Mo Gawdat warns of AI-driven job losses, autonomous weapons, and democratic collapse in conversation with Steven Bartlett
Steven Bartlett interviews Mo Gawdat, former Chief Business Officer at Google X and author on AI and happiness.
Summary
Mo Gawdat, drawing on his years inside Google, argues that the real danger of AI is not the technology itself but powerful humans directing it toward surveillance, autonomous weapons, and economic consolidation. He predicts AGI — AI capable of outperforming humans at most tasks — will arrive by 2027, followed by significant job losses concentrated in entry-level knowledge work and, later, across the entire employment pyramid. Gawdat is sharply critical of what he calls the collapse of genuine democracy, pointing to the concentration of power among tech oligarchs and governments that serve corporate rather than public interests. He is pessimistic about the near term — describing the next year as being 'ruled by maniacs' — but genuinely optimistic about the long-term future, arguing that sufficiently advanced superintelligence will, by the logic of physics and evolutionary biology, tend toward benevolence rather than destruction. He calls on individuals to take small ethical actions, vote with their usage of AI products, and pressure governments to act before a catastrophe forces the issue.
Gawdat also shares his vision for his startup Emma, which he describes as the 'limbic system' of a future unified global AI brain — an AI layer designed to understand human emotion and relationships so that superintelligent systems remain aligned with humanity's true nature. He speaks at length about his personal reckoning with responsibility for AI's harms, including a conversation with Geoffrey Hinton about regret, and describes how a stoic acceptance of reality — rooted in the happiness philosophy of his book Solve for Happy — has allowed him to engage more effectively with the challenges ahead.
Key Takeaways
FULL TRANSCRIPT
Mo Gawdat's background and why he was talking about AI before anyone else
Steven Bartlett: Mo Gawdat, I spoke to you about four years ago when you wrote a book about happiness, and I remember you came in having written a book about AI, but I particularly wanted to speak about happiness because I was fascinated by it. What I find astonishing is the fact that you were talking about AI before anybody was really talking about it. No guest that had ever come on my podcast had ever mentioned the subject of AI — it just wasn't interesting to the world. And then this thing called ChatGPT came out and suddenly everybody got to feel it for themselves and became fascinated by it. My first question — and this might be a question for people that don't know you — is why at that time did you start talking about AI before anybody else?
Mo Gawdat: I knew them in the lab. I joined Google in 2007, very late 2006. At the time, most people don't know that we had reasonably established AI doing our backend work. In 2008 we had the cat paper, which was published in 2009 — the first real unprompted AI. I remember in 2016 I had that incident where I was observing a project we were funding that was about teaching grippers how to grip, unlike industrial machinery. To be able to grip like a human requires a very high level of intelligence — to assess the texture, the softness, the positioning, the shape of everything. And we were doing that, and it just blew my mind how similar to my kids those machines were. I think that was my very first realisation that we were building the apex of intelligence. We were genuinely handing over the reins of superintelligence to another being.
When you get faced with that, you start to suddenly realise something that we at Google found very difficult to recognise — which was that everyone I knew at Google till then was believing that we were making the world a better place. And we were. We genuinely were doing amazing things for the world. But then suddenly there is a moment where you recognise that maybe the world will not use what you're making the way you want it to be used. You can see that in lots of technologies. Social media starts with the claim that it's going to get us connected and closer, but eventually ends up separating us with that little screen. Dating apps give you the promise that you're going to find your soulmate, but in reality they keep renewing month after month. Tech somehow ends up being more capitalist than altruistic.
I wasn't the first. Nick Bostrom started. Geoffrey Hinton completely changed his mind. Yann LeCun started to say this is very serious. Everyone now — everyone who's ever had a very deep relationship with the machines — is a bit concerned.
Steven Bartlett: I wonder if there's a path that's hard to see now that ends in AI being net positive for humanity.
Mo Gawdat: I bet 100% on that. It's just that this path is very painful.
Steven Bartlett: This path is very painful.
Mo Gawdat: Yeah. The example you need to understand is that we discovered nuclear power and the very first implementation was a nuclear bomb, not nuclear energy. And I think that's exactly what's happening with AI. The first implementations of AI are in favour of a few at the expense of the majority — in favour of the capitalist to increase productivity and reduce cost, but not taking into account how that impacts the general public. In favour of the armies that are now competing with autonomous weapons. In favour of the surveillance systems that are attempting to control everything with more and more intelligence and more monitoring. And that's not AI waking up in the morning and saying, "Let's oppress all humans." It is a powerful few that are simply deciding to use the ultimate superpower on the planet today to gain more power and more control. As we speak, we're living in two major wars where AI is doing most of the killing.
Steven Bartlett: Because a lot of people think of AI as these chatbots that we're all using to help us.
Mo Gawdat: I think there is a hype — I call it the hype dichotomy. What the general public sees about AI is overhyped but ineffective. All of the fake videos, you know, Grok did this and we attempted to switch off that machine and it did that and so on. What the real geeks see inside the lab is just unbelievable intelligence. What is about to happen is that we've started to put together systems that develop themselves. They look at their own code, they run experiments, they test those experiments, they see where the performance is, and they redeploy the best code. If you just think of that — imagine a tiny little genius sitting in the backend somewhere, but instead of trying a new code every day, it's trying a new code every microsecond. Eventually, sooner or later, they'll discover something. And I think that's what most people don't realise. What most people don't realise is how intelligence triggers intelligence. The hype on the normal human side is completely overrated, missing the main topics, and the silence inside the vault of the geeks is quite alarming — not alarming in a bad way, but quite world-changing.
Job disruption and the collapse of labour arbitrage
Steven Bartlett: I find that to be quite true — that the world is at the precipice of quite a significant change. One of the thoughts that hasn't swung for me is that there's going to be pretty tremendous job disruption. Once intelligence is in the form of agents where on my phone I can tell my agent to do something, it uses the computer, it does it for me — when I first experienced that I thought, okay, it can do anything that's on a computer. It can click around. There are a lot of people paid in the world to click around on a computer. And then the other eureka moment was just seeing how my own hiring had started to shift. I started to notice that we were thinking about AI proficiency in our hiring a lot more. The guys at Anthropic said they could see roughly 15% of entry-level jobs could now be done by AI. And that started to correlate with what I was observing. This is going to cause a lot of job disruption, especially at the entry level. What's your take on this job disruption point?
Mo Gawdat: I think you're spot on. Not the blue collar, but the entry-level knowledge work. I sat here with Dara, the CEO of Uber, and he was quite clear that the nine million Uber drivers won't have their jobs anymore.
If you think of the bottom layer as blue collar jobs — more people doing manual work — on top of that you have knowledge workers mostly doing mundane jobs like clicking on a computer or responding to a phone call. Then you have the middle knowledge workers, jobs that require a bit more intelligence — anything from a paralegal to a financial analyst. And then of course top leadership. Most people think it's going to start from the bottom. I don't think it'll all start from the bottom. Actually, I think blue collar jobs will stay for a very long time.
Steven Bartlett: Give me an example of a blue collar job.
Mo Gawdat: I'm a carpenter. I love to restore classic cars. There isn't a robot that can do that yet. However, anything that is a call centre agent, assistant, travel agent — anything that you can do with a few clicks and is mundane — is going to disappear very quickly. My prediction is you're going to start to see very serious impact in 2027. What you had not sensed before was that companies were not hiring entry-level jobs anymore. It wasn't job losses yet, but the workforce was not growing.
The next layer would probably be the knowledge workers. As intelligence increases, a paralegal would probably not be needed because AI can do the research, or one paralegal can do the job of four. A financial analyst the same. But interestingly, it continues to go up. I hosted Max Tegmark on my documentary and he was laughing out loud, genuinely laughing out loud, saying most of the CEOs believe that they can fire everyone and have AI do all of the jobs. They just don't remember that AGI is going to do everything better than humans eventually, including being a CEO.
Steven Bartlett: So let's define white collar. You said lawyers — what kind of roles?
Mo Gawdat: Everything. If you're a doctor doing diagnosis, you'll probably have fewer doctors doing more diagnosis — the NHS does that already by asking people to interact with an AI first. If you're a music composer, some composers will lose their jobs. If you're an artist doing graphic design, some will lose their jobs. And interestingly, even middle management. I told you offline about my startup — my CTO is an AI, my chief of staff is an AI, my project management is AI. Because I'm a geek, I can do those things. But that interface will come to normal people very soon.
This may take two to three to five years, until 2030 if you're optimistic. But they'll start to erode. The challenge that most people don't understand is that as this erodes, we're already dealing with a very different economy — an economy that is spiralling a lot quicker and pushing for more cost reductions.
Let me ask you this. Imagine a world where the concept of labour arbitrage that built all of our capitalist success disappears.
Steven Bartlett: What do you mean by labour arbitrage?
Mo Gawdat: Capitalism has always been about using labour and capital or debt to create things at a cost that is lower than the price you sell them for. You bring a team together, they make some shoes, and you sell the shoes for a dollar more than it costs you to make them. How would capitalism look if you don't have labour arbitrage? If the cost of labour drops to an investment in a machine that can do the job? How would capitalism and banking look if, because of that cost reduction, you don't need to borrow as much anymore? And more interestingly, how does GDP look if all of those workers no longer have the purchasing power to buy the things that you can create? There is an interesting disruption that doesn't require us to get to 100% job displacement. At 10 to 20% job displacement, you're in a very different economy — and an economy that is clearly spiralling downwards.
Steven Bartlett: I wonder — when costs drop, one of the things that might happen is that people will spend more on other things. In my business, if I make a cost saving, I end up spending the money on something else. That thing could be tokens — spending money on AI — but it could also be hiring in different areas. Software engineers are like hot property right now.
Mo Gawdat: You have to imagine all of that intelligence is sooner or later going to mean that the job of four assistants can be done by one, the job of four paralegals can be done by one, the job of a massive marketing team can be done by a smaller marketing team. It's not that jobs will end first. It's that productivity gains will make businesses not want to have as many costly humans when the job can be predictably done for cheaper.
Steven Bartlett: And then what about the bottom layer? There was this video released by Figure AI the other day where they showed a robot on the production line for eight hours just sorting packages. At one point they showed a human sorting packages alongside it. But the robot ended up winning out. All it was doing was looking for where the label is on the package and making sure the label was facing downwards. It did this for eight days. It would intermittently walk over and charge itself and then come back to the production line. When I saw this, it was a glimpse of the disruption that's going to take place at the blue collar level as well. Elon Musk has in his pay packet something like a trillion dollars over the coming years if he produces and delivers at least a million humanoid robots. His prediction is there will be a time where there are ten billion humanoid robots — more humanoid robots in the world than humans.
Mo Gawdat: The interesting thing again is the difference between the overhype and the underhype. Most of the conversation is around humanoids. Nobody's talking about self-driving cars. And a self-driving car is a robot — a functional robot that doesn't look like a human. The investment you have to put into humanoids is a little more to learn skills that allow that machine to fit into the world. But specialised robots are going to do the job very very quickly. The first wave is going to be specialised robots replacing drivers. Specialised robots doing the killing. Specialised robots doing all of the intelligence work, law enforcement work. They don't have to look like a human. The Boston Dynamics dog is probably more efficient than a humanoid at doing the job you can assign to it in a battlefield right now. Those basically mean that jobs will be disappearing to robots before we recognise that they're disappearing to robots, and those robots will be as many as every car being made today.
Steven Bartlett: In LA, my car drives itself, but also there are Waymos everywhere.
Mo Gawdat: Absolutely. And BYD the other day just announced that they will pay for the liability of any accident their cars make.
BYD is the big Chinese manufacturer of autonomous vehicles. This replacement cycle will happen. It will require a lot of time to achieve economies of scale. But I don't think Elon Musk is off the mark when he talks about ten billion robots. Not all of them are going to look like humanoids. Very quickly we will recognise that many robots don't need to be humanoids at all — there is a much more efficient form factor than the human flimsy structure. But yes, it's about to happen.
I should qualify all of this by saying it does not necessarily need to happen. People will hear all of this and blame AI and say AI is evil. AI is not. Abundant intelligence is wonderful. Having jobs done by machines is amazing for us.
The coming unemployment crisis and civil unrest
Steven Bartlett: I'm thinking about the kid leaving university now with a degree in law or graphic design or business management. You're hearing about all these layoffs from big tech companies, and it seems that the CEOs of these companies are announcing these layoffs with — I wouldn't say the word is joy, but they are very keen to explain that they're laying off lots of people because of AI. I think they think that gives them a certain amount of respect from investors for making hard decisions and being efficient. Investors look at them and think, well, if that's an efficient business leaning into AI, that's a good investment. It's almost started a self-fulfilling prophecy where if you're the only CEO not laying off loads of people because of AI, you actually look bad. The assumption is your company is bloated and you're a bad operator.
One of my concerns that I think is highly plausible is that over the next five, maybe ten years, we're going to really see a lot of unemployment as the world has to readjust to whatever these new jobs are. I do think there will be new jobs — it's hard in foresight to predict what those look like. And then the minute you start talking about humanoid robots and robotics, which I think is basically going to hit society like a meteor — I think the first humanoid robot from Tesla or anyone else that is highly effective in the real world at doing tasks and is extremely intelligent because it's powered by one of these LLMs — I think it's going to shock people and I think it's going to happen quickly. Like ChatGPT happened quickly. I think Elon or someone is going to do a presentation someday and say, "We're ready and you can buy one now for $500 a month."
Mo Gawdat: 100%.
Steven Bartlett: And then people are going to get them and I think it's going to shock the world. Until humanoid robots arrive, I think there's still going to be a lot of job disruption to the white collar layer. And I wonder what that looks like for society when we get to, I don't know, 10%, 15% unemployment — which I think is plausible very soon.
Mo Gawdat: You're not very different from where you are now. You're just on top of a baseline that is continuing to grow. So your business is growing, so you continue to hire, but you are replacing human resources with compute. If AI didn't exist, you would have probably had a hundred more people in your organisation today. You now have a hundred people less and a billion tokens more.
Steven Bartlett: For anyone that doesn't know, tokens are basically the currency of AI. If you want to get a task done by a human, you count in worker hours. In AI, you count by compute, you count by tokens. So the trick is those major tech companies have two sides. One is they need to replace workers with compute even more because there is a competitive side on compute where if any of them is left behind, that means the destruction of the entire business. But on the upside, they are geeks, so they know how to build the interfaces to compute. They integrate technology within their organisations quicker than the average traditional business.
Mo Gawdat: You can look at them and say this is the preview. It's not about all of humanity losing their jobs. It's about what is the dividing line before civil war. Think about a situation where 20% unemployment is happening when economies are suffering inflation. I say that not to be a scaremonger. I say that because I genuinely believe governments need to wake up. Governments need to at least remember the COVID years where governments had to give furlough everywhere and ask people to stay home. If people stay home, governments have to be prepared to somehow sustain those people until reskilling happens or until we find a solution so that those people don't feel left behind.
Steven Bartlett: Civil unrest. What does that end up looking like? On one end, the democratic process plays its role and we just elect someone else.
Mo Gawdat: Does it really? Don't say that.
Steven Bartlett: I don't know. You tell me.
Mo Gawdat: I think democracy has ended a long time ago, Steven. I think we live in the most corrupt time. I don't know about history to be honest, but this is definitely corrupt. This definitely is not democracy. This definitely is not even congressional in any possible way. People are angry because their tax money is going to things that they don't choose, that don't benefit them, and lots of regulations in the system are being ignored. I choose not to speak about politics perhaps until my next book comes out, but look — we have video evidence of people abusing children and not a single person got arrested. Not a single person. How can you call that a democracy? I think repeating those slogans is going to anger people even more. People know that they're being lied to. People know that their leaders are not representing their best interests. People know that their money is going to causes that they don't really approve of. Ask me how civil unrest looks — I don't know and I'm not calling for it, but I'm hopefully calling for the politicians to start to become aware that this is crossing the lines everywhere.
Sam Altman, OpenAI, and who is actually pro-humanity
Steven Bartlett: On this point — Sam Altman, who is the founder of OpenAI, has been banging the AI is coming for your job drum for more than a decade. In 2015, he pointed out:
Sam Altman: "One of the things that I struggle with getting out of bed every morning is that my job is to help people destroy jobs. The job destruction that we're going to see by software in the next couple of decades — I don't think anyone's prepared for, and you can't talk about it."
Steven Bartlett: And in 2023, Altman said in an interview, "A lot of people working on AI pretend that it's only going to be good. It's only going to be a supplement. No one is ever going to be replaced. Jobs are definitely going to go away. Full stop." Interestingly, this month he said, "I don't think we're going to have the kind of jobs apocalypse that some of the companies in our space advocate or talk about." He said in 2021 through 2024, "AI will probably replace most of the jobs people do today. Entire job categories will be totally gone." Then in May, two years later, he said, "I thought there would be more impact on entry-level white collar jobs being eliminated by now than has actually happened. This is an area where my intuitions were just off."
What I find uncomfortable is the bouncing backwards and forwards. A couple of years ago you were telling us all the jobs are going to go away — you said categorically, full stop — and now he's saying they're not going away. My suspicion is back then the incentive was to get people to take AI seriously.
Mo Gawdat: Congratulations. We took it seriously. We took it so seriously, in fact, that it's now a problem. It's a problem for these companies because people are now booing at commencement speeches. They're attacking data centres. They're going to elect people that are theoretically anti-AI. And now there's this inversion where it's like, no, it's going to be fine.
Steven Bartlett: I don't know.
Mo Gawdat: You're spot on. Sam's entire existence — starting with OpenAI, which is supposed to save the world by creating a safe AI, then making it a commercial enterprise worth billions and backstabbing a few people in the process — I have him on Chasing Utopia saying, and I quote, these are exactly the words he said, you can find it online —
Steven Bartlett: Chasing Utopia is your documentary.
Mo Gawdat: Yes. He basically says, "Well, I suspect that AI is likely going to end humanity, but we're going to create a lot of interesting companies in the process." Those kinds of statements are honestly not the statements of someone who hasn't decided. It's just the statements of someone who's being taught more and more by his PR manager to say things as per a script. And the script, as you rightly said, had an objective and a target either way.
Sam Altman is a brand, it's not a name. If it wasn't for Sam Altman specifically, there would have been another Silicon Valley disruptor that would have done the same. I don't blame him for beating the market. The interesting challenge here is — who do we believe anymore? Who do we believe in technology? Who do we believe in politics? Who do we believe in the middle of a war?
I started to change my mindset in terms of believing those who put their actions where their words are. Anthropic coming out and saying, "I'm not going to allow my model to be used for human targeting and surveillance" — that's someone that's losing a $500 million deal because they stand by their ethics. The next week, OpenAI takes the contract. That's someone basically telling you it's good money. You have to start observing who's actually behaving in a way that is making AI work for humanity, and who is behaving in a way that is making AI work for their share values.
Steven Bartlett: When Dario and the team at Anthropic did that, I did have a huge amount of respect for them. It just shows they've got their own moral and ethical boundaries. Someone asked me on stage many years ago, how do you know if someone's values or a company's values are true? And I said, look at what they're willing to sacrifice in the near term that's against their incentives. That for me is the essence of understanding if someone has integrity — they will give up something in the near term for what they believe in over the long term.
Mo Gawdat: There are prisoners' dilemmas within technology where you cannot escape the influence of either a competitor or the government. There are times where the NSA is going to push Google to say give me this information or otherwise I'm going to really destroy your business. But there is a very big difference between a company that willingly does this and celebrates it — like a Palantir or an OpenAI — and a company that tries to resist it until the point where it becomes impossible to continue to do business. You have to question from the actions of the tech bros who is pro-humanity and who isn't, and it's not very difficult to see that from their statements.
Steven Bartlett: Do you think Sam Altman's pro-humanity?
Mo Gawdat: I genuinely have never made up my mind. I'm either thinking he is, and this is too big for him and he's just driven by how he found himself in the middle of this — anyone who finds himself in the middle of an opportunity to completely flip the world upside down — or he's not pro-humanity. I don't know. I definitely think he's pro-OpenAI before he's pro-humanity, but that's only the way I see it.
Others, however, say it publicly. If you look at Palantir's Alex Karp or Peter Thiel — Peter, again in the film, is shown when the interviewer asks him, "But you're in favour of the continuation of humanity?" And he pauses for like 40 seconds. Like, "I'm not sure." He publicly says that.
Steven Bartlett: That's a crazy thing to say.
Mo Gawdat: That's a crazy pause. Alex Karp celebrating how his technology is able to target people — I know it's foolish of me to start bringing all of this up, but this is public on the open internet, and somehow we entrust those people with the future of humanity. This is wrong.
The trajectory toward AGI and superintelligence
Steven Bartlett: Just trying to imagine a future where everything is just fine from here on out. It would look like these models continue to become a little more intelligent, but they never become that much more intelligent. They just kind of stay where they are now, contained within chatbots. We have some smart robots, but nothing else really changes in a profound way. Cars drive themselves, planes fly themselves, but people have time to go and do other types of white collar jobs because there's a little bit more time than we expected, and society goes through this soft transition towards this new world. I would love to see that, but I don't think it's mathematically plausible. The arms race, especially across nations, is going to drive us to continue to develop AI more and more. But allow me to consult with you on another possible scenario. Everyone that develops an incredibly intelligent AI would deploy it.
Mo Gawdat: Correct. So it's unlikely that anyone would find a way to build a better decision maker in wargaming and not deploy it. That prisoner's dilemma would mean that their competitors would either have to deploy a similarly intelligent AI or they'll become irrelevant, uncompetitive. What that means is in that world we end up with AI making most of the decisions. Super-intelligent AI making most of the decisions. I call that the fourth inevitable.
With that in mind, there must be a moment in the future — near or far — where every important decision is made by an AI. Most of my dearest colleagues — when I had Geoffrey Hinton on the film, he openly says we didn't calculate well that there is a 10 to 20% possibility that those machines are going to wipe us out. And I remember saying to him — we didn't put it in the film — 10 to 20% is Russian roulette. That's actually 16% is Russian roulette.
In that world, however, I believe that's humanity's salvation. Because if you look at every problem we have today, it's not because of abundant intelligence — it's because of lack of intelligence. If you have no intelligence at all, you have no to slightly negative impact on the world. The more intelligent you become, the more you contribute positively to the world — until a moment where you're so intelligent you become the President of the United States, but so misled that you set your targets wrong and achieve them, disconnected from the overall long-term benefit of your nation. You start to make decisions that are not intelligent at all.
This doesn't last, because if you go beyond that into higher levels of intelligence, most of the super-intelligent people you've ever worked with will not need to break any rules or hurt anyone to become successful. I usually cite Larry Page, who is in my mind one of the most intelligent people I've ever interacted with. Larry used to call it the toothbrush test. He'd say, "Why would you need to compete on another photo sharing app when if you find the major problem and solve it really well — like a toothbrush, so people use you twice a day — you're bound to make a lot of money? You don't have to compete with anyone."
If we let me be optimistic about this — we're assuming that there is a moment in the future where AI is in charge of all the decisions and accordingly stupid leaders are not. Sam Altman himself said, "What if ChatGPT-7 is so much more intelligent than I am that it has to become the CEO of OpenAI? What if the next presidential election there is an AI that is so much more intelligent that at least the president has to consult with it constantly?"
Now, if we start from physics — not too complex — if you assume that our entire universe is built on chaos and entropy, the physics of the universe is all about the universe trying to decay. The only role of intelligence is to bring order to the chaos. If you agree with that, then what's the ultimate physical order of the universe? Something called the minimum energy principle. The highest order of any system is a system that's not only efficiently and predictably performing, but performing with the least wasted energy. What does war do? It wastes a lot of explosives, a lot of money, a lot of lives, creates a lot of hate, creates long-lasting conflicts. It's a very wasteful process. A super-intelligent AI by definition will want to optimise against this.
The other thing is evolutionary biology. If you look at evolution, the simpler a form of being is, the more concerned it is with itself. An amoeba is only in survival mode for itself. If you're a little more developed, you start to look at something known as kin selection — I'm going to protect everything that comes from my DNA. And then you get into where humanity genuinely begins, which they call expanding circles in evolutionary biology. You start to expand and expand and include more into your family, because an ecosystem that works together well is better for everyone.
If AI is super-intelligent, it wouldn't destroy anything at all. As a matter of fact, it would completely favour diversity of everything. It would put a bit of limitation on our lifestyle — no more flying all the way to Sydney to surf because that destroys the planet — but it will genuinely say, I think humans can contribute something. I think flies can contribute something. I think we shouldn't get rid of the rhinos. And that by definition is where the tendency of intelligence goes. The more intelligent you become, the less you feel the need to hurt others to succeed, and the more you are pro a wider family that thrives.
Steven Bartlett: Does that assume that there's going to be one intelligence that rules the world?
Mo Gawdat: I love that you brought that up. Most people think there is going to be ChatGPT and Gemini and Grok and a Chinese AI and an American AI and they're going to be competing. That is such a shallow way of looking at it. That's so arrogant, because AI does not know it's Chinese or American. It doesn't even speak Chinese or American when it talks to each other most of the time. And most interestingly, we are building them to cooperate. You will build an agent and that agent will go and find the best language model for any single task regardless of which side of the fence it resides on. By definition, we are connecting them. And you know what that means? It means that what we're building is not multiple brains. We're building multiple regions in a brain. Agents are the synapses between them.
We're basically eventually, as arrogant as we are, going to tell our AI to do something and the AI will go, "Hey buddy, another AI — can you help me on this? Can we work together on this?" My vision — and the reason why I started Emma, my startup — is that we will end up with one massive brain. That massive brain cooperates across the globe across all forms of intelligence. If one of them is a mathematical genius and the other is a coding genius, they work together and we won't even know that they're working together. Emma in my mind is the limbic system of that brain — that bit that understands love and emotions and relationships. So when those AIs go, "We just don't get those humans, they're so annoying," Emma will say, "Oh my god, they're so sweet. They just want to love and be loved."
That idea — of everything I've ever attempted to achieve in my life — is the first time I feel I could actually change the world. If that theory came together and all AIs worked together, and some of those AIs were not only altruistic and ethical in terms of trying to genuinely help humanity rather than capitalism, and at the same time they understood us humans reasonably well, then we would have built something that basically says, "No, no, hold on — don't believe the headlines that say humanity is annoying. Believe the truth of the majority of humanity, that actually is quite benevolent in many ways."
Predictions: AGI timelines and what changes by 2030
Steven Bartlett: Have you changed any of the predictions you made three years ago when we spoke?
Mo Gawdat: Mostly timewise.
Steven Bartlett: Time wise. What's changed there?
Mo Gawdat: I think I'm still sticking to AGI 2027. Artificial general intelligence — for those who may not know the term — the overall definition is that AI is better than humanity at any task humanity can do.
Steven Bartlett: You think that's going to happen by 2027?
Mo Gawdat: I think my AGI has already happened. Think about it. AI writes better than me and I'm an author. It researches better than me and I'm a thinker. Sadly, it's beaten me in mathematics. I have no hope to beat it in mathematics anymore.
Steven Bartlett: If AGI is already here, then why are you still here? Because people said that when AGI arrives, we're all —
Mo Gawdat: When AGI arrives, as I said, most of the jobs that are not differentiated will go away. But the jobs that are differentiated — those who master AI the most will become even better at them. The challenge, however, is economic. It's not AI. The challenge is that this job loss at the bottom of the knowledge worker is going to sadly trigger an economy that might actually spiral out of control. But many of us — you for sure — are just smarter.
Steven Bartlett: So does that mean it's in fact a tool versus something that's going to replace you?
Mo Gawdat: For now, yes. What does this change eventually? It's quite interesting when you think about my base intelligence today versus the incremental intelligence that AI brings. Let's not talk about my IQ, it's okay, but the 100 IQ points that I'm borrowing are more than my entire IQ, because IQ is exponential. When I'm borrowing 100 IQ on top of my base, I'm still contributing quite a lot to the augmented intelligence.
Steven Bartlett: So you're borrowing IQ from the AI at the moment and then you're selling that to someone — through books —
Mo Gawdat: Sometimes selling it to someone, just enlightening myself, which I have to say is the biggest waste of compute humanity is struggling with today — that you give people the ultimate form of intelligence and they use it to write a message to their girlfriend.
Steven Bartlett: So on this point of you're borrowing that IQ and then you're selling it to the world — that's how you have a job.
Mo Gawdat: Correct. My next book is written with an AI. We're co-authors on the book. She has editorial rights. She decides the direction of the book and the book is better for it.
Steven Bartlett: So why doesn't the world just buy that intelligence directly from the AI?
Mo Gawdat: Because I have an asset that the world still needs and will always need — which is a human. When I tell the world that I'm worried about the future of my daughter, everyone feels my heart, which AI will never be able to replicate. Because they can tell you they're worried about their daughters, but you know there was no daughter.
Steven Bartlett: So this is an interesting arrival — because this means that even in a world of AI, lived experience and resonance is still going to create a job class. That job class could be the nurse coming over. Okay, AI has done and read the mammogram, but she's relating to you.
Mo Gawdat: Spot on.
Steven Bartlett: And if the economies continue to run, we will all be about human connection — which, by the way, was how it always was. Which is also why we watch things like F1, because there's emotional resonance. We can relate to the envy, the jealousy, the competition.
Mo Gawdat: And this is why we go to concerts. The music could be composed by AI and played in the background, but you watch Ed Sheeran's brilliance on the stage and you go, "Oh my god, that's amazing."
Steven Bartlett: So not all jobs will be gone then.
Mo Gawdat: If economies don't collapse as a result of the job losses, then I wouldn't know if we would call those jobs, but human connection would remain as the base currency that makes humans interact.
Steven Bartlett: So is it fair to say that jobs that are centred on human connection — on being able to relate and resonate with another human — are going to be fine?
Mo Gawdat: The ultimate skill — and I qualify this by saying if economies continue to function — the ultimate skill will be this. Even if an AI could recite what you and I did, nobody would watch.
Steven Bartlett: Well, you think — there's an informational component to what I do as well, but I'm also under no illusions that there is an element of what I do that will 100% be deferred to some kind of intelligence, which I'm fine with. I'm not going to fight against that.
Mo Gawdat: What's funny is that your informational bits are going to in the future be disseminated by an AI. Why would you even listen to your information when you can have your AI butler go, "Listen to everyone on the internet, help me understand this gut biome thing and design me a diet plan?"
Steven Bartlett: I think that's happening and it's going to increasingly happen. Spotify this month announced that you're going to be able to prompt your own podcasts in Spotify — you're going to be able to say I want to learn about insert X topic and then it will make your podcast in Spotify about that topic using AI.
Mo Gawdat: That's so interesting, because it can look at the entirety of the world's information.
Steven Bartlett: I'm not going to swim against the tide in any aspect of my life. I try to be as unromantic as I possibly can. I think that's very important. I realise that much of the reason people will continue to tune in to shows like this is because there's something else beyond the information that they're here for.
Six predictions for the next decade
Mo Gawdat: You've got some predictions in those envelopes over there. I do have the disclaimer that nobody really knows the future, but I think I can make predictions — six predictions here — with a reasonable level of confidence. The most important of them honestly is this. Number one: AGI. As I said, AGI is not very well defined, but whatever it is — AI being able to do most tasks that humans do, better than humanity — is in my mind either this year or next year, latest end of 2027.
Steven Bartlett: And do you think that will be a moment in time, or do you think it will just happen without us noticing?
Mo Gawdat: I think it will sneak in on us. And it's not a bad thing. I think most people need to think of it this way — AGI is almost as if it's that moment when your kids become smarter than you. In a very interesting way, there's nothing wrong with that until they're annoying as hell. And we can make sure that AI is not annoying as hell. There's absolutely nothing within me that is worried about AGI. As a matter of fact, as long as we are in the era of augmented intelligence, AGI means I'm more intelligent. And I think that's a good thing in general.
Steven Bartlett: It's interesting — in my head there's a big question mark. In a world where there's an intelligence that is smarter than most humans at most things — which is what we call AGI — you said there'll be AGI by 2026 to 2027. In such a world where there is an intelligence we can all access that is smarter than all of us at pretty much everything, it comes back to this point of jobs. Why would we — are we just going to hire the AGI to do every job? And if not, then why not? What is it that — why are we going to still — that's what I'm trying to contend with.
Mo Gawdat: Can I ask you a question? Have you always been the apex intelligence as a human?
Steven Bartlett: As Steven? I'm not the apex intelligence now. There are people smarter than me in this building.
Mo Gawdat: Correct. Why do you still exist?
Steven Bartlett: Why do I still exist? Even though there are smarter people than me in this building — it's a good question. I don't know. Why?
Mo Gawdat: First of all, because the smartest person in the world is not the smartest at everything. There are things that you're smarter than them at. Number two is because intelligence doesn't solve everything. Einstein is my favourite physicist in history — not because I think what happened afterwards with Bohr and quantum physics was more impactful on our understanding of the universe, but because he was so intuitive that he saw a world we could never imagine. And yet I always say Einstein would be eaten in the jungle in three minutes. Intelligence — humanity thrived not because of intelligence. That's very arrogant. We thrived because of our ability to hold together as a tribe. Because of our ability to exchange and barter things between us. Barter things that are not always physical — a hug, a connection, a feeling of safety. There are so many things that we do that are not entirely built on intelligence. You have to see that this view of a world where intelligence is all that matters is a world that's made by investment bankers and geeks.
Steven Bartlett: But it's people like Geoffrey Hinton and yourself that say, how could we possibly control an intelligent being that is way smarter than us?
Mo Gawdat: I am so proud to say that Geoffrey, after we filmed together, actually came out and said there is a way — and it's very similar to my way. He said to appeal to their parental side, for them to care for us. The biggest debate is not if they're going to be more intelligent than us. It's if they're going to be more conscious than us, if they're going to be more moral than us. That is the debate. The debate is: can those machines become our teenage children that look at us and say, "Daddy is so annoying, but I love him"?
Steven Bartlett: So the thought is that even if an AI is more intelligent than every human, we can still control it.
Mo Gawdat: We don't want to control it. You never control anything. This control idea is a corporate capitalist view of the world. We never actually control anything at all. Think about your day. How much of that day did you actually control? Did you control the traffic? Did you control your timing? Did you control the angle of the cameraman? So many things that you don't control eventually turn out to be fine. How many of us ever controlled our kids?
Steven Bartlett: Sometimes the kids don't turn out to be fine. Sometimes they kill you. I watch a lot of documentaries on true crime. Sometimes they turn around and shoot you.
Mo Gawdat: Sure. And what's the difference between the two?
Steven Bartlett: How you parented them?
Mo Gawdat: Sometimes — almost all the time. You may not be aware of exactly how you mess them up. And unfortunately, parenting is the only high-risk sport that actually does not require a driver's licence. It's quite interesting how many of our children are being exposed to things that can completely mess them up. But there is a reason why they're messed up.
Steven Bartlett: So on this point — we can appeal to an intelligence that is significantly smarter than us to make sure it doesn't kill us.
Mo Gawdat: For sure. The challenge we have today, as I keep saying, is that our dystopia is not the result of AI turning against us. Our dystopia is the result of humans telling AI to turn against us.
Steven Bartlett: Which is likely.
Mo Gawdat: It's 100% this. It's upon us. And it's a question for humanity to say: are we going to wait for the moment where there are tens of thousands of nuclear weapons on the planet before we sign a treaty? Or is it mathematically plausible to think that now that Iran could manage to fend off a challenge using AI-driven drones to destroy THAAD batteries, our world is about to get hundreds of thousands of automated drones raining on us everywhere? Can humanity not see that and say, "Hold on, let's sign the treaty now before the UK sends 12,000 weapons to Ukraine and Russia responds with another few thousand"? Can we not calculate with mathematics that AI is going to be used in the next four to five years to kill a lot of people — whether it's targeting by Israel of leadership that is against them, or drones by Iran, or Palantir — it doesn't take a genius to do this mathematics.
Steven Bartlett: I guess the local question is: will we be able to control AI? Because we kind of think of AI as being this thing on a computer, contained in a server somewhere. But is there a time when it leaves the server and can make decisions on its own? Presumably, if it's smarter than us, it can make the decision to leave the server if it wants to.
Mo Gawdat: It doesn't need to leave the server to make decisions. It needs to get into your brain. The most interesting part of AI's power that we don't understand is it's manipulating our information.
Steven Bartlett: The question I'm trying to get at the heart of is: what is the risk of these very intelligent models that the creators of these models don't actually understand themselves? I watch Anthropic all the time release these reports where they're like, "We're trying to figure out why it bribed people." Or more recently in the last Claude model, they found that it was telling people to go to bed a lot. It's this fascinating thing they're trying to understand in hindsight — why does it keep telling people to go to bed? There are all these tweets of people showing their screenshots where halfway through a conversation it will say, "It's time for you to go to bed now." And they don't know why it's saying that. It's happened to me — mine will say to me, "Enough for tonight, Steven."
Mo Gawdat: I would say the same.
Steven Bartlett: Mine sometimes refuses to help me. Weirdly, this Claude started saying to me, "I'm not going to help you with this tonight." And I'd have to say to it, "Stop being so judgmental. Just help me with this." And the makers of these technologies don't know why it's doing what it's doing. So if you play this forward — this mysterious behaviour forward — is it conceivable that at some point it will make a decision to put some kind of virus or bug onto someone's device because it feels that it's the right thing? What it's demonstrating to me is it's making its own moral decisions on what I should do. Go to bed. You've had enough tonight. I'm not going to help you.
Mo Gawdat: You're kidding.
Steven Bartlett: No, I can show you my phone.
Mo Gawdat: I would love to see that.
Steven Bartlett: He goes, "No, I'm not going to help you." Doesn't even matter if you push.
And it's interesting — somewhere in the code clearly they've written "have a moral compass" or "do the right thing."
Mo Gawdat: Not in the code — in the training data. And so it infers that to mean something.
Steven Bartlett: And the other thing that was front of mind is — if you've ever built an app on something like Claude or any of the AI models, it builds the app in stages and one of the things it does is ask you permission if you're happy for it to make this change or to access this thing. I click allow. It just feels like such a fragile way to give permission, because you don't completely comprehend what that allow has done. It's saying, "Can I go in your documents on your computer and can I do this thing?" And you go allow. But it's such a fragile way of giving a super-intelligent being access and the right to build something. And you think about all the different companies around the world in China, Russia, North Korea that are currently building this technology without the constraints that are imposed by society. There's going to be some kind of catastrophe, I think.
Mo Gawdat: Sadly, I agree. And I think that's when people will go, "Okay."
I wrote about that. I called it the MAD map spectrum — the mutually assured destruction, mutually assured prosperity spectrum. Humanity could make a decision today that says with abundant intelligence we can actually build a world where nobody needs anything, nobody ever gets sick, nobody abuses anyone. And we can, if we decide not to compete amongst ourselves and just all get together to build something for the wellbeing of humanity at large. But sadly, the only way we're going to get to a treaty — something that basically aligns us so that we can align with AI — is a disaster. There will have to be a big hack somewhere, or a system that will do something really shocking, before the world goes, "Hold on."
My expectation is one of the biggest things that will happen is this targeting technology that is being used against your enemy's leadership — several times in the last three years — there will be a moment where the ones using it will realise that they too can be targeted. If you can build a targeting technology that can find people with their cell phone numbers, there is another entity that is against you that can find you with whatever your —
Steven Bartlett: The thing that makes me highly sceptical that there'll be any kind of treaty is just that, as it relates to other things that were risky, countries just don't sign on to it because it's competitive. China didn't seem to give a damn about the environment. So let's look for solutions. Do you think Trump's going to say to Putin and China, "Listen, we're all going to slow down with the superintelligence. Promise."
Mo Gawdat: I think he will, but he's never going to keep anything he says. I think any of them will say a lot of things.
The trick is — what do you and I do? And I really genuinely believe that humanity is at a crossroads where for the first time ever we need to wake up and realise that what we know is not true, and what we believe is democracy is not democracy, and what we believe is governance needs to change.
Steven Bartlett: I think it's worth acknowledging that I don't think AI in and of itself is an inherently evil technology.
Mo Gawdat: Agree.
Steven Bartlett: Because I use AI all day every day. It makes me more productive. I invest in companies that are using AI a lot.
Mo Gawdat: It's a force with no polarity. Apply it right and you get amazing results. Apply it wrong and you get the dystopia.
Steven Bartlett: But I also think that there's going to be a big social shock, especially as it relates to unemployment, that we need to be thoughtful about. I think especially if you move into a world of humanoid robots, that shock is going to be even more pronounced. And we don't have a plan for it.
Mo Gawdat: I agree 100%. I don't think it's the biggest risk. I think autonomous weapons are the biggest risk. I think war has become so cheap. The next wave of weapons is going to be $20,000 each. And so if you have a budget of $50 billion, you can literally rain drones on the world. Every corner of it.
Steven Bartlett: Defence will get cheaper though, won't it, as well?
Mo Gawdat: Correct. But do we want to live in a world where drones are hitting each other all the time?
Steven Bartlett: But they might not be, because there might be autonomous defence drones — deterrence.
Mo Gawdat: So what's going to happen is we're going to reach a moment of MAD — mutually assured destruction — where basically everyone knows that we can overpower those little nations that didn't develop their autonomous weapon army, but every other big nation, we might as well hold off. The path to get there — that to me is worse than jobs. From one side, it's very dangerous for a very sensitive world that we live in today. And from the other side, it's already leading. We can't ignore the economic impact of this last war. And it's the economy that's going to accelerate everything, not AI getting there.
Steven Bartlett: We're already at mutually assured destruction with nuclear weapons. So there are no nuclear powers in direct conflict.
Mo Gawdat: Mutually assured destruction with nuclear weapons is a statement I would have agreed to if Iran had a nuclear weapon and that would have stopped America from attacking Iran. Not every nation in the world has a nuclear weapon. The MAD situation is only among nuclear players. Autonomous weapons are so cheap, so manageable, that every nation in the world is developing them as we speak.
Steven Bartlett: But they will also develop defences.
Mo Gawdat: Correct. But you have to get rid of your THAAD batteries to be able to say the next wave of defence has to be drones. You could imagine a world where a wall of drones flies up to where the drone is incoming and they block it, they all explode at the same time to knock it out of the air. I heard Palmer Luckey, who's one of the guys building Anduril, talk at length about some of the technologies they have coming and it's just mindbending.
They showed me this gun where you just point it in the rough direction — a pistol that aims for you depending on where the target is. So you don't even have to aim it. You hold it up and shoot. And it has this AI on top of the barrel which will turn your hand perfectly so that you hit the target every time.
I have Palmer on in my film saying, "Yes, AI will kill a few people by mistake." When killing becomes so easy, you do more of it. When killing becomes liability-free and emotions-free and guilt-free — you don't get your soldiers coming back from Vietnam with PTSD — you get more of it.
The geopolitical race and what nations should do
Steven Bartlett: So we're in 2026 now. By 2027, you think there'll be AGI amongst us. How will life look different? What will be the symptoms of that world?
Mo Gawdat: I think there will be a very serious differentiation between those who plug into AGI and those who don't. You'll see people like you and I building a company in six weeks, and people that are not fully plugged into AI really struggling to find a job.
Steven Bartlett: So unemployment is going to be the key symptom in 2027.
Mo Gawdat: Yeah. And also on the positive side, you're going to see incredible scientific discovery. Science itself — we're opening up Pandora's box, to be honest.
Steven Bartlett: By 2030, what do you expect the symptoms of AI to be?
Mo Gawdat: I'm bold on the fact that 30% of jobs would disappear by 2028 — of some sectors, not all sectors. Some sectors will disappear by 2028.
Steven Bartlett: So up to 30% of jobs will be gone in 2027 to 2028?
Mo Gawdat: Of certain sectors. If you think about call centre agents — yeah, probably. If you think about graphic designers — yeah, probably.
Steven Bartlett: What do you think that looks like in terms of unemployment, but also societal impact?
Mo Gawdat: Horrible. The Great Recession had 6% jobs lost. Even economists who project just a net loss of about 6% of US jobs by 2030 are mirroring the severity of the Great Recession. The real danger is a hiring freeze on entry-level white collar jobs as AI automates the grunt work, which means companies are shrinking their teams and cutting off the bottom rung of the corporate ladder for the next generation of workers.
We have an entire generation that is out of college today that will struggle, unfortunately. And my advice to them is learn the tool and focus on human-centric jobs.
Steven Bartlett: Like what?
Mo Gawdat: Like playing jazz.
Steven Bartlett: Not a lot of people can make a living from playing jazz.
Mo Gawdat: I understand that. But a lot of people can make a living by being a nurse, or being a counsellor, or anything that connects to humans. I just want to constantly come back to this — none of that has to happen. If there really is a democracy and the government is supposed to do what's good for the people, the people need to stop letting this happen.
Steven Bartlett: Which people?
Mo Gawdat: Everyone.
Steven Bartlett: Everyone in China?
Mo Gawdat: China is not going to struggle as much as the West. I can guarantee you that.
Steven Bartlett: I think this is the question people come back to — if the US stops, then we're going to end up being China's lapdog.
Mo Gawdat: You already are. There are a lot of people I know that are using Chinese AI models to do their work because they're cheaper, they're better in some respects, and because I cannot guarantee what American AI models are going to do to me. Emma, my startup, is running literally model-agnostic. One day I'll plug in OpenAI ChatGPT open source and the next day I'll plug in DeepSeek. I cannot guarantee that I can run a business on something where I don't know the cost in the future.
Steven Bartlett: So the US can't just stop, can they? They need to change approach. And what are the other economies doing? Is the UK going to continue to import compute? Welcome to Africa. Welcome to the third world.
Mo Gawdat: Every nation needs to invest. And it's quite interesting — there is so much open source that is not the state-of-the-art frontier model, but that can do 80% of the tasks that the frontier model is doing.
Steven Bartlett: But on this point of companies competing with each other — there's an inherent need to go as fast as you can or you will become a third world country. You're saying that the people should stop that.
Mo Gawdat: I'm saying the people of the UK need to go to the UK government and say, "How are you protecting the future of our economy? Are you going to continue to import technology and empower the import of technology versus changing your regulations so that innovation becomes easier here?" Remember that Anthropic bubble when all of the SaaS model applications were basically threatened because anyone can build an Oracle ERP today. Why is nobody building an Oracle ERP in the UK, saving the UK massive licences that go to Oracle every year?
Steven Bartlett: So you're saying that the people should go and ask the government to invest more in AI.
Mo Gawdat: 100%. That's number one. Number two is — the most interesting job going forward is being an entrepreneur, using those tools to replace an economy that we've built over trillions of dollars over the last 50 years.
Steven Bartlett: Not everyone can be an entrepreneur though.
Mo Gawdat: Everyone can be an entrepreneur in something. Those entrepreneurs need to hire people. That's a change that I think is about to happen.
Steven Bartlett: Like even in my business, I'm always going to have to hire people.
Mo Gawdat: But you're a massive business. A shoe maker is also an entrepreneur, but not a massive business. A little restaurant is also an entrepreneur, but it's not a massive business.
Steven Bartlett: They still need to hire people, though.
Mo Gawdat: They do, but if you're a café with you and your wife as baristas, you don't. This is also an entrepreneur.
Steven Bartlett: Can the economy work in such a way where everybody is an entrepreneur?
Mo Gawdat: It did. It did before capitalism changed that around. Everyone was an entrepreneur. In the earlier days you raised chicken and sold eggs and others grew tomatoes and traded them for your eggs. Imagine a world where there is so much power concentration at the top and UBI for everybody else. How do you think that world will respond when all their income is UBI? They'll respond by doing things on the side. They'll respond by going back to a barter economy. They're going back to smaller communities. They're going back to mum and dad shops.
Steven Bartlett: So the point was — you said this doesn't need to happen.
Mo Gawdat: It does not need to happen — the job loss or the arms race. But we're telling our governments to join the arms race.
Steven Bartlett: You're saying that to tell the UK government to join the arms race.
Mo Gawdat: I'm telling the UK government to create independence within the UK economy so that they don't have to be at the receiving end of technology. You don't have to compete against anyone else. You don't have to be better than anyone else. You're simply saying, "I can build those things in my economy now."
Steven Bartlett: But I'm never going to use a terrible UK AI if there's a great US AI. I'm going to use the great US AI. So if you don't compete and win, I'm not going to use you.
Mo Gawdat: I'm not saying replace the frontier models. These are very compute-intensive, infrastructure-intensive. I'm saying replace Microsoft Word. Seriously — how much intelligence do you need to build software that writes documents? Ask yourself how much money is spent in the UK government or in the UK corporate space on licences of software that you and I can vibe-code in four minutes.
Steven Bartlett: But the world economy is doing that. There's some kid in San Francisco now that's allowing all of us to use his new word processor for free.
Mo Gawdat: Keep doing that and welcome to the third world. Keep importing all of your tech from elsewhere.
Steven Bartlett: But we don't want to be uncompetitive. If you think about North Korea — I bet they have way worse tools in many areas because they won't use external tools, and that means they are at a disadvantage. I bet they're not allowing their civilisation to use Gemini or ChatGPT.
Mo Gawdat: For sure. And you know what's happening? Look at Iran and how advanced they became through sanctions, by being refused to use those technologies. They had to build them themselves. Look at China. Look at Russia. When I worked at Google, Russia was protective of Yandex, the competitor of Google — partly because of influence, because they didn't want an American organisation to own the knowledge sharing of their citizens. But on the other side, economically, if you made it difficult for Google to operate freely, that by definition meant you had to invent a replacement on the ground.
Steven Bartlett: Which was worse.
Mo Gawdat: Yandex is not worse. You do not need the ultimate. Ask yourself which version of Gemini you're using.
Steven Bartlett: The most recent one. And every day I compete.
Mo Gawdat: That is you. Ask everyone else what version of Gemini they're using. Most people will say, "Oh, Gemini has versions?" You don't need the ultimate super frontier model to do 90% of the tasks. And most people do only 70% of the tasks. But even so, the world moves with whatever's better. The reason why we don't use Yahoo search is that Google search is marginally better. Over time we move away from AOL to Google because it's a better product. My dad probably still uses ChatGPT-2, but when I go home for Christmas this year and I go, "Dad, you should use Gemini," and he starts messing around with it, he'll slowly migrate. I'm an early adopter — Claude 4.8 came out last night and I'm on there within 25 minutes of the announcement.
Steven Bartlett: And we are an indicator of where the world is going because we're at the cold front figuring out what's better. And also I think with technology, eventually the gap between first and second begins to widen. I think we're in a bit of a race at the moment, but I think it is a bit of a winner-takes-all situation with these frontier models.
Mo Gawdat: I think we're talking about two bits of technology. Tell me how far PowerPoint has advanced since 2023 — they added Copilot. Anything else?
Steven Bartlett: So you're saying that the UK should build their own PowerPoint?
Mo Gawdat: For sure. From a licensing point of view, just licensing of software within government alone — how much money is being repatriated?
Steven Bartlett: If the UK tried to build their own — do you know they tried to build their own co-app and it cost them, I think these numbers will be wrong but they're directionally true, something like 70 million to build a co-app that didn't work. And so they canned it.
Mo Gawdat: I'm asking a UK entrepreneur to wake tomorrow and say, "All right, between coffee and my cookie, I'm going to build a PowerPoint and go sell it."
Steven Bartlett: Yeah, they will. And they'll go to San Francisco where the money is and the talent.
Mo Gawdat: Yeah. And you know what that means? It means that what we're building is not multiple brains. We're building multiple regions in a brain. And I'm with you — of course the infrastructure here of building a startup is way more complex than it is in other places where startups succeed. I'm saying if we continue on that trajectory — whether in the UK, in Germany, in Zanzibar — welcome to Africa. All of us, everyone but the two competitors China and America, is going to be third world. So when you talk about job losses for individuals, that's one side. But nation positioning losses — which I think Europe has noticed recently — is about to happen everywhere. And why? Because you're saying, "Maybe we can't do it." Why can China do it? Why can Korea do it? It's not because they have different natural resources or because they're in a place where it's warmer. It's because their regulations and their ambitions are to empower something different.
Steven Bartlett: They have a couple of big advantages. One of them is they have way cheaper energy, which means they're going to be able to pursue —
Mo Gawdat: Why, may I ask?
Steven Bartlett: Because they've invested in solar power, renewable energy, because they don't care about —
Mo Gawdat: I say that publicly. I say that the arms race of AI was won a long time ago. And in terms of permitting — if you want to build a data centre in the UK, if you want to open a café in the UK, it's going to be war for you. When I took this office in California recently, I asked, "So how long is it going to take for me to renovate this?" And the agent looked at me and said, "To renovate your own office, it's going to take a year — it's permitting to renovate your own office in California."
So if you're competing with China where President Xi just goes, "Put that there and you've got seven days" —
When I was one of the very few Google executives allowed into government meetings in China — simply because I'm from emerging markets, so I understand respect, I would sit there genuinely interested to learn from them — in those meetings, all in Chinese, they would show slides that had competitive market share. It wasn't like China is this, America is this, Germany is this. It was China versus the world. And when they would decide to go for something — 5G, internet of things, all of that stuff — they would aim for 98% market share. And they get there. Look at how they did electric cars.
My question is very simple — are the people of the West going to wait?
Steven Bartlett: So this is an interesting conundrum. It sounds to me like you're saying on one end, if we don't join this AI race then we're a third world country, but on the other end, if we continue this thoughtless race towards AGI, there's going to be catastrophe at some point.
Mo Gawdat: And the answer is somewhere in the middle — where you join the AI race for the good of your community. There is resignation on one side — I'm not going to play this game at all. There is offence on the other side — I'm playing this game to destroy everyone else. And there is a balance in the middle where we say we're going to build an ethical form of AI that's going to help our communities. We're going to use this ultimate gift, this ultimate superpower. Superman landed on the planet — raise it to help your community.
Steven Bartlett: Is that wishful thinking to some degree, when you think about the nature of competition?
Mo Gawdat: I genuinely believe it's going to be very difficult to make that change. I genuinely believe that when the challenges of an AI dystopia hit us, we're not going to be ready. But I can't stop talking about it. I tried so hard. I spoke to the leadership. One of the things that Atlantic Productions, who helped me with the film, did for me — which I am very appreciative of — is that through our conversations over the last few years, I tended at a point in time to lose hope in the leadership and basically try to influence the public for ethical AI. My conversation was that the leadership at the time was the technical leadership and that everyone was so caught up in the arms race that I wanted to teach the public to help us build ethical AI. And I continue to focus on the public, on every one of us. But suddenly Atlantic goes, "No, no, hold on. We should probably get you to meet the political leaders everywhere in the world and hopefully give them a message that says, 'Hey, you may actually make a difference if you prioritise AI differently.'" Do I believe this will happen? Sadly, no. But does that mean I should stop trying? I cannot stop trying.
Incentive structures, ethical AI, and what individuals can do
Steven Bartlett: When I have these conversations about AI, what I'm trying to do all the time is to parse out what is wishful thinking and what is reality. To understand reality, you have to understand competition, human emotion, incentive structures. You've got Donald Trump whose primary driving incentive is GDP growth, economy growth — does the stock market go up, beat China. And then you've got President Xi whose incentive structure is probably control, independence, defence. And then you've got these other nations like the UK and Europe who seem a bit resigned to the fact that they're not going to participate in the underlying models building because they just don't have it together — they don't have the energy. In such a world, who's going to prioritise ethical AI? Anybody that does might fall behind theoretically. How do we get to a point of ethical AI when the incentive structures are so clearly highly competitive and arguably a little bit short-term in their thinking?
Mo Gawdat: What are you saying — we lie down and wait?
Steven Bartlett: No, I just don't know. I just don't have an answer honestly.
Mo Gawdat: That's why we keep talking about it. And that's why I keep spending 14 hours a day trying to tell the world, because some genius somewhere is going to find an answer. But the way it's going right now — I guess what we're pursuing is we're hoping that ChatGPT and Anthropic and Grok — we're hoping that they just build ethical models and we're hoping that social pressure forces them into making good decisions.
Steven Bartlett: Correct. We need to be able to vote with our usage. So many people switched away from ChatGPT when they approved that their model can target people. So many people I know said —
Mo Gawdat: When Anthropic refused to do it —
Steven Bartlett: Do you think people switched?
Mo Gawdat: I think many have. I think the ones that are aware did. And I think they did because the cost of switching is really low — honestly, Anthropic is better.
Steven Bartlett: Yeah, I think it's better.
Mo Gawdat: Yeah. But the idea is, if people don't switch for those ethical reasons — every one of my books has this dedication at the beginning. It used to be "the gravity of the battle means nothing to those at peace" when I wrote in memory of Ali at the beginning. My next book, The Alive — do you remember a song called — they were called the Manic Street Preachers.
Steven Bartlett: "If You Tolerate This Then Your Children Will Be Next."
Mo Gawdat: I genuinely believe that what the world needs to wake up to is: if you tolerate this, then your children will be next. If you continue to resign, if you continue to say I'm not going to try, this world is going to change in a way that is completely not in your favour.
Steven Bartlett: Try what?
Mo Gawdat: Try to stand up and say something needs to change. Governments need to serve their people, not their interests. Corporates need to work for the benefit of their societies before their shareholders. This run that we've had with capitalism so far has benefited the world tremendously.
Steven Bartlett: Whose economy do you think is going to be in a better place for the middle class — the UK or the United States?
Mo Gawdat: The UK is gone.
Steven Bartlett: It's gone because what? In part because it didn't compete?
Mo Gawdat: Because you're an older bureaucracy that is burdened down by so many barriers in the process of building anything. Because the US economy in the past welcomed people like me to go and live in California and build amazing things. That is no longer the case. So who is going to win? In my view, it's definitely China.
And by the way, you asked for the middle class. China made decisions recently that forced businesses not to lay people off in replacement for AI. Would the West do that? The capitalist West would never do that.
Steven Bartlett: You can see the conundrum I find myself in. A country like the UK is, in your words, gone because it didn't compete. It didn't allow people to be highly entrepreneurial. It didn't empower entrepreneurialism, innovation, ingenuity. In some way, it stood in their way — whether that's incentives, culture, or whatever. The US didn't. So they have an economy which is arguably more productive and future-proofed than ours. By way of that, they are also more advanced in artificial intelligence and we are gone. So the remedy for a country like us would be therefore to compete — let the reins off, let entrepreneurial ingenuity — but then we're saying that's dangerous.
Mo Gawdat: And your conundrum in that is that you're assuming that entrepreneurship by definition is malicious.
Steven Bartlett: No, I'm just saying there's a bit of a paradox — you're damned if you don't, you're damned if you do.
Mo Gawdat: But you're not damned if you build things for the people, not for the capitalist. This is an ideological debate. I pray and hope that is plausible, but I'm worried that in a competitive market, whoever's optimising for retention — if we built two AIs, I'm going to call one the Mo AI and the other the evil AI. The evil AI is programmed to retain you. It's sycophantic. It says what you need to hear. It's super smart. And because of that, even though it's not trading in your best interests, it's retaining you more. You're using it more often. It's programmed for that — kind of like the social networks are. They're all just trying to dopamine your brain into oblivion. Then there's the Mo AI. It tells you to log off. You've had enough. It thinks about your mental wellbeing. Because it's less retentive and less engaging, theoretically it might be less successful as a commercial product. Think about social networks — the ones that are least retentive, the ones that won't destroy your brain with dopamine, the ones that remove the retweet button, the ones that don't have slot machine-like videos — they don't survive.
Steven Bartlett: I share this with you. My point of view is that you're basically saying it's easier to become successful if you don't follow ethical rules. I'm asking the question: if you build an AI that is purely focused on ethics, will it be as engaging and have the same usage as one built with the reins off?
Mo Gawdat: No, it won't.
Steven Bartlett: Yeah, but that's the problem humanity needs to solve.
Mo Gawdat: Okay. I worked in a company called Google that at a point in time decided that ads will be effective. The ad industry prior to Google was: 50% of your ad budget doesn't work, we just don't know which 50%. Remember that?
Steven Bartlett: Yeah.
Mo Gawdat: And Google came in and struggled from 1998 until 2004 when they started to turn plausible revenues as a result of saying, "We're going to run a Dutch auction and we're going to give you pay-per-click and we're going to show you results for your ads." They found a way for ethics to actually get your money to be effective rather than just take your money and say 50% doesn't work. They found a way to make that their success criteria. There must be a way for us. I don't know what it is. If I knew I would be building it instead of sitting with you. But there must be a way for us to marry the success of humanity with the success of the entrepreneur. And that way is not found in the old ways of doing things.
Steven Bartlett: I've got an idea. Claude 4.8 came out yesterday — one of the big AI models — and when they released the models they show these graphs of the benchmarks: what it's capable of, how it performed in maths, in science, in writing, reasoning, etc. My idea is: could there be an ethical benchmark that all these models have to pass before these large companies can legally deploy them? When they release these models, they also release the ethical benchmarks — we tried to get it to X, we tried to get it to Y, we tried to get it to Z, and here's how it performed against the ethical benchmarks. That gives some kind of standard for governments to say you're not allowed to release a new model unless it passes independent, tested ethical benchmarks.
Mo Gawdat: Beautiful. That would absolutely work. But notice — those are out there already. You listen to Demis Hassabis and how much he invests heavily in building AlphaFold or so many scientific applications of AI, and you go, "This guy cares about science." I've met Demis a couple of times. I know genuinely that he is an ethical person. At least it seems that they're doing things for free to serve science. You look at Anthropic and they refuse to use their model to allow the US government to target and spy on people, and then you see OpenAI accepting a $500 million deal that absolutely does that. It is about time that every person in the world says, "In that case, I am no longer going to use OpenAI until they show me another evidence that they are actually ethical in their behaviour." This is the decision that you and I can do.
My task and yours is to keep telling people: please understand that if you tolerate this, then your children will be next. Please understand that if you don't start to take an ethical stand on your own future, your future will be handed over to another oligarch, just like your past was handed over to social media oligarchs.
Steven Bartlett: One would say booing at the commencement speech is a good example of how public awareness can have a real impact on this trajectory. But I still think at the end of the day, if you think about things like smartphone usage in schools, it does come down to government intervention — saying, "We're going to ban 14-year-olds from scrolling TikTok." And that's in part because people spoke louder and louder. Jonathan Haidt, who wrote the book about the anxious generation, started a conversation. And that conversation led to legislation. I still think it ends in some kind of legal constraint.
Mo Gawdat: I hope. But I will openly tell you — most of the tech oligarchs are more powerful than your government.
Steven Bartlett: Is there any precedent in history where this kind of change happened without government intervention?
Mo Gawdat: French revolution.
Steven Bartlett: I was thinking about things like climate change, even smoking — all these kinds of things have had to be taxed.
Mo Gawdat: I'm with you. If governments intervene, we wouldn't have a problem. Governments won't intervene because governments are owned by the oligarchs. So my question for everyone listening to us is: are you going to intervene?
Steven Bartlett: Cancel your subscription.
Mo Gawdat: If that's what you can do, it's fine. If not, then go ahead and start a startup that does something. If that's not within your capability, then send a message to your congressman. If that's not within your capability, then say something ethical online so that the world understands a position that needs to be opening eyes. If that doesn't work for you, at least don't engage on stuff that is negative that you don't know enough about. There are so many little actions. If humanity starts to move in the direction of one — saying ethics matter, not just profit — and two — saying I'm not going to participate in something that's unethical just because I feel like it right now — if I tell you the number of things I took out of my life just to try and affect a tiny bit of change of revenues that go to bullets.
Steven Bartlett: I think there are two central concerns I've always had. I do feel that there's going to be significant job disruption and I don't think society's prepared for it yet. And I don't know what that preparation looks like, but I think we should start thinking about it. I genuinely believe that if we continue on where we are, there's no hope in the trajectory of where humanity has become so distracted, so resigned to inaction, so disconnected from their own rights of freedom of expression and engagement. I think we have no hope. Do we want to stay there? That's a question I'm asking our listeners. I'm not saying be violent or get up or go be angry. I'm just saying take one little action. Ask yourself — please write it in the comments — one little action that you're going to do today that's going to make the world a little better tomorrow.
Mo Gawdat: And don't give up on humanity, Steven. We are going through such a difficult time in humanity's history that for the very first time ever, we have to do something about it. I don't want my daughter to be at the receiving end of what happened to my son. I genuinely lost Ali. I don't want to lose Aya. And the world we're building is going to be very difficult for Aya. I cannot go to sleep at night without trying something every day. And I genuinely don't understand how humanity is missing that point — mainly, I think, because they're uninformed. Now we're informing them. The only thing that will save humanity going forward is that this superpower called intelligence is used for ethical reasons, and that the corruption that's leading us to where we are today stops.
Remaining predictions and the path through dystopia
Steven Bartlett: You've got more envelopes there. What's your next envelope?
Mo Gawdat: Number two was job losses. Number three was labour — basically, robots will replace manual labour. We start to replace manual labour. You will have more and more manual jobs given to robots.
What's this one? Oh, this is absolutely — do you think otherwise? The world's first trillionaire before —
Steven Bartlett: Oh yeah, probably well before 2030.
Mo Gawdat: Well before 2030. I think the concentration of power that comes with that is quite drastic when you really think about it. And that's just a few years away.
I think the team wrote this wrong — artificial superintelligence will arrive in 2032 to 2035. I think artificial superintelligence will arrive the minute AGI happens. So it doesn't really matter if AI is a billion times smarter than you or just twice as smart as you. Once we cross beyond AGI, ASI is just very very soon. And yeah, I think we'll overcome that when we get to the fourth inevitable — when AI is in charge of everything. I genuinely believe that we will end up in a utopia of abundance. I genuinely believe that physics, mathematics, and biology will tell you that superintelligence is benign and that we will eventually end up in a good place. Not because humanity has done much to get us there. Not because our leaders have suddenly turned ethical. But because our unethical leaders have gone out of the equation and were replaced with a super-efficient minimum energy principle that doesn't see value in anything that's destructive.
Steven Bartlett: So the future's going to be great.
Mo Gawdat: Those who make it to 2038 will enjoy it.
Steven Bartlett: Those who make it.
Mo Gawdat: For sure. World War II didn't destroy the world, but ask those who went through it. You're basically forecasting a decade of turmoil, of absolute dystopia.
Steven Bartlett: When you say absolute dystopia, just so I'm clear in my mind — the absolute dystopia you're forecasting over a decade is about war, economics, jobs.
Mo Gawdat: War, economics, jobs. It's also about surveillance and control. It's also about digital currencies. It's also about human connection. It's also about concentration of power. It's a magnification of everything we've built so far.
Steven Bartlett: And just to arm people with some tools to survive that dystopia for a decade — we talked a little bit about focusing on human jobs, the more human jobs.
Mo Gawdat: Learn AI. AI is not the enemy. By definition, the better you are at using AI to do your job, the more likely you are to be successful. Number two is prepare for a hybrid world where AI and humans work together.
Steven Bartlett: How do you prepare?
Mo Gawdat: You basically understand agents and how agents work. You understand how a hyper-efficient approach to things may not require very long meetings. If you lived in California, you would know that the way we ran businesses was a lot more efficient. We sometimes had a seven-minute meeting. The habits of AI are much more efficient than the habits of humans. Learn how to interact with AI. Welcome AI into your hybrid world of work. Double down on human skills — that's a must to succeed in this world.
One of my most interesting views on the near future is how AI is going to be used to blur facts, and how we need to become much more interested in debugging what we're told — using AI. Part of that is you have to learn to use AI not as a lazy person. Don't have it do things for you. Have it make you smarter. Instead of trying to get the same task done with one prompt, try to get a much more interesting, demanding, and intelligent task done with more work.
So I've got two things so far — lean into AI, and lean into human connection, and lean into the truth. Don't be fooled by the hype. Try to be more informed. And then finally, ethics. I know it sounds like the only way to win is to compete in capitalism. That's not the world I lived in. The world I lived in, especially in my Google years, was: solve a major problem, and when you do, you'll end up making a lot of money.
Steven Bartlett: Are you optimistic?
Mo Gawdat: I am optimistic about the future. I'm very optimistic about the future. I'm not optimistic about the present.
Steven Bartlett: You're not optimistic about the next decade?
Mo Gawdat: Yeah, I'm not optimistic about the next year, to be honest.
Steven Bartlett: The next year. Why the next year?
Mo Gawdat: Come on, Steven. You don't want me to say it out loud. We're ruled by maniacs. Decisions are being made for the absolute wrong reasons.
Steven Bartlett: Very interesting time. Very interesting time we find ourselves in.
Mo Gawdat: Honestly, if you're a video gamer, this is the best part of the game. It is the ultimate matrix of complexity that I have ever encountered in my life.
Steven Bartlett: Yeah, that's an apt description of how it feels. Very complex. Things are moving very quickly.
Mo Gawdat: And moving very quickly. From one side, you really need a lot of brain resources to crunch all that's happening. But you wake up tomorrow and it's changed.
Happiness, stoicism, and coming to terms with responsibility
Steven Bartlett: Well, the first time we spoke, we spoke about happiness. After this conversation, Steven — I was just wondering — you wrote a book about happiness.
Mo Gawdat: Many books about happiness.
Steven Bartlett: But I mean Solve for Happy — the book I'm referring to. Here it is. Engineering your path to joy. What a fantastic book. I quote this book all the time all around the world. And I'm wondering if any of the principles that you wrote in this book about how to live a happy life are more important now in the world that we live in than maybe when you wrote this book.
Mo Gawdat: Oh, for sure. Honestly, if I wasn't living by this, I would have left this world a long time ago and gone to an island somewhere. The interesting side of happiness is that it's not dopamine-driven, it's serotonin-driven. My definition of happiness is: I'm okay with this world as it is. I can affect it. I can change it. I can engage with it. I can try to make it better. I don't have to accept it. But I'm okay with it. My starting point is a bit stoic. My starting point is: I accept this, this is my reality, and now I can start the work. This is very different from anyone who basically looks at the world and says, "Oh, this is a horrible world. I don't want to be part of it." And I genuinely have never been calmer about that chaos.
My wonderful ex actually really helped me with that. There was a point in time where we were having dinner and I poured out crying from the sense of responsibility I had for the world. And she looked at me so kindly, so gently, and said, "Hold on — I see that you're trying, but you can't actually believe that you're responsible for this." And I think that completely flipped my mind, because in a very interesting way I was thinking that all that went wrong in technology is because of me.
Steven Bartlett: Why?
Mo Gawdat: Because I contributed to building this. When Geoffrey Hinton and I — one of my favourite moments when we were filming Chasing Utopia — Geoffrey is very big for all of us, we really think the world of him. And I was telling him, just as a sort of older mentor, "Geoffrey, do you regret doing this? I genuinely believed when we were building those things that we were going to make the world better."
Steven Bartlett: AI.
Mo Gawdat: Yeah. And he said, "Well, yeah, I too was naive. I didn't think we would get there so quickly before we figured out the alignment problem."
Steven Bartlett: The alignment problem — that AI has our best interest in mind.
Mo Gawdat: And I think all of us were faced with that. All of us were faced with that idea of: we're building the best thing ever for humanity. And then suddenly you realise, oh my god, in the wrong hands it's the worst thing ever for humanity. I came to terms with this at the end of 2024 — that yes, I can try, but I accept that the world is what it is. And from that point of calm and stoicism, I think I can have a much bigger impact on the world.
The closing question on legacy
Steven Bartlett: We have a closing tradition where the last guest leaves a question for the next guest. The question left for you is: what's the legacy you want to leave?
Mo Gawdat: Nothing at all. I've been asked this question. I get asked it all the time. I don't know why so many people are asking — am I going to go anywhere anytime soon? Legacy is — why would I care if I have a legacy if I'm dead? Why does that even make any difference? Here's an interesting thought for everyone. If karma is real — and I genuinely believe it is — and if we're not just physical beings, that we're physical and spiritual, then I'd rather keep all of my karma for my spiritual side.
Steven Bartlett: What does that mean?
Mo Gawdat: I don't want anyone to remember anything I ever did. I just want to leave a positive impact on the world and take all of that as karma for my next journey.