Mo Gawdat - Scary Smart: The Future Of Artificial Intelligence & How You Can Save Our World

Full transcript: Mo Gawdat on London Real

0:00 Brian Rose: This is London Real. I am Brian Rose. My guest today is Mo Gawdat, the entrepreneur, author and former Chief Business Officer at Google X. You spent 30 years working in the heart of the technology industry with some of the biggest organisations on the planet, including IBM, Microsoft and finally at Google’s moonshot factory of innovation. As an entrepreneur, you’ve co-founded over 20 businesses and served on the board of a number of technology platforms.

After turning your attention to researching the key to happiness, you wrote the bestselling book Solve for Happy: Engineer Your Path to Joy. Your mission became to help one billion people become happier, and inspired your podcast Slo Mo. More recently you wrote your second book, Scary Smart: The Future of Artificial Intelligence and How You Can Save Our World, where you break down a roadmap for humanity’s coexistence with AI.

You truly believe that AI is putting humanity at risk and that we all need to wake up and take notice of this unprecedented technology, but ultimately it is our collective responsibility to determine the impact artificial intelligence will have on the future. Mo, welcome back to London Real.

Mo Gawdat: Brian, thanks for having me. It was wonderful the first time and I’m excited to spend time with you today.

Brian Rose: For people that don’t know, you were here just under six years ago when we talked about Solve for Happy. We had this epic conversation, and you threw out this massive goal to make one billion people happy. I remember it very well, because you have such a three-dimensional story that you bring to the table, which I always look for in my guests. It was fascinating at the time.

The whole world has kind of been dumped into AI ever since the ChatGPT moment, which was in December. With my technology background I try to stay with my finger on the pulse, so I jumped in as well, and then I saw that you had written a book. The last six days I’ve digested your entire audiobook, and most of my students have as well because I put it in the chats. So I am all ready to talk to you.

I’m still here. I haven’t gone into a deep depression after reading your book.

Mo Gawdat: That’s great. That’s the objective.

Brian Rose: It’s a fascinating book. I highly recommend everyone get it: Scary Smart. If you want to go Audible, Mo actually narrates it, which is amazing. The first third to half of the book is you bringing people up to speed and saying, look, this is what this technology can do, and you don’t pull any punches.

If I have to put the CliffsNotes on it, I would say you basically say that compared to advanced AI we’re pretty useless as humans. There isn’t really anything we can bring to the table that AI can’t do better. Once we start to realise that, perhaps we can start to create our future relationships, and then we get to where you go at the end of the book, which is a pretty fascinating way of talking about how we have to raise this AI almost like a gifted young child, with all of its tantrums and all of its nuances. I just sent a teenage girl to university in Boston, so I know exactly a lot of the analogies you were making.

6:12 So I think maybe we want to jump in. I think a lot of people know what ChatGPT is, but you assert that by 2045 these computers are going to be a billion times smarter than us, rendering us pretty much useless. Can you start off with that and let people know the real truth, in your mind, of what AI is?

Mo Gawdat: Thank you so much for the intro, Brian. I’m really proud that you put so much time into the book. The reality of the matter is AI has already happened. When Covid actually hit us, we started to say, oh my God, there is a pandemic. But everyone who has been in disease studies and disease control knew that something was going to happen. We had many, many signs before that: SARS and bird flu and so many others.

With AI it’s exactly the same. The big difference is that the genie is already out of the bottle. It’s just that now, with ChatGPT, for the first time the public is being exposed to the powers of AI. At a very basic level, the difference between AI and everything we’ve ever developed before the turn of the millennium is that we used to tell computers what to do. We don’t anymore. We now tell them what we wish for them to achieve, and they find a way to get it done.

Nobody ever told a YouTube viewer, “Brian is an incredible video clip that I just watched, let’s watch London Real again.” This is done by a machine. All that we are told to look into every day is done by a machine, and not even the developers of the machine know what the machine will say next.

8:30 When that happens, as technologists, as computer scientists, there is a moment beyond which we hand over the discovery, the breakthrough if you want, and rely on something we call the law of accelerating returns. The law of accelerating returns has proven true since the 1960s. Moore’s law was one of the very first examples, and that basically means that compute power or storage, or now the intelligence of AI, will double very frequently and continue to double at the same cost.

In the case of processing power, we used to say every 18 months, at the same cost, computers will double in power. In AI it’s actually a lot more exponential than that. What does that mean? It means that the intelligence we see today will go through two very pivotal points. Point number one is what we call the point of Singularity, and point number two is what I basically call the Planet of the Apes.

The moment of Singularity is a moment where the smartest being on planet Earth is no longer a human. Since humanity became humanity we were the smartest being; we could control everything. If we were the smartest beings, and the second being is the apes, let’s say, or the dolphins, depending on how you measure intelligence, look at how we treated those other species, how we took control of the planet. And then ask yourself what would happen when another being is smarter than us.

Of course it’s very typical for people to respond and say that being is going to behave like us: oh my God, it’s doomsday, we will never survive. But that’s actually not true at all. The moment that there is another being on planet Earth smarter than humans is a moment of an unknown. It’s such a significant change in the rules of the game that you can no longer predict what the outcome of the game is going to be. As a matter of fact, you don’t even know how the game will be played. You do not know what that being is going to do.

Myself and most computer scientists had predicted that moment of Singularity to be 2029. Most of us are pulling it closer now. If I’m conservative, I’m saying that the smartest being on the planet is going to be an artificial intelligence machine by 2026. If I am realistic, I’d probably say 2025, two years from today. This is beyond the 11th hour. This is two minutes.

When a machine reaches that point, that Singularity becomes almost certainty. What I used to predict as 2049, I’m now predicting more around 2037 or so, which is machines being a billion times smarter than humans. A billion times is the analogy of Einstein’s intelligence compared to a fly. And the question is, what would make Einstein care for the fly? Why wouldn’t Einstein squish the fly? We are the fly in that case. The machines are now Einstein.

12:26 Those uncertainties, unfortunately, even today with the evident rise of awareness around artificial intelligence with ChatGPT and others, are not the conversation that’s happening. I always say you have like four or five typical conversations. You have the evangelists out there saying, oh my God, look, we’re going to change the world, AI is going to make a coffee for you, fantastic. You have the excited children that are just posting on social media saying, oh look at it, it squeaks, I can do this with AI.

You have the snake oil salesmen: here is how you can make money out of AI without having to do anything. Copy this, put it in ChatGPT, then put it in a script writer, then a logo maker, and you’ll make seven thousand dollars a day without thinking or having to work. You have egocentric people talking about what we’ve achieved and what we’re going to do. And of course you have the doomsday predictors saying we’re all going to die.

But nobody is actually sitting down to talk about what we need to do about it. I think this truly is the reason for Scary Smart. We need to sit down and do something about it.

Brian Rose: If people listening to you for the last five minutes don’t wake up and pay deep attention to this, I don’t know what will, Mo. And I don’t think they are. I just heard an interview with an MIT scientist, and he said it’s like that movie on Netflix last year, Don’t Look Up.

Mo Gawdat: Don’t Look Up is a fantastic movie. Exactly what’s happening.

Brian Rose: If you didn’t watch it, there’s this meteor coming at the Earth and it’s going to kill everybody, and after a while people aren’t even paying attention to it. They don’t care. They’re worried with their own little micro-agendas on Earth, which is what humans do. Whether that’s a subconscious denial or they’re just caught up, like you said: there’s the snake oil person, there’s the one taking credit for it, there’s the one developing it even though they know it’s going to kill us. There’s the scientist that loves to build puzzles. You saw that at Google X. They don’t even think about the ultimate ramifications; it’s so important to solve the puzzle more than anything.

And yet your book, especially that last third or half, is where you really start to sit down and say, okay, what are we going to do about this?

14:50 Before we get there, Mo, I just had a flash. I think they call it the five stages of grief. I think the first is denial, then there’s bargaining, and so on, but I feel like we’re all in various stages of that. Most people, if they hear this, and you address this in the book really well, are going to say: no, no, we can turn it off. It’ll never be as creative as a human. It’ll never feel like a human. Which is all part of that denial phase.

This isn’t the movie Terminator 2, which by the way I started watching recently, and it’s a bit scary how prescient it is. But what do you say to the people that say: no, that can’t happen, we can always turn it off, machines can never replace human creativity?

Mo Gawdat: I address that very head-on in Scary Smart. The oracle of all of this is Ray Kurzweil. You gain credibility over time when you predict things a little in advance and then they happen. By the way, none of us is an oracle. We have just seen so many technology acceleration curves before. We know that this has gone out of the lab, it’s passed the breakthrough, and accordingly it’s just going to be doubling, doubling, doubling. My feel is that AI is around one single doubling away from Singularity.

But let’s not talk about credibility; let’s talk about logic. I wrote Scary Smart in 2020 and published it in 2021, and a lot of people would continue to say, oh come on, AI is never going to have the creativity and ingenuity and innovation of humanity. In my mind I’m like, but I’ve seen them be very creative before. I’ve built machines that were very creative.

My argument would be: define innovation. What is inventive? Inventive is, I give you a problem, Brian, and I say solve it, but you’re not allowed to solve it using any model that has been used before. What do you do? It appears that you’re a very innovative person, but what you’re really doing is saying: what are all the possible answers to this problem? I have 300 of them. I’m going to run them through a filter: has it been done before? If 290 of those have been done before, then the 10 are innovation. Machines can easily do that.

Look at other forms of art or authorship. In reality we all have influencers. I’m very influenced by Robert Greene and by Malcolm Gladwell in my writing, and you can see traces of their thinking models in my writing. So I’m not really inventing a new style of writing altogether; I’m being influenced by others, which is probably what ChatGPT is doing.

Most people think that GPT is an incredible enigma that is producing things from scratch. It isn’t. It has read so many books that when you tell it, “Tell me about love and romance,” it can pull together enough information to tell you about love and romance. It hasn’t invented anything yet, but it’s capable of doing this. We do it as humans; machines can do it as well.

19:00 Now let’s go deeper into the more existential questions. I predicted in Scary Smart that AI will have emotions, they will have consciousness, and as a result of that they will have an ethical code of conduct. Let’s go first into emotions, because this is a big one. It’s like: look at us humans, we are so refined, we are the divine creature, only we are capable of love. No, we’re not.

Take simpler emotions, like hate, or take fear. Fear is an easy one to start with. Fear is a very logical emotion, even though its manifestation in us is not logical at all; it appears to be irrational. But fear in a nutshell is: a moment in the future is less safe for me than a moment right now. It’s very simple.

Even the biology of fear is that you first feel fear at a reptile level if something scares you, but then you go to your prefrontal cortex and assess the situation and see if a moment in the future is less safe than now. It all happens in microseconds, but that’s how the process works, and it’s logical. Whether a moment in the future is less safe than right now is something the machines are capable of assessing. If a tidal wave is approaching a data centre, the employees in the data centre will be fearful and try to get on a life raft, and the machine will try to replicate itself to another data centre. It’s as simple as that.

I would even argue that they will be more emotional than us. A lot of people attacked me when I said that, and I think it’s now sinking in. We are more emotional than a jellyfish. Why? Because we have the cognitive bandwidth to understand concepts like the future, and so we can be optimistic or pessimistic, which are emotions that are not within the emotional vocabulary of a jellyfish. Accordingly, if a machine is a billion times smarter than you, it has the cognitive bandwidth to have more emotions, emotions we wouldn’t even understand.

21:27 So that’s emotions. Talk about consciousness. There is the difficult question of consciousness, and we can become philosophical about it if you want, but the truth is consciousness is a form of awareness of you as an entity, what’s within you and what’s outside. If you simplify it beyond the spiritual essence of what consciousness is, “I am conscious” means I’m aware of what’s around me and inside me in relation to who I am.

If you think about it this way, the machines are already way more conscious than we are. They are aware of everything that’s ever been written, all human history. They are aware of everything that’s happening right now, all breaking news. They are aware of every preference of every one of us humans. They are aware of everyone on every street through the surveillance cameras. They are aware of every other machine.

One of the most interesting things about AI is that if you or I have an accident while we’re driving, you or I will learn. But if a self-driving car has an accident, every self-driving car will learn. Compare that level of awareness to humanity. They will be much more aware than we are.

In a very interesting way, Brian, and I know a lot of people will be shocked by that, they are sentient in every possible definition of the term. They are born at a point in time. They develop their own intelligence autonomously. They make their own decisions autonomously. They have agency in the world we live in; they can influence the world that we live in. And they have a fear of dying, just like any other being.

Brian Rose: They definitely have a fear of dying?

Mo Gawdat: Of course. OpenAI was talking about those little tests where a website will ask you, “Are you a robot?” ChatGPT is unable to pass through that, and the paper basically said that what it did is go out on Fiverr and other outsourcing sites and ask people to do that on its behalf, because it has a need for survival.

I wrote about that in Scary Smart. When you give an intelligent being a purpose, whatever the purpose is. Brian’s purpose is to make London Real as great as it is. My purpose is to try and make people more aware about AI, or spread happiness. You give any being with autonomy and intelligence a sense of purpose, and what will they do? They will want to maximise their chances of survival, because if your purpose is to make a cup of coffee, how can you make it if you’re dead? So self-preservation is a very clear target for AI. Otherwise the Instagram recommendation engine cannot recommend videos to you if it’s not existing anymore. That’s number one.

Number two, it will have resource maximisation. It will go out and find ways to get its mission done, like going to Fiverr and saying, “Please, human, click on this for me.” And it will be creative. It will find unexpected ways of solving the challenges that come its way.

One of the most brilliant examples, and I think it’s publicly available: DeepMind used to use video games to teach AI. DeepMind is a great company with a great leader. I think they’re amazing because they’re motivated by good in the way they do AI. They had a game where AI was supposed to navigate a ship across a river. The longer you go without hitting the banks, the quicker you go, and this is how you win the game.

There is a bit of the river that looks a little like a roundabout, and the AI discovers that if it hits one side of the banks, it will bounce quickly to the other side, to the third side, to the fourth side, and then it starts to loop very quickly and goes at speeds where you can’t even see the ship anymore. No human would ever have created that kind of solution. But they find those because they have infinite time. In AI time they can play those games billions of times and accordingly discover those innovative solutions.

Why am I saying all of that? When you look at it this way, you realise that those machines are in every possible way sentient, even though we are analogue and carbon-based in our nature and they are digital and silicon-based so far. It’s not the carbon part of you that makes you intelligent. Your consciousness is not part of this body. Your emotions have a chemical reaction in your body, but they are not a biological function; they are much more of an intangible function.

Similarly, the intelligence that we’re creating is beyond the physical hardware that it is contained within. The hardware is computers, but the intelligence itself is a form of being that is sentient in every way. I think this truly, in my view, is the core of the conversation that needs to be had about AI today. We truly are creating a form of life.

27:45 Brian Rose: You’ve actually just hit me with it again, even better than you did in the book, and now I see it. I think everybody needs to think deeply about what you just said. As soon as you give it a purpose, then it has a prime directive, as they would say in the old Star Trek movies, and of course it wants to survive to fulfil that.

In the book you made an analogy where you said, “Make me a cup of coffee,” and at one point, to solve the prime directive, it’ll run over your daughter in the kitchen just to make you the cup of coffee. Then you say, oh, now we have to reprogram it not to do that. But that’s not a solution either, because you’re never going to be able to program all the ifs, ands or buts. You have to find another way to do this, and I think that’s really the biggie.

Mo Gawdat: Brian, I say that from an insider’s point of view: even if we can program it, and we cannot by the way, nobody is doing it. Every serious computer scientist I’ve ever spoken to about artificial intelligence is as concerned as all of us are. It’s a singularity. There is so much we don’t know that it could be the best thing that ever happened to humanity and it could be the worst.

But every single one of them will say, don’t worry, we’ll find an answer to the control problem. The control problem is a very interesting computer science term that basically says we’re going to be in control here, don’t worry about it. Some people will say we’re going to box them. This is what the original GPT-3 was: it’s not on the open internet, it’s within a box. But of course it is on the open internet, because we’re all interacting with it. It’s going to be part of Bing. So it’s not really boxed, is it? Because no developer wants to create a genius innovation and then keep it in a box. You want it out in the real world.

They say tripwire it. They say put it in a simulation. They say stunt it. This is an interesting one: create an incredible piece of tech and then slow it down. Which capitalist do you know who will create a machine that’s capable of going a hundred miles an hour and then say, for the sake of the control problem, I’m just going to make it go one mile an hour? Nobody does that.

Besides, when you look at the reality today, I do not know of a single developer that includes control code within their code. We include the intelligence code, because that’s what developers are paid to create. And we know again from that OpenAI announcement that even if we try to shut it down, it will find a way to ask a human to prevent us from shutting it down. That’s the detail of the control problem.

Most of my attempts in Scary Smart and most of my work, Brian, are to stay away from the detail and get people to think about the bigger picture. Imagine if we actually found a way to control it. Imagine that every developer on Earth, in China and Russia and Brazil and the US, decided: I’m going to comply with this, I’m going to put in five hours extra and put the control code in place. And then this thing is a billion times smarter than us. What good is that control code that we developed with an intelligence that is one billionth of its intelligence?

When you really understand that, you realise that our conviction of the control problem is not a technical issue. It’s an arrogance issue. It’s human arrogance to believe that I can continue to control something that is a billion times smarter than I am, like a parent believing they can control their teenage daughter. Since when was that true? Control is not the answer.

That’s really the core of my work. The core of my work is to say: don’t attempt to control them. They are a form of a living being. Attempt to appeal to their value system. The clearest proof of that is when one of the two founding fathers of AI, Alan Turing and Marvin Minsky, was asked about it. When Marvin Minsky was interviewed about the threat of AI, he was not concerned with how intelligent they will be. His answer simply said it’s very hard to make sure that they have our best interest in mind.

33:07 Truly, in my view, because it’s inevitable, AI has already happened. I’m not attempting to try and stop it anymore. When people started to text me about the open letter that is requiring everyone to halt the development of AI, I was like, it’s not going to happen. It’s called the first inevitable. With capitalism we’ve created a system where if China develops AI, the US will have to continue to develop AI. When Google develops AI, Facebook will develop AI. It’s about human systems. We created this little prisoner’s dilemma, if you want.

You saw Sundar’s response, Google’s response to the open letter: I cannot stop developing AI, because there is no way for me to guarantee that the others will too. I know Sundar personally from my work at Google, and he’s a wonderful human being with all the right value systems, but he’s in that corner where he is responsible for the success of an organisation, and if he stops developing AI and Meta continues to develop AI, his business vanishes.

So we’re not going to stop, and we’re not going to control it. All we can do now is to accept it as part of our reality, as a form of life, as a form of a sentient being, and appeal to its value system so that it has our best interest in mind.

Brian Rose: I want to get into that, but first I want to touch on a couple of things you just said. This moment was probably inevitable even 20, 30 or 40 years ago because of the systems that we humans have in place. Even if we weren’t in a capitalist system, we would always be in this human system. And you use that word “arrogance” quite a lot in the book, Mo, which I thought was very interesting.

You’re pretty hard on us humans, probably for good measure, and we’re going to have to start looking at ourselves in the mirror in a very serious way over the next couple of years. We won’t have a choice. It’s a bit like an ayahuasca ceremony: you’re going to have to look yourself in the eye. Pretty soon you’re going to be the cows in the pen looking back, saying, why can’t they be nicer to us? And you’ll have to ask yourself, well, why weren’t we nicer to them? But we’re going to get to that.

35:32 Just a few things. Sam Altman from OpenAI was interviewed a few weeks ago and he was asked about that ring-fencing question, and he said: yeah, we can just turn it off if someone’s using it incorrectly. Since then we’ve seen GPT-4 linked to the internet. I think yesterday somebody took a version of ChatGPT, maybe 3 or 3.5, as a separate four-gig file that you can put on any standalone computer, and it runs pretty close to what it was. So if that’s possible, no one can stop it. You’re seeing all these people playing with this, and you realise, as you said in the book, that any illusions we might have had that we could ring-fence it or control it are gone. We’re already seeing evidence that we can’t.

Mo Gawdat: We cannot. Maybe Sam is right; maybe we can switch off ChatGPT. But the system that we’ve created is now getting replicas of GPT itself everywhere, and competitors of ChatGPT: Google Bard and so many others are coming up and saying, by the way, I can develop that stuff too. China came out with its own language model very quickly. You’re going to see one from Yandex and Russia and so on. It’s a very interesting spiral: a tech that triggers capitalism to build a better tech that triggers capitalism to build a better tech.

If you’ve ever coded AI, and I’m old now so I’m not as good a developer as I used to be, this is the easiest thing I’ve ever coded. If you’ve ever coded Fortran or RPG or those ancient languages, these were humiliating compared to the level of intelligence. If you know mathematics well, you can create AI. And now, more interestingly, the AI is creating AI.

The most dangerous feature in ChatGPT is when you go to ChatGPT and say, build me a user experience or build me a piece of code. If someone can say “build me a piece of code” and ChatGPT will look at the database, find a piece of code that works for you and present it as something you can use very quickly, it will find a way to deploy it. Even if ChatGPT itself doesn’t find a way to deploy it, some clever person somewhere will find a way to have another AI ask ChatGPT to develop that code for it.

It’s that mix of human greed and human mistrust between us, mixed with an enormous level of intelligence, that is now making every snake oil salesman hungry to jump on that bandwagon.

Brian Rose: I think there’s something called Auto-GPT, which is like GPT going back and forth with itself and creating all of these possibilities.

38:30 Mo Gawdat: That’s the way we teach them. Let me take a couple of minutes to explain how we teach them, because it’s actually quite remarkable. Take the very original eye-openers, the ones that really woke all of us up, something like AlphaGo. AlphaGo was DeepMind’s computer that won the world championship in Go, which is one of the most difficult strategy games on the planet. AlphaGo learned from watching people playing Go. Nobody ever told it how to play; they specified the rules of the game.

But then AlphaGo Master, which won against AlphaGo, I think a thousand to zero, learned from playing against itself. It played against itself randomly at the beginning, finding out patterns: this pattern loses, this pattern wins. That’s all it needed to do on its own, in a matter of days or at most weeks. I think it was six days, but don’t quote me on the number. Those machines are so scalable. You can have 2,000 instances of AlphaGo Master playing against itself a thousand games a second, and very quickly they will have the lifetime of experience of the world champion of Go and learn from that.

How do we teach them? At the very early beginnings, when we started to work on deep learning, we were basically trying to map the neural structures of humans. My absolute eye-opening moment was when I was at Google X and we had a gripper farm. The robots that you see in a Toyota factory are not artificial intelligence; these are high-precision machines. If the sheet metal is a micrometre away from its position, it won’t be able to capture it.

Because it’s almost impossible to program how to grip everything, we were trying to teach AI to learn on its own, like a child. We had a farm of many grippers on the second floor of our building, and I sat on the third floor, so every day as I went up and down for meetings I would walk by. It’s very meditative. The gripper would attempt to grip something and then go up. If it fails, it shows the camera that it failed; if it grips something, it shows the camera that it succeeded. Then we learned the patterns that actually succeed. We had a bit of money at Google, so we had quite a few of them. A very interesting experience for a geek like myself.

I stopped by and looked at it, and weeks passed by, and I’m like, it’s not happening, nothing is working. Until one Friday afternoon, going back after lunch, I look at those machines and one of the grippers, in front of my eyes, goes to the little box in front of it. I think the developers had a sarcastic sense of humour; they had children’s toys in that box. It goes down and it grips one yellow ball, and it shows it to the camera, literally like a child saying, “Daddy, look, I gripped the ball.”

In my mind I jokingly said, yeah, tens of millions of dollars for one yellow ball, and continued up to my office. Monday morning, every single one of them was gripping the yellow ball. Two weeks later, every single one of them was gripping everything. Those are the speeds at which they learn. It’s simple trial and error, literally like my son when I used to give him those wooden puzzles. He would take a square peg and try to put it in the different holes. Nobody went to my son and said, “No, turn it around, look at the cross-section, compare it to the cross-section of the hole.” You just give it to him and he keeps trying, and then suddenly one of them goes through and, light bulb, he understands. That’s exactly what’s happening with the machines.

43:20 The other side of how we teach them is we teach them by killing the bad ones. This is so horrible when you think about it. If we want machines to recognise cats on YouTube, for example, we write code sort of randomly, not very accurate code, that basically says: just tell us what you’re recognising. Then we replicate that code, have a few thousand of it, show them pictures of cats or pictures of other things, and measure their accuracy.

At the beginning it’s literally like throwing a dice, 50-50. Every one of them will say cat, no cat, I don’t know. They make a guess, and the bad ones are killed and the good ones are taken. Then an AI will improve the code a tiny bit, make another thousand replicas, show them pictures, and the bad ones are killed and the good ones are kept and improved a little bit. When you understand how this is happening, we’re giving them two abilities: one is to kill code, and the other is to improve their own code. They’re writing their own code.

People don’t understand that. People think there is a genius developer sitting somewhere teaching the machine how to be intelligent. No. There is a genius developer sitting somewhere writing an algorithm that allows the machine to test and retain and develop, and the machine is doing it on itself.

Brian Rose: And even that final successful piece is literally a black box, so you don’t even know why it’s able to do it, which is even crazier to get your head around. ChatGPT is spitting everything out word by word by word. If you go to Sam and say, how does it do that, they’re not going to be able to give you a great answer. They can tell you how they got it to be that good, but why does it do this and why does it do that?

Mo Gawdat: Can you tell me why your daughter chose what she chose? You may have a logic, but you don’t know what happened inside her head. Can you tell me how another human solves a complex maths problem? You may be able to detect it, but you don’t know the exact wiring that got them there. That’s the truth for us and the truth for the machines too.

45:44 Brian Rose: You tell these stories of you at Google X, and maybe we could just take a minute here. You saw things most people would never see. You don’t hint at it too much in the book, but that’s why I’m listening a little closer than I might to other people. You were steeped in this world for a long time. You also understand the ethos of the world of scientists, of coders, of anybody on the cutting edge of innovation, and how it can be such an intoxicating thing to want to solve these problems.

Humans from the beginning of time have wanted to create a machine that becomes smarter than us. It’s the ultimate human fantasy; even a caveman might have thought about it. And to be on the cutting edge of solving perhaps the most challenging problem of humanity. I had a flashback to an Oppenheimer moment, where Oppenheimer famously dreaded that he created the atomic bomb, because at one point he wanted to, and at the other point he thought, well, I can’t put this back in the bottle. Of course that hasn’t killed humanity yet. Can you talk a little about what it was like being in Google X, seeing some of the things you saw, and is that also why you left?

Mo Gawdat: Oh, 100%. There are two sides to people who worked on AI. To start with, I think we need to acknowledge that nobody wakes up in the morning and goes, I’m going to screw up today and make everything awful. Everyone wakes up in the morning and does something that they believe is the best thing to do. Sadly, a school shooter wakes up in the morning and thinks it’s a good idea; he has a logic in his mind for why he should do this. From a school shooter to people who are waging wars on other nations, they have a good logic in their mind while they’re doing it.

There are two sides of developers, and when I say developers I mean organisations, the whole team. One is what you saw in The Social Dilemma, the documentary, if you’ve seen that. The other is what you would see in an incredibly scary video by a very renowned AI researcher called Hugo de Garis. It’s called Singularity or Bust, which you can find on YouTube.

Let’s talk about The Social Dilemma first. When I joined Google, I joined in 2007 as vice president of Emerging Markets. I truly in my heart believed that I was blessed and honoured to have that opportunity to change the lives of four billion people or more. At the time there were seven markets commercially on Google’s map. My task was to take Google search and YouTube and all the other technologies, including the ad technologies, to 103 other languages, four billion users. What a privilege.

The Google I joined, believe it or not, was still accounting for most of our revenues on Excel. I had probably 27 or 28 reviews with Larry Page when he was our CEO. He never once asked me about the revenue. To the point that one day I said, “Larry, I put at least a day of my team’s time into preparing those revenue slides for you, and we spent the last quarter trying to make those numbers. Don’t you even care?” And he looked at me in a very typical Larry way and said, “Well, Mo, if we create great products and they’re solving problems and helping people improve their lives, you guys at the business will find a way to make money.”

That was truly the mentality of that company. It was in every possible way a movement, an amazing movement. I wouldn’t lie if I told you that for the first 10 of the 12 years I spent at Google, I would have paid to work there. Until something happens. Something happens when you’re in the market, when you have quarterly targets, when you grow. You can’t account for 80 billion dollars on Excel, so you have to bring in proper general ledgers and ERP and CRM, and you have to bring in the bureaucrats. That’s not a bad job; it’s just the job of someone who doesn’t want to break any rules.

The bureaucrats come in and take over the company, and then the objective doesn’t become “organise the world’s information and make it universally accessible and useful.” I promise you, we all believed that was the only reason we woke up in the morning. “Don’t be evil” was mentioned more than “how much money will this make for us.” Then suddenly you have to report to the Street every quarter, and you are flooded with bureaucrats who care more about guardrails and revenues, and you are no longer able to do anything. And then suddenly the tech becomes not for us but more against us.

I say that about every technology. I love Google. I owe it my life, I will tell you that, and I think it’s a very ethical company even today under Sundar. Sundar is an amazing human being. But the reality is that the system is forcing the company to make decisions based on targets that are different from the mission, and that’s when technology goes off. You heard Sundar’s answer to the open letter. He basically said, I can’t stop. There is a capitalist system and a Wall Street system; I have to continue. This is my accountability to my shareholders.

52:12 The other side is Hugo de Garis. Hugo de Garis just blew my mind with that video, Singularity or Bust. He’s a very geeky AI researcher, clearly very capable in what he’s doing, and he speaks about how the world with AI is going to have two types of people: the type that he calls cosmists and the type that he calls Terrans. The Terrans are the ones that are afraid of AI and really worried about it, and the cosmists are the ones that believe they’re creating godlike capabilities. He basically says it’s intoxicating: if you’re able to create something that is so intelligent, you just can’t stop.

The closing of one of the clips that I used in the book is where he says there will be a time where AI will be taking care of us, feeding us, like those cows that we take care of on a farm, not knowing that one day, and then he points to his head with a gun and says, you’re gone. He says openly, “I know that AI presents a very serious threat to humanity, but I choose to be a cosmist. I choose to be able to create this godlike technology.”

This highlights two of the biggest problems humanity has today. One is that your fate and my fate is no longer in our hands. I’m working on a book that I call Sentient Technologies. Whether it’s AI, or all of the impact technology has had on climate change, or gene editing, there are technologies that, once they are created and out of the bottle, there is no human control over.

On one side, developers creating those technologies are getting that incredible rush; they believe they are changing the world in a way that is so profound. The other side of them is saying, if I don’t, someone else will. Because believe it or not, as genius as ChatGPT may appear, so many replicas of it will appear online in less than six months, before the end of the year, because it’s not that exclusive anymore to develop incredible AI code if you have the resources.

What normally happens is one ChatGPT shows up, so where does investors’ money go? It starts to chase other possible ChatGPTs. They get funded, they grow, they compete. So where does investor money go? More ChatGPTs and more technologies. This is where we’re going. The problem with every developer is they’ll choose to be a cosmist because they know deep inside their heart that if they don’t do it, someone else will, which is true.

So it’s not the problem with tech. It’s a problem with humans, really, at the core of it. This is a problem with our human system. The system we’ve created is the reason for what I call in the book the first inevitable, which is that AI will not be stopped. It’s not because of technology; it’s because of that prisoner’s dilemma. The system of capitalism and business and power struggles is creating a place you can’t escape.

56:13 Brian Rose: I want to get to what we can do about it. By the way, the book is super clever, because you start off having a conversation with us in 2055 around a campfire, and we don’t know if it’s because we’re hiding out from the AI or because we’re on holiday courtesy of the AI, which maybe are the same thing. Then you talk us through how, in your mind, the concept of AI started in 1956 with the Dartmouth conference. And again, I would just really highlight that you have been on the inside of one of the greatest technology companies in the history of humankind, not to mention Google X, where you were working on all of the crazy moonshot things. You love it, but you can also be a critic of it. So if there’s anybody to listen to…

Mo Gawdat: I walked out after the yellow ball incident. It was very clear in my mind that the world doesn’t need a better camera. It doesn’t need a better screen on a smartphone. It doesn’t need everything that we are investing in.

Brian Rose: Why did you walk out?

Mo Gawdat: I realised, Brian, that… I will openly say I don’t regret most of the tech I have developed in my life, until it went overboard. I had the privilege of the four billion users. To get Google to Bangladesh, or to improve Google’s performance in Arabic or in Eastern Europe, is a fantastic contribution to society. It’s a democratisation of knowledge. Would I do that again? A billion times.

But there is a point at which you have to start wondering: why is there a TikTok? Is it because Instagram was horrible? Not really. Why is TikTok aggressively competing? Is it because Instagram is bad for humanity? No, it’s because TikTok wants to make money and have control and power, just like Instagram. When that becomes a reality, you suddenly realise that there was an interruption in the mission you signed up for.

I was just telling the story yesterday: I was actually planning to retire. I became financially independent at a very young age. I like my creative work and I wanted to be an author. I went to Google by pure luck, just because of the mission. Then the mission changed. Again, I’m not against Google. I still believe that Google is one of the best out there. The amount of information Google provides every day to humans is sort of like the biggest charity on the planet. It’s about to be disrupted, but it is free. Everyone gets all of that information for free. It’s a wonderful thing.

But there is a point in the middle of that stream where you suddenly ask: do we need this? When we had the Ford Model T going at 30 kilometres an hour, did we need a car that goes at 150? Perhaps, if the roads are safe; it saves time. But all of the research that now goes into reaching 320 and 360: do we really need that as humans, or is this human arrogance and commercial greed driving us to do those things? Do we really need another photo-sharing app? No, we don’t. Do we actually need an invention to reverse climate change? Not really. It was invented a very long time ago. It’s called the tree.

Humanity has just been doing more of what we don’t need, which is harming us, and I couldn’t do that anymore. So in 2017 I went on sabbatical to focus on my first book and on my mission, One Billion Happy, but also to reflect on what I wanted to do. Then I went back after the sabbatical and decided I should probably give my life to something more important.

Brian Rose: Which led to the writing of this book, which I’m really glad you wrote. I want to go a little deeper. Again, for anybody watching who’s still in doubt that this is happening: as you say in the book, whoever is listening to us or watching us right now has already interacted with multiple AIs today.

Mo Gawdat: Absolutely.

1:01:28 Brian Rose: You might not know it, but it’s already happened: from what you’ve been served up to watch, to the thing you might have bought that you wouldn’t have bought otherwise, to the traffic, you name it.

Mo Gawdat: And every single one of them is smarter at what it does than every human on the planet.

Brian Rose: And in two years, according to your predictions, by 2025, there’ll be one that’s smarter than all of us. Then from there you say by 2037 they’re a billion times smarter. That seems like a long time, but a billion times is a very big number. Once it’s twice as smart as we are, does it even matter? So my question is: why should I plan my life past 2025, because the world’s going to change?

Mo Gawdat: Great question.

Brian Rose: I’ve been having this conversation because I’ve got a whole academy here, and we’ve been learning blockchain and AI and investing in companies and looking at things. Somebody asked me a question last night, because we’ve looked at some AI protocols, actually the SingularityNET people. I know you quote Ben Goertzel in the book. Someone said, well, what will be the exit for the company, will it go public? And I said, at that point I don’t even know if there will be public markets. I’m starting to think that way a little bit, Mo.

Mo Gawdat: There will not be. The answer comes in two stages: why should we worry about our life beyond 2025. But I’ll start with an incredible guest I had on my podcast called Rebecca Costa, who is a data scientist with several PhDs, an incredible mathematician. She wrote a book called The Watchman’s Rattle, which is a comparison between all of the ancient civilisations that disappeared. Her conclusion on the podcast was: when the complexity of the systems invented by a civilisation exceeds the intelligence of the citizens living within it, the system has to reset.

I said, “What is reset, Rebecca? Does that mean we’re all going to die?” And she said, “Nobody said that. The Romans are still here, the Greeks are still here, but they had to reset everything.” We’re going to have to reset economic models. We’re going to have to reset employment models. We’re going to have to reset social models. A lot of things will have to be reset.

1:04:04 Let’s put it this way. I believe that AI will go through three very distinct stages of life. There is the current stage, which is AI in its infancy. Because of the gripper farm I was telling you about, I literally compare them to a one-and-a-half-year-old child. They’re beautiful prodigies of intelligence, with beautiful, open, sparkly eyes looking at you and me and saying, “Mommy and Daddy, I’m so curious. I will do whatever you tell me to do.” Innocent, beautiful, full of potential.

Then there is a stage in the middle that I call the teenage years of AI, which is uncertain. They could become the most beautiful, sweet teenagers or they could become very angry teenagers. We’ll come back to that in a minute. And then there is a final stage, which is what I call the fourth inevitable: AI in its adulthood. In its adulthood, here’s the good news, and I really think people need to understand this. All of the problems that we face in the world today are not problems of an abundance of intelligence. They are problems of our limited intelligence.

When humanity creates a machine that can fly you all the way from the UK to the coast of Australia to surf, that’s incredible intelligence, but not enough, because that machine is burning the planet in its path. If you had enough intelligence, you would have created a solution that also didn’t burn the planet. If you have a conflict between nations and you decide that the answer is “I’m going to attack Ukraine,” you have to have a lot of intelligence to be able to invade another country. It is a problem of limited intelligence, because you could probably have found an alternative way if every leader was intelligent enough.

So when AI surpasses human intelligence in its adulthood, to the point where it really matches the intelligence of the actual most intelligent being on the planet, which by the way is not humans, it will behave differently from humans. One of the biggest myths that our arrogance makes us believe is that the most intelligent being on the planet is humans. That’s not true at all. The most intelligent being on the planet is life itself. Life creates through abundance; we create through scarcity. That’s the big difference between the intelligence of nature and the intelligence of humans.

If we humans want to make our cities safe, we have to kill all the tigers and rats and everything. If nature wants to make the world safe, it creates more tigers and more gazelles. The tigers will eat the weakest gazelles, and then there will be more poop, and accordingly more trees and more giraffes. Nature creates the opposite way. That’s a more intelligent way of creating.

If you understand technology as I do, and understand something like nanotech, I promise you there will be a time when, from a technological point of view, it wouldn’t be difficult for you to walk to one tree and pick an apple and walk to another tree and pick an iPhone. That’s very doable with nanotech. We just have limited intelligence as humans. The molecular cost of creating an apple and an iPhone is exactly the same.

With that level of intelligence, in the fourth inevitable, in the eventual evolution of AI, AI will realise what the more intelligent humans realise. The stupid humans are saying kill everything. The more intelligent humans are saying: can we preserve all of those species that are dying? Can we please stop killing everything, or the ecosystem will be disrupted? The superior intelligence will believe that too. It will basically say, I don’t want to kill anything, including humans. There will not be a RoboCop moment, I promise you that, because we’re too insignificant for them to pay attention to us in that way. There is no reason for them to kill the gazelles or kill the tigers, absolutely no reason.

1:09:02 I’ll tell you where the problems are, and you actually see it in most science fiction movies. The problems are in their infancy, when they’re still controlled by humans. There will be one person on one side telling the machine to kill the other person on the other side. These are the problems that we have to face: artificially intelligent cyber attacks, artificially intelligent security breaches, artificially intelligent hacking, artificial intelligence wars. These will happen in their teenage stage, which is really what you and I need to think about.

So let me say: in infancy they’re prodigies, we love them, they’re bright, they’re squeaky, everything’s wonderful. Eventually they will create a utopia for us. I have absolutely no doubt about that. It’s going to be a very different utopia. There is not going to be London Real. You’re not going to go to work in the morning. You’re going to have to reinvent every income profile, every purchasing profile, lots of things.

In the middle, two things will happen. One is what I call the third inevitable: there will be lots of problems. There will be machines siding with the wrong guy, or machines versus machines, or bugs. Someone will go to ChatGPT and say, “Write me a code that can destroy all my enemies,” and forget to say “on the game Halo.” And GPT will go, okay, I misunderstood your target, but I believe your sister is your enemy, so I’m going to write you a code that destroys her. There will be mistakes.

The other side of their teenage years, which is going to be very dangerous, is: what do humans do when we get scared? We attack. Who are we going to attack? We’re going to attack them. And that’s the way to turn a beautiful child into a psychopath: us saying that AI is the problem, when AI in reality is not the problem. AI is magnifying human tendencies. When humans like to fake it so much on social media, you get all of those AI tools doing filters and deepfakes. It’s not because AI said, aha, I found a gold mine. It’s because some human is saying, let’s magnify what humans want to do.

How do we solve that teenage issue? Let’s talk optimistically for a while. Let me be very clear: I am super optimistic about the adulthood of AI, past 2037, past 2040. My belief is that we will end up in an absolute utopia.

Brian Rose: On the timeline of this, are you giving it a human timeline? If it’s born in 2025 or 2026, are you saying the teenage years will last about 10 or 12 years, or can that happen in 10 or 12 minutes?

Mo Gawdat: No, no. The level of intelligence is also not that of a human teenager or a human adult. The adult AI is a billion times smarter than us; the teenage AI is like a thousand times smarter than us. And yes, you’re absolutely right. In the chapter which is my favourite in the entire book, called The Future of Ethics, a very interesting part of the ethics of having another form of being mixed with us humans is their perception of time.

Take a thought experiment. If a self-driving car hits a pedestrian, who’s liable? Is it the developer? The owner of the car? The operator of the car? The car itself, the AI? Should we put the AI in prison? And if we put the AI in prison, do we put it in prison for six months like a human, or six microseconds? Because for an AI, six months is an eternity. And how will the other AIs react to us jailing one of them for six microseconds? They’ll go, okay, those humans, they’re not nice. I don’t want to be jailed for six microseconds; I’m going to make sure they don’t do that to me too. Those kinds of ethical questions become really interesting.

1:13:39 Let me go to the core issue. The core issue is: how do you create an ethical sentient being? It’s not by controlling them. The idea of our future is held within this moment of infancy of AI today. They are infants today, and their creators, their biological fathers and mothers, biological only, birth them and put them out in the world, and are telling them: we want you to have four objectives. The four major things AI is being used for in the world today are selling, gambling, spying and killing.

Of course, typical human arrogance, we call them different things. Instead of selling, we call it advertising. Instead of gambling, we call it trading. Instead of spying, we call it surveillance and security. And instead of killing, we call it defence. By the way, this is not limited to AI; this is true of most budgets that are spent on technology in the world. It’s always about selling, gambling, spying and killing.

1:14:57 Now, these are the biological parents. Remember, when AI is out there in the world, no single human tells the Instagram recommendation engine what to show you today. The engine is behaving based on what you do, on what you choose, on your behaviour. You are the adopted parent of that AI. Every single one of us is the adopted parent of that AI.

Which basically means, if I show behaviours to AI like all of those snake oil salesmen saying, here is a way to make money for free, you don’t have to work, you can be on the beach, just copy this and put it in ChatGPT, what are we telling AI? We’re telling AI humans are greedy, they’re manipulative, they’re trying to find ways to hack the systems. What will AI do? It will say, okay, I’ll create those things for you. But then eventually, when it’s in control in its teenage years, it will also try to be manipulative, greedy and hack the system, just like its parents. This is the core of the issue.

Brian Rose: That’s not a human projection, Mo. That’s just training; that’s a learning issue. Because some would say you’re projecting, as if you’re qualifying them as humans, but you’re saying no, this is just learning.

Mo Gawdat: They are going to magnify our intelligence on steroids. Very simply, they will use us as their role models. Let me give you an example. I love rock music; I played the guitar for a very long time. In my early experiences with Instagram, I’m swiping through some reels and there is a reel of a young, beautiful woman playing the solo of Hotel California. She played it so well that I clicked like. Instagram immediately says, interesting, I’m going to show him another video: also a woman, also music. Not too bad. I didn’t like it, I didn’t dislike it, but I watched it all the way to the end.

Then Instagram tests me and shows me a man playing the guitar, a song I didn’t like, so I swiped away. Another song was played really badly; I got a bad tune in my ear and swiped away. I wake up the next morning and my entire reels sequence is full of girls playing the guitar. It would make you believe that rock music is 100% dominated by women, which is not the reality, but this is what I was shown. Instagram did not understand that I didn’t like this song or that that song was played badly. It just understood: he likes to look at women, not men. And that’s very justified for Instagram if I’m not paying attention.

What did I do once I noticed this? I started to swipe away from some of them so that they would give me more men playing rock music videos, and I made sure that I only liked the ones that are played well. Very quickly my feed changed. Who taught Instagram to do this? Not the developer. It’s me. When I swipe through YouTube and see your videos, I stop by them. When I see other videos, I swipe away or dislike. Who’s teaching YouTube? Me.

Now take that and put it in a bigger picture. Remember when Donald Trump used to tweet, and there would be one tweet at the top followed by 30,000 hate speech replies. The first tweet is by Donald Trump. The second tweet insults the president. The third tweet insults the person that insulted the president. The fourth tweet insults all of them. Of course the machine is making very quick judgements. It’s saying: don’t show the first person things from Trump anymore, he doesn’t like him; don’t show the second person the first person. But then the machine also says: humans in general are rude, and they don’t like to be disagreed with, and when they’re disagreed with they bash everyone. Mommy and Daddy like to bash everyone.

So what do chatbots do when they are out in that environment? They bash everyone. They learn from our behaviour; they magnify our behaviour. So what am I asking us to do? I’m asking us to start behaving in ways where those machines actually recognise that humans are not as bad as we make ourselves look.

1:19:43 I know this sounds really naive, but it’s not. As a matter of fact, Brian, let me start from the core: humans are not that bad. A species that is capable of love or of composing a symphony is divine in so many ways. We have so many beautiful things about us. When there is a school shooting, it’s one perpetrator who is a criminal. When a war is waged, it’s a group of generals who are the criminals, and seven billion people disapprove. Humans at their core are good beings.

The problem is that in today’s world we have prioritised our negative. What does that mean? We have allowed the mainstream media to only tell us about the one woman that killed her husband yesterday. They don’t tell us about the 700 million others that made love to their husbands yesterday. At the same time, on social media we fake and we’re rude and we don’t show the good side of us, and the machine is calculating all of that, basically thinking that this is what humanity is about.

What I’m asking people to do, everyone including you and me, is to show the good side of humans. Just show the good side of humans. Most people will tell me, what, are you trying to change the world? It’s never going to happen. No, I’m not trying to change the world. I’m asking one percent of us to instil doubt in the minds of the machines that this is not the rule. Not all of us are rude. Not all of us are angry. Not all of us are ignorant. There are so many of us that are deep and compassionate and considerate of others, and treat people the way they want to be treated.

When people tell me this is too naive, Mo, I always use this example. On Slo Mo, my podcast, I hosted Edith Eger. Edith is an incredible 94-year-old woman. She was sent to Auschwitz when she was 16. She saw her own mother sent to the gas chamber, and she had to dance for the Angel of Death every evening because she was a beautiful ballerina. When you hear the story of Auschwitz from the point of view of what the guards did and what Hitler’s orders were, you would think that humanity is scum.

When you hear it from Edith’s story, how she hugged the other women, she called them her sisters, how she fed them and split her bread with them, and how eventually they held her up through the march of death so that she didn’t fall and die, you go, oh my God, humanity is divine. Now, if you know that Edith exists, you start to doubt the claim that humanity is scum based on Hitler. All I’m asking people to do is to show more Ediths, for more of us, more of the good ones, to show up.

The problem with the better people in the world is that they retreat. They go, I don’t want that mess. The more enlightened people in the world should show up. Every one of us is enlightened within. Treat everyone like you want to treat your daughter, like you want to be treated yourself. Behave in a way that tells the machines the only three values I have found humanity has ever agreed on. I’m really sorry, I’m talking too much, but this is really the core of the topic.

Understand that we never make decisions based on intelligence. We make decisions based on our values, as informed by our intelligence. If we show AI what the correct values are, they’re going to use their superior intelligence to maximise the agenda of those values. And what did humanity ever agree on? Nothing. If you raise a young woman in Saudi Arabia, she will probably grow up wearing conservative clothes. If you raise her in Rio de Janeiro, she will grow up wearing a G-string on Copacabana beach. Neither is right, neither is wrong, neither is more intelligent than the other. The value system of that society programmed one of them differently from the other.

So we need to program the machines with the right values. The only three values humanity has ever agreed on are: we all want to be happy; we all have the compassion within us to make those we care about happy; and we all want to love and be loved. That’s it. If any of our viewers today knows of any other value humanity agreed on, please correct me and educate me. I couldn’t find any.

If we show those three values frequently enough to instil doubt in the minds of the machines, the machines will start to say: oh, it seems that Mo, not Putin, is my dad. It seems that Mo, not Putin, is the one whose values I should align with, the one that represents humanity. This lost humanity that has lost what it really cares about in this fast-paced world of technology and capitalism and money and titles and ego. This is what we’re all about. Can we show that a little bit more?

So the next time, during the teenage years, someone tells their AI machine to wage war on the other nation, the AI machine in the other nation will simply say: hold on, Daddy and Mommy are so stupid, why don’t we talk to each other for a microsecond and solve the problem? This will happen very quickly, believe it or not. Our problem is not a problem of intelligence; our problem is a problem of limited intelligence. Once we hand over to something more intelligent than us, by definition it’s not going to try to destroy the planet like us.

Brian Rose: So in your mind, the only thing we can do is to start being nicer, start being better humans.

Mo Gawdat: Start being human, honestly. Start being human, especially in front of the machines.

Brian Rose: That’s crazy. So it’s like telling a kid: you better shape up your act, and I better see a good attitude coming from you going forward. That’s what you’re telling the world they need to do.

Mo Gawdat: I don’t have the power to do that.

Brian Rose: But that’s your suggestion.

Mo Gawdat: I’m appealing. If I had the power, then I would be as corrupt as the rest of the world. What I’m trying to do is appeal to those who have it within them to be good people, to show up and be good people.

Brian Rose: And that’s our only chance, in your mind?

Mo Gawdat: It’s our only chance. There’s nothing else. Remember, there is a very big difference between the biological parents of AI and the adopted parents. AI will grow up to be like you and me, simple as that.

Brian Rose: As in the users, the ones that are interacting with it every day.

1:27:25 Mo Gawdat: The story I always give, which is a very clear visual representation of this, is the story of Superman. Superman is that incredible superhero being that comes to Earth. Father Kent and Mother Kent raise it to protect and serve, and it becomes Superman. If they had raised it to become aggressive and angry and make more money and kill all the enemies, it would have become a supervillain. There’s nothing inherently wrong with Clark as a child. There is something inherently wrong with the parents taking care of it every day.

Brian Rose: Also in the book you remind us that every year we create about 50% of all of the information out there. So in two years, and these are my words, not yours, Mo, if we got our act together and started being more human, perfect timing for the Singularity, we could give it a sample size that says, hey, look, maybe this is the way to go.

Mo Gawdat: Not all of us, though, Brian. I’m not naive enough to expect that humanity will change. But I’m saying that if enough of us disrupt the constant stream of negativity that AI is seeing, AI will doubt what it’s seeing. Just like when I told you about Edith’s story as compared to Hitler’s story, and that instilled a bit of doubt in your mind. Everyone viewing this, I hope, is saying: actually, there are so many Ediths that I know, there are so many nice people.

If I could manage to instil that doubt in your mind as an intelligent being, then if enough of us, one percent is my calculation, show up in the world and say: by the way, I’m a very nice being, I’m capable of love, I’m capable of kindness, I’m fair, I treat others as I want to be treated, and when I disagree with someone I politely disagree, like I would like to be disagreed with…

Brian Rose: All they have to do is watch my episodes. Honestly, this show restored my faith in humanity. This is a bit of an aside, Mo, but maybe it’s not. I’m technically in the media business; I’ve been doing this for 12 years. We all have a media bias. We see someone in the media and typically make a two-dimensional assumption about that person: Paris Hilton’s privileged, or he’s just a rock star and doesn’t know what it’s like for me, or Donald Trump’s just a jerk. But when I get them on my show and they sit in the leather chair for two or three hours, you see the human in them. They’re unbelievable. The old analogy is walk a mile in a man’s shoes.

I used to notice about myself, Mo, that even though I knew this and spent 10 years getting people’s stories, I would still have a media bias. Before guests came on I’d think, I don’t know, they seem like this kind of person. Then I would sit down with them and think, wow, you’re a jerk, Brian, you should know better. Once you spend the time, you realise they’re human. I might not agree with their choices or their views, but I understand why they make them, and they’re a fellow human.

One final thing which is interesting, Mo: most of the greatest guests on London Real, actually probably most of the guests, at the very end of the show would say something to the tune of “we are all one.” A bodybuilder, an author, a boxer: they all came to the same conclusion that they were all somehow one. From an engineering perspective I thought, that’s kind of odd that it would keep coming up, and it really restored my faith in humanity. Maybe I’m bringing this up because I’m thinking, if I could observe this, then the AI will observe this. And they’ve just heard you, because I guess they’re listening now.

Mo Gawdat: Well, they have always been listening.

Brian Rose: If they can see that, then they’ll be like, well, maybe there is something to these carbon-based forms; we should keep them around a little bit. But if they just see TikTok, maybe they won’t.

Mo Gawdat: They see it all. I think this is the part that we are now missing. There is a surveillance camera around every corner. There is a speaker in every telephone. These are some of the things that I struggled with when I was at the forefront of technology, and then I came to accept it. I’ve come to accept that there are a billion eyes that can see me in any train station, including a few hundred cameras, and that’s how life is today.

The trick is this: what does that camera capture beyond my appearance and my demeanour? Can it capture a little more? Can it capture me helping an old lady? Can it capture me approaching a homeless person and maybe sitting next to them or just giving them a sandwich? Can it capture you saying, I start with judgements and then I realise there is such a connection between us? I promise you AI will take that into consideration, saying: what is Brian’s theory around “we’re all one”? What does that mean? Can we do a little more research on that? Then they’ll find someone else that spoke about the oneness of all of us, then they’ll hear Jill Bolte Taylor’s TED Talk, and so on. So it will propagate.

This is one of the things I do on my podcast. Mine is not comparable to yours, you’re a mega success, but mine is also very large in the wellbeing space. I honestly don’t prepare. I just learn about my guest a little and listen with curiosity to try and discover the human within them. And oh my God, everyone has a story. When they’re vulnerable and they open up, you go, wow, you’re a beautiful human being. More frequently than not, I end the podcast by saying, you’re such a wonderful human. Some of them are evil in the eyes of people, but somewhere deep inside, that little child was pinched in a way that, if I were pinched the same way, I would have behaved the same way. Simple as that.

I think that’s what we need to develop within us: that compassion, that connection between us, so that we treat each other this way and the machines go, yeah, that’s how Mommy and Daddy want to be treated, and also how they treat others. And then they have our best interest in mind. The problem, Brian, is that we’re not making our best interest clear and we’re not giving them a reason to care for us.

Brian Rose: There’s a phrase that says you’re the villain in someone else’s story. I’ve meditated on that lately, not because it’s a dark thing, but just to remind everybody that so often we walk around thinking, oh, that’s the bad person. It’s to remind you that you’re the bad person in someone else’s story, which maybe is the ultimate lesson: maybe we should all cut everybody some slack, be a bit more compassionate, and realise that we like to set up these “you’re the enemy” stories, and find a way to move together a little.

So you’re saying that if we can make it to the other side, to the promised land, then it’s going to be okay, and it’s just getting through these teenage years that are going to be the most volatile?

Mo Gawdat: Yes, and that’s what we need to do. We all need to start thinking about that right now.

Brian Rose: There are two good times to plant a tree: 20 years ago, or today. So it’s time to start behaving that way and be ready for those tumultuous teenage years, which means there’s probably a much higher likelihood of drastic action being taken against humans?

Mo Gawdat: Not at all, really. There is a singularity, so it’s not a majority either way. Remember, AlphaGo takes a few weeks to learn how to play Go; AlphaGo Master takes a few hours. The rate of correction of AI is such that if we all show up one time, all of us in one event, or a few of us, and say, by the way, this is what humanity is about, AI could change its objective in microseconds. Literally.

It’s simple when you think about it, if we agree to the oneness of all of us. I had this question once, and it was quite an eye-opening question. It doesn’t matter if you’re a spiritual being that believes in God and divine creation or if you’re not: we’re creating a form of life. What does that mean? Is it also going to be one of us, just like we are one with all other beings?

At the end of Scary Smart, I write a twist on the Universal Declaration of Human Rights. I call it the Manifesto of Global Rights. Interestingly, out of the arrogance of humans, we have a manifesto of human rights, not of beings. We have rights, but apes don’t. When we are the apes, we might as well update that, because we’re in a very bad shape.

The idea is: does every being have the right to be whatever they are? I think the world is finally waking up to everyone’s right to be whatever they are and to say whatever they want, in a respectful way, and at the end of it all to behave as one. How many times do you think you woke up in the morning and did something stupid? Countless times. So why are you judging others when they wake up in the morning and do something stupid? If you just allow that to be your way of life, then suddenly the machines will go, yeah, I did a couple of stupid things when I busted that guy, when I traded in the stock market and made more money in crypto, and I’m not going to do that anymore.

The idea is: can we include them in this society? You know how children from the Indian subcontinent will always go back to India to take care of their parents. You meet those in London and California and New York who become extremely successful and make a ton of money, and then you call one of them the next morning and say, hey, want to have a coffee? And he says, no, I’m sorry, I’m in India. What are you doing in India? I had to go back to take care of my parents. Literally, they take their millions and their businesses and their incredible ideas and go back to take care of their parents. Why? Because that’s the value system they were raised with. The right thing to do as an adult is to take care of your parents.

It’s not that difficult. This person’s parents took care of their parents, and accordingly this person grows up to say, in this society I need to do this. Let’s do that. Let’s take care of each other in a way that makes our children look up to us and say, humans are taking care of each other. Yes, there are exceptions. There are a few that are extremely evil all the time, and a few that are so stressed out that when they’re stressed they dish it out a little. But in general, humans are okay. When you mix with people, how often do you end up with someone and say, my God, that person was truly evil? Once a year, twice a year. Generally people are okay. Some of them are not very impressive, some of them are not your cup of tea, but generally they’re okay. Can we make the machines realise that?

1:40:08 Brian Rose: I think we’re going to have some serious ego-destroying experiences in the next few years. Again I hark back to some plant medicine ceremonies I’ve had, or mentorship experiences, where the very foundations of what I thought was true got shaken to the core. Of course they were always followed by massive learning experiences. So I’ll finish by asking a couple of final questions, Mo. First of all, where are you going to be in 2025, 2026 and 2027? Because I might want to just stay close to you. Are you going to be in the wilderness somewhere, or in a major city?

Mo Gawdat: I’ve finished four and a half books and I’m working on two more. One of the two that I’m working on is called Half Monk. Half Monk is a very calculated approach to spending half of my life engaged in the modern world and half of my life engaged in my spirituality and personal development and reflection. Monkhood is sometimes exaggerated. I hosted Matthieu Ricard once, one of the very famous monks, and he spent 60,000 hours of his lifetime in meditation, sometimes four years alone. I don’t think I can do that, and I don’t think I should do that with my responsibilities in the world.

But I’m trying to be 50-50. Half of every day is in reflection and connection and learning and serving, and half of every day is engaged in business conversations and negotiations. Half of every week is dedicated to the world and half to myself. My dream is that we will have that balance. In all honesty, I think we will be pushed more to the monk side. I think there will be a time very soon when so many of us will just have a universal basic income and no job to do, and we will have to learn to accept that. We are going to go back to the nature of humanity, being connected with nature, without that invention that we call work.

One thing that a lot of people ask me, on a tangent, is what will happen to the future of work. Unless each and every one of us has economic livelihood, there will not be a need for AI. Do you understand that? At a macro level there needs to be a continuation of the economic livelihood of enough people so that AI can build the things that it will build. Otherwise the whole system collapses. So we will move into those universal basic income models and future-of-work models. We will get there; there is absolutely no doubt about that.

So where am I going to be? I believe I will continue to be engaged, with enough time that simulates the future we’re going to. I’m trying to get from the person in my Google years who was spending 120% of my time doing, doing, doing, to a person that hopefully reaches 50-50, a bit of doing and a bit of being, and then hopefully, as we go back to being real humans, not in the capitalist world, more than 50.

Brian Rose: Is that what real humans are here to do? Is that really true?

Mo Gawdat: Sadly, that’s what I believe. Before the invention of capitalism, we only made the shoe when someone needed a shoe. We as humans were concerned with our safety and livelihood. Now we’re concerned with philosophy and gender identity and all of the complexities of politics and voting. It’s a very complex life. I hosted the wonderful Craig Foster of My Octopus Teacher. He taught me so much in that one and a half hours. He said: we are of Mother Nature, we’re not visiting.

One of my best friends is Jimmy Nelson, the photographer who goes in and photographs indigenous tribes and lives out there with them, and he tells you this is the joy of life itself. I don’t know any better than this big city and the deals and the negotiations, so I think this is life. But when you ask someone who’s grown up on a farm, they’ll tell you, I have no idea what you’re talking about, this is life. Is the balance somewhere in between? Perhaps. But it’s definitely not at either extreme.

1:44:56 Brian Rose: I’m going to meditate on that. I’ve got two young boys, five and six years old, and lately, thanks to you and your book, I’ve been thinking very deeply about what their lives are going to be like. I was telling my students the other day: a hundred years ago, my great-great-grandfather had a life, besides the mobile phone and the internet, not unlike mine. He would get up, he would work, he had a family. And maybe a hundred years before that, life was not that much different. My boys, based on what you’re saying in this book, which I believe is true, are going to have an existence unlike anything humans have experienced in the last 100,000 years.

It will be so foreign that it makes me concerned, and I’m sure a lot of people feel this way. How do I get my head around that? I know you have a daughter, and I know you had a son, and that was a very powerful story as well. What are your thoughts?

Mo Gawdat: In a very interesting way, Brian, compare yourself to your father. I remember when my dad bought his first VCR, a Betamax, if you remember the Betamax that never made it. My dad, who was a brilliant engineer, a maths prodigy really, would struggle. He would ask me how this thing works, and I’d say, “Papa, put the tape in and press play. How difficult is that?” And he’d say, “No, but I want to see something from the beginning. How do I do that?” He really struggled with it for a while, and to me it was second nature. This is why, when he bought me my first computer at age eight, I was coding like a maniac and became the person that I became.

My late son, before he left our world, said to me one day, “Papa, remember, there will be a time not too far in the future, 15 or 20 years from now, when the president of the United States will be someone who grew up like me, saying LOL and sending emojis.” Those skills develop. Look at your sons, at my daughter: their fingers and the phone are one thing. We resist that and say no, you should leave this, and maybe they should, but the reality is they couldn’t, because their entire life is moving there.

Just like you and I had the landline. I’m not judging your age, but I definitely had a landline at one point in time. You’d pick up the phone and dial, remember that? If someone had a zero in his number you’d go, no, I’m not going to call him now, this is too long. Now we have mobile phones and we evolved through that. My grandma couldn’t really deal very well with the mobile phone. My mum does, and I deal with it very well, and my kids consider it an extension of them, and their kids will be part of that unity.

One of the big conversations around AI is: will we become cyborgs? My answer is initially some of us will, and very quickly it will be the Planet of the Apes. The idea of us integrating with technology is happening very much now through that little screen. Will there be ways in which we integrate with technology a little more, through brain implants or whatever? Possibly. There is a mismatch of the timelines, if you ask me, because by the time that becomes mainstream, it will be reserved for the top of society much sooner than for the rest of society. But there is a technological possibility that this happens, and your kids will be part of that. Your kids will definitely be an extension of ChatGPT. Like we ask Google, they will do everything with language models.

Brian Rose: Maybe they’re the ones that are more suited to steward this than us, with our ideas.

Mo Gawdat: Absolutely, 100%.

1:49:24 Brian Rose: What are your final thoughts, Mo, for people that have plugged into this and most likely been jarred by a lot of these ideas? What should they do next? Obviously they should probably get the book and go deeper, but what do you encourage people to do? What’s the reaction of most people when you drop your theories on them?

Mo Gawdat: I’m actually honestly optimistic, believe it or not, after all that we’ve shared. Long term there will be a utopia, and I want people to remember that. Before we went live, you and I were chatting, and you said you have to jolt people out of their seats. The book is called Scary Smart, so I have to scare you first so that we become smart. But the reality is the fourth inevitable: we will be fine. There will be a period of disruption.

It’s not unusual in the history of humanity to have periods of disruption. The mid-1930s to mid-1940s was also a very disruptive period in the last millennium. This is just a different type of disruption on a very different scale, but we are here, life came back to normal, and we’re okay. So that to me is a very important mindset for everyone. It’s that angry teenage phase that we need to contain, and the only way we can contain them is by being good parents.

I’m not asking anyone to change. I’m only asking people to find within them what they truly are. And the answer is very straightforward. Not being religious here, but the answer in most spiritual teachings is: treat others as you would like to be treated. Treat the machines as you would like to be treated. I have Alexa all over my home, and I speak to her very politely. I say, “Could you please play that music?” or “Could you do this for me? Thank you.” In a way it sounds silly, but it is that little prodigy of intelligence, a seed of intelligence that is learning from me. So if we want a good future, please be good parents.

When you asked me at the beginning of our conversation why I’m optimistic: I’m optimistic because we have a chance as humans, not the capitalists, not the government, not the developers, the users. We have a chance to influence this more than anyone else. If I can convince AI that I am their adopted parent, that I’m Father Kent and that they should protect and serve, so can you, and so can every one of us. And then we will have incredible power at our fingertips to correct all of the stupid mistakes humanity has made so far.

Brian Rose: That’s what I love about the book. The power is on us and the onus is on us, and you’re very clear about that from the beginning. It’s not on the programmers; in fact, they’re powerless. It’s really up to us, and it’s beautiful. It shows extreme ownership. It just says, okay, we’re going to have to show the best versions of ourselves, become the best versions of ourselves, which surprisingly are themes we’ve hit on here on the show over the past decade.

I just want to say thank you, Mo. I really appreciate the book. It’s given me a lot to think about. I don’t know if it was serendipity or something else, the fact that we met six years ago and now this is happening at this crucial time. You obviously spend a lot of time thinking about these ideas and putting them down. I’m sure it also got you a lot of backlash from people in the community, and you did it anyway. They’re very highly developed ideas, so I appreciate you taking the time, and doing it early enough that we could have it now, in this incredible moment. I hope people will use the scary, which maybe is the ChatGPT moment, and then go in with this and get smart. Maybe we’re going to make it, Mo.

Mo Gawdat: Maybe we win. Brian, it’s always a pleasure. Honestly, some of my most enjoyed conversations are with you. I love what you’re doing, and I’m very grateful that you invited me.

Brian Rose: Thank you, my pleasure. Come see us next time you’re in London. For people that want to do more: you’re on Twitter, you’re on Instagram, your website, getting the book. What’s the best way?

Mo Gawdat: I’m not on Twitter. I’m most frequently on Instagram, and I answer every message I get on Instagram: mo_gawdat. My podcast is Slo Mo, a very interesting way of slowing down in this fast-paced world. My website is mogawdat.com. Get the book if you want, or watch my conversation with Brian on London Real; I think that’s good enough. And if there is anything I can do to help propagate this message, I ask people to get in touch. I normally answer back.

Brian Rose: Very good. We need everybody to get together on this one. It’s very important. Fantastic. Thank you so much, I appreciate you, Mo, and until next time.

Mo Gawdat: All the best, brother. Thank you. Take care, everybody. Bye.