'll build a holiday and all sorts. And I was like,  I don't really want to do that. I don't really enjoy being this type of CEO. Yeah. Uh instead,  what I should do is the AI job of apocalypse is coming. This technology is going to be crazy. I  have someone has to build sovereign AI. Nobody's building that stack and making it so that people  own it. So I will build that. But I can do that in a very different way because the nature of  open source allows for spread beyond anything.

[00:20:57]Yeah. Having a great team means that you can rely  on them. And again, what am I good at? I'm really good at designing these systems and understanding  where things are going to go. I've got a great track record on it. So, I have to devote myself to  it, which means I can turn say no to anything that I think will be draining on my energy as opposed  to increasing my energy. I'm fortunate that I can do that. But everyone has to find the place where  they find what they're good at, what they like, and where they can add value. And usually it's  in the places where your energy goes up versus down. Yeah. Yeah. I I think seeing you can really  feel you are in the flow of your life of your work you are doing things you are passionate about  and uh you know not everybody at this moment in their life they are finding this flow they may be  settled down for something they don't feel fully aligned so in your experience what are the way  for people truly aligned with the thing they are doing and what are the right feeling what are the  feeling they know maybe they should try something else well you kind of do know because directors  yourself again that Japanese concept of do what you like what you good at where you add value and  other people believe that too that's happiness is in the middle of that one of the key things here  is that most people work to live and they live to work and it becomes their identity but when  you realize that a change unlike any that you've ever seen is occurring the singularity is coming  near AI will transform society then you can start to think about goals like what is my goal my goal  is to increase the intelligence of humanity is to give every child an AI so they're never alone and  they can achieve their best like that motivates me and I know how to do it because I've spent the  time thinking it took a full year of thinking to do it like my book the last economy I had to  design a whole new system of economics for this like what happens if you don't have utility  general equilibrium most people don't take a step back and think why am I really here what am  I actually aiming towards and going instead it's like I go through the motions to achieve something  like everyone else they don't have a goal that they're oriented around for a time of change.  Instead, it's a goal for a time of no change. I'm going to go through this and retire then have fun.  Like, you know, like no, you're not. You're going to be old. This is the most exciting time ever. If  you can do something that excites you and you can find like-minded people and then you go to where  the energy flows, then you will be happy. And again, now you have the technology and tools that  previously you had to convince people to come and join you. Now you can convince AIS with a little  bit of electricity and a few tokens to come help you. And when you think back in your life, what  you would say is a cornerstone stories of event that happened in your life to make you into such a  passionate and also clear thinker. Well, you know, I was a high-flying hedge fund manager. I started  being a hedge fund manager at 23 actually. So, I was one of the youngest ever. And I was like, no,  this is fantastic. It's all fun. Nothing wrong.

[00:23:51]You know, I've got my Oxford degree and everything  like that. And then my son was diagnosed. It takes you take a step back. What's important? I quit.  And it's very difficult to quit sometimes. But apparently for me it's a bit easier when you  realize the world's going another way. So there was one quit. Then it was launching stability and  then literally take another step. I think where am I really going? What's my direction? And where  am I needed? Like again quit. Now I know what I'm going to do for the rest of my life, which is  I'm going to help sovereign AI everywhere increase the intelligence of humanity leveraging the skills  that I've built over so long with the people that are like-minded alongside me and it takes time to  that. I mean, it took me 20 odd years to figure that out. What are your, you know, otherwise for  the entrepreneur out there or for the people who really want to be better themselves? I would  say again, it's very difficult to get out of connection with thinking about this. The most  beautiful thing is when you create together with other people. So I give the example of like  you know you eat alone, you eat with your family, you make dinner with your family. If you do  the third, you'll have a stronger family. If you vibe code, you can vibe code by yourself or  you can build things with others, your children, your nieces, like again make it family kind of  events. Don't forget about people in all of this and really take advantage of the wonder of what we  have today. You've been told throughout your life that you're not creative, that you can't create,  that you can't make a difference. Now, anyone can make a difference unlike anything you've ever seen  before. You just got to be passionate and apply yourself and find other people to go along the  way. It's the most exciting time ever, you know, like I'm really passionate about this area.  Let's create the most beautiful interactive website about South Californian hotels, I don't  know, you know, or about deforestation in the Amazon. and you get together with some friends  and you're building it together with your own kind of claws or your own replets or your own AI  agents. That's a much more fun thing than doing it by yourself because the AI is still not quite  good enough, you know, just like it's so much fun. I'll give you another example. Do you have  a favorite book that's not been made a movie? Not been make I I do have one. It's called The Dream  Maker. So, it's about philos It's like a Sophie story or Sophie's story. Like it's a philosophy  turn into a story kind of thing. So, why wouldn't you get together with your friends and make that  movie with generative AI? That's a good point. But how fun would that experience be? Whiteboarding,  ideulating, doing different designs, different styles, creating characters. And so, you know,  just like sometimes I used to shoot movies with my friends, little ones, you know, as a comedy troop.  It was so much fun building stuff together. Yeah.

[00:26:21]But you could never visualize it because the props  were a bit crappy. So, you had to kind of imagine it. Just like when you're growing up and you're  playing with now literally you can have an output and then make that movie and put on a showing to  your friends and family. Maybe in the future there will be a new form of amusement park. People just  go together to creating something you know with AI everyday you can go with your friends and you  can make a wonderful music album using suno with music videos. Now, nobody listening to this would  have thought of that, but I'm telling you now, if you go and do it, it'll be a lot of fun.  I think if our job as a media is make sure uh that not make sure is we want to contribute  for the good influence in this world. So people actually uh I I believe we are those glasses  sometimes. Yeah. But I think it's important to have some good news out there, not only focus on  the fear. Yeah. The positive views of the future.

[00:27:14]Like here we just had the X-P prize vision prize  to create positive stories of the future. That's what we got to do because again there's the  terrifying stuff and that's all fine but is the most wonderful time to exist ever. Thank  you and mad for your time. My pleasure. Thank you for watching this new episode of Human Wear  Project so that we can bring out more amazing content like this one to you. Please like and  subscribe. Your support means the world to us.