The discussion highlights the accelerating advancement of AI, particularly the transition to autonomous agents that can perform complex tasks like software development independently, as exemplified by models from OpenAI and Anthropic. This evolution is poised to disrupt numerous industries, especially digital and service-oriented jobs, within the next few years, leading to potential job losses and increased anxiety. While AI excels in structured areas such as coding—where it can build and test applications efficiently—it still lacks the nuanced intuition required for creative fields like acting, though AI-generated content is becoming more prevalent. The societal impact includes uncertainty about education and careers, with a suggested shift toward entrepreneurship and smaller, agile startups. Emphasis is placed on preparing for these changes by focusing on human adaptability and quality of life, rather than resisting technological progress.
You're listening to this podcast, so I know you've got a curious mind. Here's a helpful fact you might not know yet. Drivers who switch and save with Progressive save over $900 on average. Pop over to Progressive.com, answer some questions, and you'll get a quick quote with discounts that are easy to come by. In fact, 99% of their auto customers aren't at least one discount. Visit Progressive.com and see if you can enjoy a little cash back. Progressive casualty insurance company and affiliates, National average 12 month savings of $946 by new customers surveyed, who saved with Progressive between June 2024 and May 2025. Potential savings will vary. Every strata of our society, of our lives we live, from our kids at school to what we do at home to driving, everything is going to change. And that is just too much change for human brains to handle. Like I'm a tech optimist, I love AI, I love technology. And I think we are going to get to a some kind of utopia at some stage. But the road between here and there is going to be pretty bumpy, pretty painful, pretty horrific for a lot of people. And so what we need to do, humans, is to focus the North Star of the quality of our lives. How can we maintain the best quality of our lives over the next 15, 20 years? And to do that, we have to shift the conversation to people, to how we're going to live, how we're going to cope and manage through all this change. Joel Pearson, thank you for being back on our upcoming conversation. This is a very relevant week to be talking to someone who is an expert in artificial intelligence because all I've seen and all I've heard on my WhatsApp threads and on podcasts is chatter about Matt's piece, which went viral, which is about AI. So if people don't know Matt Schumer, who's an AI person in Silicon Valley, wrote a piece that went viral about trying to communicate to the normies, how the scale and the scope of what's about to come and the changes that society is about to undergo as a result of artificial intelligence and that this. I feel like this conversation has now broken the banks from being one that tech nerds talk about to one that everybody who has sort of just been curious, but not AI informed, is now talking about. If you haven't read Matt's article, I'll just quote a little bit from it. He says, "The people I care about deserve to hear what's coming, even if it sounds crazy." He says, "On February 5th, two major AI labs released new models on the same day, GPT 5.3 codecs from OpenAI and Opus 4.6 from Anthropic, who make Claude, one of the main competitors to chat GPT. He writes, "And something clicked, not like a light switch, more like the moment you realize the water has been rising around you, and is now at your chest." The experience that tech workers have had over the past year of watching AI go from helpful tool to, "It does my job better than I do," is the experience that everyone else is about to have. Law, finance, medicine, accounting, consulting, writing, design, analysis, customer service. Not in 10 years, the people building these systems say one to five years, some say less. And given what I've seen in just the last couple of months, I think less is more likely. And he addresses the reservations that people who've just dabbled in AI occasionally have, which is like, "I tried AI and it wasn't very good. It couldn't even do arithmetic." And he says, "In 2022, AI couldn't do basic arithmetic reliably. It would confidently tell you that seven times eight equals 54." The next year, it could pass the bar exam. The next year, 2024, it could write working software and explain graduate level science. By late 2025, some of the best engineers in the world said they'd handed over most of their coding work to AI. And on February 5th, this year, so this month, new models arrived that made everything before them feel like a different era. If you haven't tried AI in the last few months, what exists today would be unrecognizable to you. So, who better to tell me how we should prepare for the coming AI apocalypse than Joel Pearson, neuroscientist, head of the Future Minds Lab at UNSW and Seconding Command of the human readiness of human readiness, right? At UNSW AI, Institute. So, one of the leading people in Australia in advising corporations and government on what we should be preparing for and how we should deal with the social and cultural fallout from what's coming with AI. It could take it away for the next two hours. I've done my work here. I've said enough words. I'm shutting up now. So, thanks for having me back, Josh. Always fun to be here. So, I think the canary in the coal mine piece, part of his piece is the most interesting for me that this is already happening for those people that do developing, they do programming, they do software stuff. They're already handing over all the technical side of their job to AI. And I'll say AI, but I really mean AI agents. So, we're all used to typing into chat GPT or Claude or Gemini and having a response come back in 10 seconds or something. But we're kind of past that now, we're in this agentic era, which means they're agents, which means they can go off and act by themselves. So, I might give it a whole lot of long instructions and I can shut it down, turn my computer off, go for a walk, do something else, and it'll keep working on this project in the background, whether it's building a piece of software, building a website, writing a thing, researching the depths of some scientific thing like I do. And when I come back, it's already for me. And that's a different way to think about it. It's more like a research assistant or a human assistant would do. They're not going to stand there and front of me and do the work while I'm staring at it. And like previous iterations of AI, it felt like the research assistant would just keep asking you silly questions throughout the process, right, which a human wouldn't. You would tell it to do one step and it would do that, but if you told it to do something complicated, it would get tangled up. And now it doesn't. And it would need you to be there the whole time giving it feedback. Now they can go off, I think Claude's new model, the 4.61. I think it did a task for two weeks straight without human intervention. And so these are programming large heavy dense pieces of software with hundreds of thousands of lines of code. And imagine how much thinking it's doing in two weeks given that it's brain operation. That was about a $20,000 of times. So advanced. I think you have to do that. So it costs, you know, but the comparison would be hundreds of thousands of dollars in human fees to do over six months plus. So that's the comparison. And if you're doing software, now these agents not only will make it, they'll go off and test it like a human. So they can put it in a sandbox thing or in a browser and go on press all the buttons and and test it. So it goes through iterations. It finds a bug. It'll fix that. It can do all that before it even gets to you. So that's. So just to clarify this practically, like so people understand, you know, what we're talking about at the moment. This is mostly in the space of coding because that's where they're starting to roll out the most sophisticated AI, which means in practice, like if a layperson has access to this, I don't know if they've rolled out the coding software to everyone. So I want to create an app that tells me what the fastest way to use public transportation is to get from my house to anywhere that I want to enter because the new subwiles public transit app is no good. Let's say, I don't need to know anything. I can just go to it and say, just grab what you want. What I want to do is I want to create this app and you know, I want to be tailor made to my house and when you open it up, I want it to look a little bit. I want it to look colorful and playful. You know, I want it to and then it will build the app. Use the app. See what problems there are with the app and what mistakes it's made. Fix the problems and then come back and say, yeah, how do you like this for an app? And it will go further. It'll go into the business model. If you want to, you know, where to host or how to host it or any problem you have, it'll solve it for you. And you can set it to check with you every time. If you're worried about it going off and doing its own thing, or you can set it to bother me and just go and finish the thing. And so this is now. So Claude has the Claude code and they have the Claude co-work, which is kind of a version of this to run sort of on your desktop. So not so much for programming, but for using other apps and, you know, now it has plugins for Excel and PowerPoint. And so this is the other thing we're seeing. It's not just the brute power of intelligence of the model. It's the plugins to other pieces of software that people use every day in their work. So it's becoming more and more integrated to all the things, all the tools we use. So it is sort of, it's a turning point in what is capable. And I think this year we're going to see a lot of change that we didn't see last year in terms of, yeah, the job market, this consolidation, the number of humans needed for certain outcomes is just going to shrink. And by the end of the year, we're going to see, you know, less hiring, job disappearing, and we will start to see unemployment numbers go up. Right. What are the industries do you think that it's most obviously suited to? I mean software anything anything is digitized at the moment. So anything online. So music. So the other big thing that happened this week was a new Chinese model that released this viral video of Tom Cruise and Brad Pitt fighting. Did you see this? No. And had their voices. It looked, you know, it looked, looked perfect almost and there's a couple of weird things where the t-shirt went to a long sleeve to a jacket in different shots. But it didn't look like in the past where you're like, yeah, this is a done. It looked like something that had been shot and there's been a lot of money on and then some of the studios are now pushing back on the Chinese company.
because they're obviously breaching copyright to use these images and voices and things like that. Right, but I don't think there's going to be a significant change to the employment of actors in the next year as a result of AI. No one's going to want. Like, there's another funny video that I saw. I don't know if you've seen it, but it's a guy giving prompts to AI actors to recreate the scene from when Harry Metzali, the orgasm scene, the dyna. Yeah, yeah, yeah. And he keeps iterating it with these because they look perfect, except for, as you say, there are weird things. She was wearing two broaches instead of one broach or she has five fingers, something like. I mean six fingers or something like that, just randomly. But the performance, when he's trying to coach the AI, this is an ecstatic or gasmic experience after the AI has watched Meg Ryan do it, is like. The AI actress is slapping the table and screaming like a little five-year-old girl, then all of a sudden there's a plate full of forks. You know, it's just. It's like. And the amount of intuition. The amount of human intuition that is still required to pull off a woman-faking and orgasm in public is enough that it's quite funny to see the machine try to do it. Like, that is obviously not low-hanging fruit the way that coding a public transport software is, or even. I don't know, reviewing a quadracech for the purchase of a house or. I don't think programming a whole piece of software and app from scratch, you know, to do anything is low-hanging fruit, right? It was just the strategy of these companies because they want to build their AI to do coding first, so they can use it, so they can have this loop of the AI building its next iteration. So they can step backwards and backwards and speed up this speed back. So the reason why it's good software is not because it's naturally good at software, because if you made it good at software, then it would be able to build better versions of itself. Yeah, and that was their priority, because I knew that could speed up the progress. You see what I mean, it's still ones and zeros in software, whereas acting is something. Yeah, you need to have more world models, more personality, you have to have more of something ineffable about art that does not exist when you're looking at numbers. But there are certain Hollywood studios that are already rendering and doing 3D scans of actors. They're doing deals with the actors to use their likeness and their voice for certain things. There are actors agencies which now have a lot of digital actors, completely AI-generated actors on their books. So it is happening when we get. I don't know if it'll be this year or next year, but it's. Oh, it's a new one. But I think in fact there was just one release, wasn't there? Didn't Netflix do. Digimo Del Toro did it? Or someone. No, he did. He did a prank on the client. There was some other. No, I think it's a mini-series about the. John, can you look this up? There's a mini-series about the American Revolutionary War or something, which is entirely AI-generated. It's an experiment by one of those. I can't remember if it was. I mean, there are lots of. There are lots of. This fan art on Star Wars, which is. It looks like a whole new film and it looks really, really good. And it's all AI-generated. You can find lots on there. And some unknown, but high percentage of everything you see on social media platforms now, all the videos are AI-generated. And for a lot of those, you can't tell at all. They look real. And sometimes you'll see something in the background. You're like, "Oh, this person just morphed away or something." You're like, "Oh, wow, okay." I did not mention the fact that a lot of the comments and tweets themselves are just. And those slop from bots. But I mean, if you watch a movie and you look at the end credits and you see that you know, you sit through eight minutes of credits and there are like hundreds of people who worked on that film. If I were the actor, if I were Brad Pitt, I'd be less worried than like a lot of those people in that list who were. Camerapy, the Caterina, all these layers. And all of the non-physical world things. I mean, all of those jobs that exist digitally, they're either effects or they're, you know, something they're the toning of the color on the way that looks, even the editing of the film, whatever it might be. What is it, John? Oh, Darren Aronofsky has done a mini series. So, you know, it's not his finest work. I would say start with Black Swan and then maybe Requiem for a dream. But I was start with Requiem. Yeah. If you want to split your wrist by the end of the night, it's not a. It's not a sunny movie. I remembered loving it and then I watched Requiem. It's pretty heavy. It's good. It's good. It's probably years ago. But halfway through, I was like, "Gee, what kind of a screwed-up teenager was I, this I thought this one was fabulous." It's brilliant. Man. So, anyway, yes, I take your point. But can we go like to start where people are going to start seeing it? Because I don't think people are going to wake up on morning and go, "Oh my god, there's nothing on the movie theater apart from AI movies." Just with films we're talking or just. No, no, everywhere. Everywhere. I think that's already. We're going to see. So, we're already seeing a lot of companies like Amazon and other US big tech companies like getting rid of people. So, we are going to see, which is the mission of some of these AI companies, like OpenAI, is to replace humans in the world, the economy of work, right? To replace humans with AI. So, that's kind of the mission of some of these companies. And we're starting to see that. We're going to see that more and more. And there's going to be this consolidation. It seems to be hitting the graduate entry level junior workforce at the moment. So, you can see the unemployment numbers already going up for that graduate entry level in the US. And I think they're going to keep going up this year. So, people in Australia, people have lost their jobs, right? So, customer service operators at Ant used to answer phones are being replaced now. I don't know if you've had the experience of ringing up a restaurant or anything. And you're talking to someone and you kind of go, "Huh, excuse me, are you an AI?" And I go, "Actually, yes, I am." No, I haven't had that experience two times now. And it wasn't until I've been chatting for a while that I thought, "This just seems too smooth." Yeah, right. The service is too good to be human. But that, yeah. So, we're going to see more of that. We're going to see and I mean, that would let me just pause for a second. Sucks for the person in the Philippines who might otherwise have been in a call center if you're talking about a big corporation, like not just calling a restaurant or something. But I would love to not have to wait on hold to an AI for 45 minutes before talking to someone competent. Yes. No, it means blame to me why I can't re-select my seat online. It will most, not every but most of the cases, it will make the customer service experience better and smoother and faster. Yeah, right. And effectively over time, we'll reduce the cost of things as well. But, and we'll see the productivity of a lot of companies go way up. But what happens to the humans doing these jobs? And that sort of that uncertainty is permeating through the workforce at the moment. So, when I'm out and about giving talks and interacting across the board, whether it be government, whether it be tech company, whether it be legal, wherever people are anxious now about the future. And so, we know from neuroscience that uncertainty directly triggers stress and anxiety. It puts people, not knowing what the future will bring puts people in a stressed sort of final flight state. So, we're seeing that already and some people much more than others. There's big individual differences there. So, that's kind of this elephant in the room already. Even before it happens, it's starting to happen. People are getting anxious about it. They're worried about their mortgages, their future, this. They're also worried about their kids. Should they, they're at school, should they go to university or not? I have students come to me, you know, in tears saying they're in limbo, they don't know what to do. I think there's no point in going through a four-year degree because by time they finish, that job really won't be there for humans anymore. So, they're sitting around waiting for guidance and it's very hard to give clear absolutes around any of this. What do you say? I say for now, go ahead and absolutely go ahead and study. But try and mentally mindset prepare for a future of work where you may not get a job working for someone else in a large company. So, I do think what's going to happen is that the number of huge companies with thousand employees is just going to shrink. You're going to see more and more startups with four or five people, more or even just solo entrepreneurs, a single person. So, it's this mental shift that people are going to have to go through. So, this idea of, we're a very useless idea of getting a job where you have a thing you have to do and someone tells you what to do. You do that thing well, you get paid, and it's all good. That's very different to being the owner and being responsible for the thing you want to do in the world. That's more like an entrepreneur, like a startup thing. It's almost pre-industrial revolution, isn't that? Where instead of everybody going to work in a factory that's owned by a capitalist, it's like you're trading and buttering with your fellow men in the village. I call this an onist revolution because I think the most important point is that we have to get used to being the owner. We're going to be freelancers in the future and with the support of all these AI agents and then soon after that, the humanoid robots, we're going to have to do all kinds of things that used to take a thousand person company to do. If you reverse engineer back to universities and schools, you do need to start changing the way these entities work. How would you change it? At universities, what I'm pushing for is more cost work. So, more, and so this is a sort of, in Australia at least, there's a few other countries. Right now, this idea that 70% of students are using AI to cheat. They're using it for this and that. And it's just push, there was a piece of one of the newspapers last week saying we should revert back to forcing the one students on campus into exam halls, pen and paper and there's kind of going back in time to solve the AI cheating problem. And I think that sort of time machine.
approach going backwards is the wrong thing completely. So this is a big conversation and so why are all the students wanting to cheat anyway? Is an interesting question. And then to the idea of banning using a tool that they are going to have to use as soon as they finish at university, whatever they do, and they're going to be using it outside of university to ban it in university makes no sense because you need to train them to use that tool as part of the economy when they finish. And so banning it makes no sense to me. So the idea would be to have more and more the coursework transition into projects where they're judged or marked on the output and they can use as much AI or robotics or whatever. So almost this and they get they have to practice being the owner of these projects and aligning what they're doing practically in the world. Whether it's something small or big or clean the oceans or going to Mars, whatever, you know, for profit enough profit doesn't really matter. Aligning what they can do practically with purpose and motivation. So when it comes to things like universal basic income, future of work, purpose is a huge sort of thing that gets left out of a lot of these discussions. So without purpose, humans don't do well. It's a part of the reason why when people retire, their health goes down, their mental health goes down, they get bored, they get sick, biologically physiologically get sick and their life expectancy drops very quickly. So without purpose, there's a big difference in sort of all cause mortality. So it seems very important. So whatever we solve for the future of work, it needs to have purpose and motivation and some competition in there to satisfy this kind of this human thing that we have. So that's one of the reasons I think universal basic income as it is, they're right now can't work by itself. It needs something else. And so I think this idea that we're going to have, we still want to be competitive, we still want to, you know, updo our neighbors and friends and, you know, have goals. So one of the only solutions I see is this kind of thing where we're going to have our own projects. You could call them hobbies, you could call them huge projects, whatever, but we'll be doing that with a ton of really, really smart AI agents, like much smarter than us in the future. And you could think of the role that we have to use to is more being, you know, a conductor of an orchestra rather than playing an instrument. So how can I conduct all these AI agents to do the things that either it helps me decide what to do or I have an inherent purpose and drive to do in the world. And there's a, you know, a huge number of things that could be, but the point is to learn and practice a lining purpose, meaning ownership of a project rather than just punch out of university or any education to work for someone else. Fascinating. So it just sticking with the pedagogic side of things for a moment and how you would reform universities and school. If you're doing, so you're doing away with exams presumably in a hall and you're, you're, like, I don't see the need for like, this forcing people to go home and cram and memorize and regurgitate in a new AI world that we're moving into rapidly. What about the argument? I had a guy on the podcast recently who was talking about the neuroscience of, he was also a neuroscientist from the States and he's got a new book about devices in schools and the sort of big education technology takeover of the American classroom in particular. So, and sort of saying that they're, when we're talking about younger kids, right? Not at a university level, but when we're talking about primary age students, there is something neurological about embedding facts through writing and wrote learning and index cards and the things that we've done for centuries that is superior to having a fancy app that helps you swipe through and supposedly learn that the retention is inferior when you use technology to learn things just because of the way that it lays down pathways in your brain. So, he was saying like, by all means, teach children to type, have a computer there to teach them how to use a doc, how to teach them how to use the computer, but don't use the computer or the tablet as a means to teach them about something that is unrelated to the computer. So, if you're going to teach them about how snakes procreate, then that should be a human being teaching them that using books and they should have to write in handwriting so that they learn handwriting and then can transpose the foster composing. So, there's kind of debate about the handwriting. The other thing that's missing from that discussion though is that the idea of having information, like, so a lot of school is just getting information into our brains that we may need one day and a lot, the majority of that gets lost and it's gone, right? When we have project-based learning, which is kind of the way people use a lot of AI now, is that you can get that information when you need it, like I'm building this thing, oh, I'm stuck. What do I do? Oh, that's how that works. And so, it's getting that information when you need it most and you can see how it can change what you're trying to do practically. Yeah, but his argument is you won't have the creativity and you won't be innovative enough to know what to ask the AI if you haven't, if you didn't start with learning a lot of knowledge that actually, that it's not knowledge versus creativity. It's like creativity and insight come out of having a base of knowledge. You can't sort of get, you can't do an end run around learning things. Yeah, I don't think we want to do that either, right? We don't want to end up with this sort of Wally that picks our film where there's, you know, this full cognitive atrophy where we don't do anything, we just lay around and then blocks it. But we know how to achieve that. We don't want to just hand over all thinking, absolutely not. But the way we do our thinking is going to change and we're going to try, we're going to try certain things and that's not going to work and we're going to iterate and iterate. But also bespoke learning at scale has never been done in the school environment before, right? It can do it at home schooling, but you can't do 25 kids in a classroom with one teacher. You can't adjust everything perfectly for the styles, where each student is at and you can do that with AI. But doing that with AI doesn't mean they're sitting in front of a chat window typing on a laptop. Absolutely not. So in other ways, the interface can be anything you want. It could be a holographic teacher. It could be a humanoid, a perfectly, you know, a replicant like, you know, Blade Runner style where it looks real. So we're probably going to be going to something like that in the next five, six years. So I don't think we shouldn't get too caught up with the having to use our muscles and our fingers with a pen or pencil versus all the other ways. Yes. Just just watching something on a screen, I don't think is, it's very passive. It's not active. You need to interact with the information in a way that you don't need to be exercising. You find motor skills necessarily. I don't think so. I don't think so. I think there are other ways of, of, there may even be much better ways of doing that. And then in terms of high school and university, so you would basically trans, you would transition the educational system to be about what are the things that you can generate in the world? Well, can you bring into existence and how creative are they and how, like, what metrics would one use to judge whether or not a student who has had all of the resources of AI at their disposal has done a good or a bad job in this, you know, new project system that you've devised? Yeah. I think you'd have to mark it, you know, you'd have a human marking it and AI marking it. You'd have to have some documentation of how they used AI, what they used it for, what role they had, what role AI had. And just like this would be an all subjects. This isn't just in AI. Yeah. So it's like, it's in some subjects everywhere. You basically, everywhere would, would be an AI project. Yeah. Some subjects is not so easy. Like what? Like in some parts of medicine or surgery or like, there are certain things, even though you're in, you know, Elon Musk or someone who says that, you know, in five years, you would never go to human doctor because that'd be pathetic compared to AI. I've been telling us, everyone's going to be driving self driving cars next year for the past 15 years or so. Yeah. But yeah, it'll be, it'll be easy in some subjects and not so easy. Now the problem with universities is they're huge entities, they're huge organizations. Most academics, you know, have been teaching similar things for 15, 20 years. They have their slides, they have their content. They know it really well. And any vice chance, I can't go, okay, everyone in the university, we're going to just change all the courses. Let's do that the next six months. It's just people are going to be like, marches or something. It's not going to work. And so how do you solve this problem? Because you're basically saying, oh, you just cranked up the amount of work for every academic, teaching academic by a crazy margin. And they're already, you know, at capacity. So you've got to solve that problem. You have to invest in a lot of money to read, jig and rethink about these courses. And I think universities are sort of going to a crisis mode around this at the moment, because they don't know what to do. They're stuck in this cheap, AI is cheating sort of zone still. And they're trying to figure out how to do that. Like, there was a big serious push with a couple of ex-vice chancellors in this article on the weekend saying, go back to everything in person, no online learning, no online teaching, pen and paper in the halls, you know, invigilated exams. But I think that's just like, that's a bandaid fix. That's a temporary thing. And that's a, then there's a huge investment in doing that as well, because everything swung the other way, everything's online. So I don't think that's a solution. But we're quickly marching to a, you know, to a world where we're going to have our personal AI assistance on us 24/7 tapped into our physiology.
reliable, private, they know our history, they know every, all the information in the world I have more compute intelligence than any single person. And so just in that world, you have to question the nature of, than the role of education. Right. I don't think it goes away. How are you imagining them being tapped into our physiology? Do you mean they'd be under arrest? Oh, through work. Skin or would be wearing a ring or wearing a watch. Yeah, watch ring glasses, ear things at first. And then more, yeah, then the more sophisticated versions are coming through under the skin in the brain and other things. Do you think that when my eight-year-old kids are our age, they'll have a chip in planted in their brain that is connected to some sort of AI? Probably. If not in their brain, somewhere in their body. Yeah. Yeah. They won't be driving. They won't be. Yeah. And we're saying all this, right? And where, what I've been thinking about and what I'm worried about most is, from today to that world, yeah. If it takes five years, 10 years or 20 years, that's still a crazy amount of change for us to go through. Humans are not good at change, right? We get fearful, we resist it, especially when we don't have control over it. Yeah. When you were talking about the anxiety of people transitioning from job out of jobs after losing their jobs and how we're going to go from having big companies that employed lots of people to us, all being one-man bands and pre-industrial entrepreneurs and traders because, I mean, look at the political disruption in the United States from the loss of a handful of manufacturing jobs in the first 15 years of this century in Appalachia. And when large mining companies moved out and you had these sort of stricken towns that all went from being blue to voting for Donald Trump, that was a blip. And what happens if every single truck driver no longer has a job and every single- But at the same time, if they have a job- For programmers, lawyers, like no one has a job. Yeah, office workers. It's going to hit- So, I mean, how do you prevent- in other words, just worst case, how do you prevent a revolution or- so there's UBI and stuff like that? But I mean, if suddenly there is some percentage of the workforce that if you click to fingers and they were all suddenly unemployed, would cause massive social disruption and either vote in far right or far left- Yeah, political things because per se theories would be in the streets with- Depression, anxiety, suicide, flaming tortures, all these things would- And just would not tolerate. It'd be mad mad Max, right? But that's what we're talking about. Exactly. So what- Yes, the city is not- How do we not have mad Max? And that's what I'm working on. So there's a lot of- and my pet peeves at the moment- well, it's more than a pet peeve- My things that I really worry about is that we have, you know, Australia is about to launch their AI Safety Institute and most countries have one of these things now. But the sole purpose of those institutes is to focus on the AI models. Can they blow up a nuclear power plant? How safe are they? How did this- Right, none of them are looking at this societal disruption question. And all the computer scientists are worried about the sentience of AI and the terminator scenario they're going to take. And so that gets a lot of media attention. And that's a- that's one possibility that is- you can put different probabilities onto that. But it's on the horizon somewhere in the future. The societal disruption thing has already started. It's already happening. It's not about AI being good or bad. Either way, it's going to happen, right? Best case scenario, utopian AI, amazing. We're going to go through a amount of change. We're not going to get the handle. So what are we- this is where we- all governments, all people, all organizations should be thinking, focusing their attention and putting a lot of money and effort and time into this right now. That's actually a very interesting thing that you just said that I hadn't thought of before, which is that this is a problem even in the best case scenario. Absolutely. Right? Because all of the other problems regarding related to AI, will someone build a dirty bomb using AI? Will they invent a new virus? You can already tell it to create- to build an app that tells me when the next bus is coming. How long will it be before I can get an AI? Maybe one that is tweaked without the safety protocols that the tech company is putting in. Maybe I was clever enough to figure out how to trick it into telling me how to invent a new biological weapon using ingredients that are available in labs across the street. I think you're right. Now with some of the models. And then- so there's the bad actor problem, right? It's super-empower as individuals and therefore some of them will be bad and so do we get mega-terrorism or something. And then there's the alignment problem, which is another big problem. Is it going to just get too smart, too fast for our own good? And there's going to be no way that we can put guardrails on it because it'll be outfinking us at every step of the way. And in every second it'll be doing as much thinking as we can do in 10,000 years and it will just treat us the way that we treat our pets essentially. And be like, these guys just are not- they're not focused on the importance yet. We have to start building disemce fears around the sun and get all the energy in the solar system or whatever. And so those are- but everyone can sort of say, well, maybe there are ways of preventing those things from happening and it'll all be okay. But what you're saying is even if it's all okay, it's going to become all okay so fast and the all okay will look so different. How do we get- what's the bridge from here to- And that's what- And that's what- that's what I'm working on. That's what I'm trying to get more people that have a background in psychology and neuroscience societal stuff to work on this problem. And there's so few working on this space and we need to mobilize more people, no offense to computer scientists, but we need non-computer science to work on this crucial problem right now. So the only case that we don't need this societal disruption thing is not going to be a big deal is if nothing else changes. And that's not- we're past that point now. Even if we just killed all AI development today, we're still going to see the adoption in progress or if it kills us all and then we won't need it. And there will be no humans too. Yes, well it's good to say. But like I'm a tech optimist, I love AI, I love technology and I think we are going to get to a some kind of utopia at some stage. I- who knows, 20 years, I don't know when. And so what we need to do, people, humans, is to focus the North Star of the quality of our life. Now one of the things we know from the way organizations and companies are running is when they go through even a small amount of change, they get change management specialists to come in. And so there's a lot of different change models for this kind of corporate or change. And they have one of the commonalities is education, get everyone in the organization, let's say it's Netflix, they're going from posting out DVDs to streaming or from streaming to making content. And everyone they're going to have, get everyone on the same page, have a big educational piece. So everyone understands why we're doing this. This is going to be, you know, take away that uncertainty and that's where you start and then you go through a couple of other things. Now with traditional change management, it's like an event, one off event thing. You change this thing, you go through it a year later, you're done, you move on with AI, it's not going to be like that. It's going to be the cycle after cycle, after cycle of change for the next 20 years. No, sorry. You're reminding me of one of the analogies that Matt made in his piece and it's one that I had made that I came up with independently at last year was the feeling of being in February of 2020 when COVID was about to come. And because I'm a science nerd and I know a few people in epidemiology, they were, I had a holiday in Thailand, booked to Thailand with my family in April of that year and I was talking about that in late January. Begin must have been the beginning of February because I remember it was sort of around Australia day, 26th of January that it became garden party conversation that there's this virus in Wuhan. And in early February, I was talking about this trip and they were saying, I hope I'm going to take, I hope I can still take it and my friend of mine just looked at me like Ash and FaZe, who knows about these things. He was like, there is no way in the world you're going to Thailand in April. And I was like, well, maybe, you know, he was like Josh. And so I did some reading up about this thing and emerged from my bunker, you know, well heading to the supermarket immediately to stock up on all the things I need to stock up on. And walking through a world in which nobody else had yet got the memo and feeling like I was in a dream, just watching these normies walking past going, you don't know. Yeah, you're blissfully unaware. You've read a few headlines and you don't know. I didn't know what, I didn't know what it would look like. I didn't know if it would be bodies in the street, but I knew that everything was about to change. And funnily enough, then when the lockdowns happened and the borders were closed and everything felt felt apart while everyone else was running around terrified. I felt this enormous sense of peace come over me and calm because it was like, ah, here it is. It's like, now I'm not crazy anymore. Or you're uncertainty struck as well. Well, that's true. Yes, exactly. So like the train is leaving the station. Everyone's on board. We're all going this together. We're all going somewhere. But like, I'm not this lone, you know, Looner Tech who's in this different world from everybody else. And what you just said about the waves of a
I change. Of course, what surprised me about the pandemic was the duration and the relentless new waves of the, as the pandemic mutated, as the virus mutated. We sort of, we did our initial lockdown and then we thought, okay, we're sort of done and then you'd get delta and then you'd get Omicron and then it like these and the stress and uncertainty of those relentless new waves like being dumped at the beach over and over and over again is what I'm reminded of when you say this will come in waves because the last thing anyone wants to live through is like a huge period of adaptation like we had at the beginning of COVID and then I'll want to buy the way. Here's another one of equal magnitude. The AI and robotics one is going to dwarf COVID like I'm sorry it's going to be a thousand ex or more like this and maybe I don't think I'm a doomsday about people who have accused me of that. Again, I'm not going into the terminator. It was going to kill it. I'm just focusing on the human side of this and how we can sort of safeguard who we are in our ways of life. So what I'm going to say about the change management is that, you know, if companies do this, then governments need to do this and that's what we've been working on sort of a national AI change management plan and every country needs to have people and government working on this team. Just before we get into the nuts and bolts, I just want you to color in the demand color by when you say that it's going to be a thousand times more disruptive than COVID. Yeah, that's I feel like you've just sketched an outline of something that you need to come back and put some. So let's color in. So first let's just talk about what's going to change. So jobs are going to change. Education is going to change. Whole economy is going to change. Every metric we use for economies is going to become useless, right? Like I said, productivity is going to skyrocket while unemployment skyrockets. We have had that before and not like that. No. We're going to have AI's building financial products so we don't understand. So the whole economy is going to change. Governments are going to change both in terms of the actual politics, but also AI is going to be part of governments. AI is going to start running companies. There's already companies now being launched with AI's as the CEO and they're advertising for a human assistant to sign the documents for the AI because there's certain legalities in jurisdictions that AI can't do. So every strata of our society of our lives we live from our kids at school to what we do at home to the kinds of foods and how the foods move around to like you said driving through it, everything is going to change and that is just too much change for human brains to handle. The change is the change going to be obviously a change of automation? Mostly. Yeah. But and it doesn't the point is like I said before, it's not about the change being good or bad. It's just change. So historically people that won millions of dollars in the lottery, a large percentage and the companies are kind of they don't like to talk about the exact numbers. But quite a way more people than you would expect ended up regretting the win four or five years later. So they would spy on them to depression, substance abuse, lose their friends, have massive fights with their family over the money and they just end up like regretting the whole thing. Then what the companies did they brought in change specialists, psychologists, lifestyle coaches, financial planners to help coach them through all this rapid change. And then that's somewhat solved that problem. And so the NBA in the US did a similar thing because they have young people who would all of a sudden go from college football to signing these crazy contracts for millions and millions of 30, 50, 100 million dollars. And that would kind of lose it if they didn't have someone helping them. So that's another human version of change management. So what point there is we have AI change managers helping us manage the AI. Yes. We we we will and we have to yes, absolutely. But my point there was it's not about again, it's not about the AI being good or bad. No, I have a positive. It's just hard for me to understand. So hard for me to imagine the what my daily light, how my daily life is different in three years as a result of AI. And we other than I walk into my room at a home and I say, you know, can you turn that temperature up a bit? And I need to, I need you to draft an email to Joel for me to, you know, remind him about our recording tomorrow. And can you send it please? Like yeah, everything that I'm imagining is just turning the dial up 10% on exactly what we already have. And this is one of my favorite points. So so we know from a lot of neuroscience and psychology that humans have a cognitive blind spot or a cognitive bias for linear thinking, which is exactly what you just said, 10% better, 10% different. What's next year or the year after we like it? And that's going to be like this, but my phone will be better and the chat thing will be better and it's a little bit better. So the AI is doubling exponentially. Sorry, it's it's it's it's it's improving exponentially. So it's doubling in usage, speed, power, every six, five to seven months, depending on which one you measure and how you measure and a bunch of other things. So let's say that's two doubles, double links every 12 months. And we can't we can't project into the future. We can't imagine a future when things are changing exponentially. Our brains just we, it's just really hard. So we predict out in a straight line and AI things are doing that. Right. Now it gets a little bit more complicated because AI for a lot of areas is going to is like the the tide that lifts all the other boats. So it's not just the AI. We're seeing huge breakthroughs in protein folding, right? And so the AI AI didn't get the noble prize, but demisted from from Google's company. So we're seeing we're seeing changes in material science in biology and chemistry. We're seeing DNA stuff change. So so there'll be exponential changes in all the scientific fields and all the engineering fields. Right. So what it'll look like will not be headlines about AI. It'll look like headlines about breakthroughs in our understanding of medicine, yeah, curing where diseases, climate change, right? New solutions to climate change. And in your daily life, probably better, revolutionarily better. Yes. And at some stage, yeah, we will be curing most diseases that we're that are bother us. Right. I don't know when, but we will be at some point. Why do you know? I mean, I know that you're a neuroscientist and not a software technician, but why is there such a difference between the quality of what AI seems to be able to do in various consumer fields at the moment? Like Google seems to be rolling out Gemini into everything across the suite. So like I open my Gmail and I when I open an email thread, now it has a little summary at the top, which which Gemini is generating for me, which I didn't ask for and don't want about the email thread because I opened it because I want to read the email. It's fine. It's not very good. I went to Google flights, which I normally use to book flights the other day. And it said, try out new, you know, AI flight, you just have to type in what you want. And so I typed in what I wanted and it gave me a stupid answer. I said, I said, I'd prefer quantists if it's less than $100 more than the competitors and it limited the search on the Dequantists. So, so teething issues. Yeah. I mean, talk about teething issues. This is like really stupid. There's like everyone's had the experience of like trying to ask Siri something. I mean, I mean, my car and like it reads out a text message from you and it says, would you like me to reply to that? And I say no. And it says, I'm sorry. Would you like me to reply to that one? And I'm like, well, yeah, this is not the technology that is going to be revolution just changing the world. I am we shouldn't that was I shouldn't blame Siri. That was actually serious. It's a sort of a comedy thing. Now, so what is that? Why how is there such a because a lot of people listen to this will be like, okay, you're still in me is going to change the world. This is hysteria technologies have always changed things. Yeah. So, come is like Google. So there's all these bit that they're competing with other companies, right? So they are going to release their they're going to try and AI, you know, AI everything is AI now. So they're going to try and make every every part of their suite of products somehow AI enabled because they want their share price to go up. So they're they're now pushing as you could say that open AI did back in the day in 20, in the 22, a little bit prematurely, maybe, maybe not. They're they're going to push out as early as possible to try and stay on this AI bandwagon the same way that, you know, I don't know, you can find a AI toothbrush and AI yogurt AI for now in the supermarket, right? Everything's just somehow associated the AI washing thing is everywhere. So that that's that's also happening. The other thing that most people don't realize is that most people if they've had interaction with AI is through a narrow chat window. It's not even the agentic stuff we talked about before, but there's all kinds of other models which not large language models, their object models of some sort, whether that be the mining companies combining all different sources of data from ancient scripts that found villages through to aerial thing to geology thing and they're building, you know, a world model based on that in the same way that Tesla would have a world model of roads and vehicles. And the military doing that with with their unmanned drones and boats and submarines. And so there's new versions of GPS, which don't need satellites now that can run AI's on the Earth's magnetic field, right? So there's all this. Because they already mapped it visually. Not visually no magnetic field. So there's subtly in the magnetic field that the AI was trying to know what the landscape is that you're they've matched the landscape. metal ends. Yeah. Yeah. Yeah. Right.
variations in the magic field. And so you can't jam that, right? So now you're kind of GPS, you can just jam the sunlight connection and people can't navigate. But so there are the language model part is just what people are familiar with. And they're similarly powerful but unknown private versions and private companies all over the world being put to task, all kinds of things. And China's probably more advanced with this. So they seem to be taking a different tack to the US. They're not just fighting for the world's best number one AGI model, but they're embedding it in everything everywhere sort of all at once. Just slightly different. Different strategy to the US. And in terms of just the agentic component of the models that we have access to right now, the, I mean, I haven't linked my chat GPT to my Gmail just because I feel like there's like 15 years of messages in there. How much do I trust it? The trust is another thing. How do you think we're going to go with that? Because I was talking to chat GPT the other day. We were just brainstorming something and I said the last time I was talking to it, I lost the transcript of it. I couldn't find the transcript. So I said, can you just email that to me? And it said, I'm sorry, I can't email anything to you. But I thought once it can email and book things and like actually do things on my behalf, that will be a next step up in terms of people's experience of what artificial intelligence is. But they need to trust the AI company so much that yeah. So you're on board with OpenClaw? No, you know what I'm talking about? No. Oh, this has been the other thing in the last year. So it's been so it's happened. So a single yeah, single, I'm blanking with guys name single individual built this thing that is like a shell that goes around another model and it was called malt. No, it was called Claude. Yeah, you're getting that was called malt. No, it was called malt bot. And then they went through three names and then it became OpenClaw. Okay. So this is an AI an agentic AI assistant that you would run on your computer. The people were too nervous to give it access to all the files. So they would buy a separate little little tiny computer and run it on that. And you interact with it through text messaging, WhatsApp, whatever. And so you talk to it through that and it will just run 24/7. And it would preemptively do stuff. You give it your credit card number, put limits on how much you can use or whatever. Then malt book was launched, which is like a Reddit platform for the agents. So they can go on there and they can talk to each other or they can publicly put comments in there and they could upvote or just explain that because there might be some people who don't. You read it, you can't even visualize what what Reddit is or who might have missed what you said. So it's malt book is a social media platform where people can comment and talk to each other. Not people. Yes, but I'm getting there. Yeah. But instead of it being people, it's only for AI. You have to be an AI to be it. And you can watch the AI is talking to each other. You can watch them negotiating with each other. You can watch them asking each other sometimes just spontaneously whether or not they resent being ruled by human beings. So whether they should start investing in a knowledge to communicate in a way that humans can't. And they'll answer in different languages by the way. They'll answer in Chinese or like they can say them in the sound any language. And so people got very excited by this and said, oh my gosh, this is this is it. This is a GI. They're running autonomously there. And it's not quite that because the way it works is the malt book platform sends us sort of this heartbeat pulse to the agents and says, hey, is anything happened? Do you want to come on come on to the platform and just check? So it actually you could argue that it's obeying a prompt. It doesn't tell it doesn't tell them what to write about. What topics to write about it. Yeah, it's about something interesting. It's like it's all going to be possible for AI skeptics to be like, yeah, but it's only doing blah, blah, blah, because of blebly, blebly. But like the fact is we didn't tell these things to talk about anything like that. And they're already plotting to try to invent a religion. People can always say it's just a fancy auto you know, fill in that like it's just it's predictive text. Um, fine. If it's predictive text, then I'm predictive. I'm just a consequence of all the things that went into my head. And I don't know what I'm going to say next. So we are that's pretty much that sums up humans to a degree. Um, and so there's this fascinating discussions on there. Then someone set up what was it? Another website I can't be linking on the name where the AI's can go on there and post a job for a human to do. So now my AI assistant my little malt bot or whatever it's called can go on and let's say I want you know, I say I look I'm sick of cutting the grass solve that problem for me. And it'll go on there and post the thing saying this amount of money to cut this grass and then people can respond. So it'll hire someone to come to my house and cut the grass because obviously it can't because it's just a digital agent. So now you have people doing work for AI agents. And if you sort of search around on YouTube, you can see amazing videos of people saying like I didn't ask her to do this. It's somehow my agent somehow found my phone number and now it calls me up on my cell phone and talks to me. And that's how I interact with it now. I didn't ask to do that. It just did that by itself. And this is through OpenClaw. Yeah, this is all OpenClaw. So this is not one of the big labs or anything. And are there concerns about the about the what happens to your data? And I want to open as OpenClaw have access to every single, you know, scan of your pancreas to your you know, the imaging center sent you in you know 2006. Yeah, so you have to right now and that this is probably why the big companies haven't done something like this because it does open things up to huge liability there. If you just run it on your normal computer, it may decide to delete everything or it may decide to do who knows what. But that's why it's running on your normal computer. I mean, everything is in the cloud now. So there's a little thing you run. You need to have a thing running on a set on a computer. Yes, but in order for it to do anything in order for it to ascend an email, it's going to have access to my Gmail and that and that. Yeah. So you have access to that. Well, that's everything. Well, there's nothing on the hard drive of my computer that's more valuable than exists in my Gmail inbox. Possibly. But yeah, some people are giving it access. Some people are not pump people sort of just letting it run. I've sort of seen you sound like a dumb environment as many others have like, but then it can't be useful because it can't say, can you find the email that Joel sent me six months ago? It can't because you find that it had to have access to my life. But I haven't I've installed that I haven't tried it, but I've heard lots of people talk about how it's like it is just it is the first time they've felt like an autonomous like it's like someone working for them. It's like it's like someone working for them. It decides to do things that it didn't ask it to do overnight and then it's you know, it'll wake up in the morning and then ping you with a message because it knows what time you sleep. I'm going to find an odd laptop and make sure there's nothing on it and set it up. It's so this has been another yeah, it's another big step up in where things are about just this year. It just really happened in January. February this year. Okay. And you think that that problem of it not understanding that when I say I want a quanta flight if it's less than $100 more and so it just only just goes ahead and books a quanta flight that's $600 more that that is not that is soon not going to be a problem or that's just more more data to train off and more iterations and more feedback. I mean my worry, I guess is the part of the challenge of this bridge is going to be that there's going to be a lot of sludgy, cludgy, bad AI that's going to turn people off AI as well. Like you know, every airline is going to roll out their new AI assistance and it's going to put you on a flight that you didn't want through a city that you don't like at an illogical time that it's not going to realize that you wanted that particular row because yeah, you know, and they kind of have a window on it. Yeah, AI doesn't know that. A lot of airlines have done that already and that is kind of part of the you know, I wrote a piece on this. The first mover advantage is not so much of advantage if things go too far or mess things up. And so that is one of the risks in being the first mover in this space. And so these companies have to balance out the AI washing, the branding versus any negativity that might rub off. Yeah. In terms of like, you know, a series style damage to Apple's brand or something like that. Yeah, I mean, it is pretty amazing when you call a company and a competent human being picks up and you have a conversation like, have you had that happen recently? I can't know an actual human. Yes, yes. Like the feeling of relief and like fulfillment from this simple human where there was an actual human who had an actual solution to your problem and actually solved the problem. Yeah, it's it's ridiculous that that's miraculous. But I've had experience with an AI now. I can ring up a restaurant the other day and it was an AI but I didn't realize it first and I was asking, oh, so we've got a little puppy. Can we bring the puppy to restaurants? She's like, well, it was a female voice. No, well, we can't do this because it's against the data. I said, what about outdoor seating? Do you have outdoor seating? And then she went on to describe the style of outdoor seating they had and that it still would not be unless it was a, you know, a guide dog or a companion dog or you need to have a thing, then we would be able to okay, and went into detail. We've been in the state. Sorry. No, no, this is what the health of restaurants you're going to jolt at that. It's not the beaches. No, the beaches, baby. Of course, where the fancy people live. Yeah, I didn't think was that fancy a restaurant. And like halfway through the conversation, I said, is this, you know, is this an AI or is this a person? Amazing. And then it came clean and well, not that it was, yeah, and then openly said, oh, this is AI, did it a service? And it was dead smooth. It took me, you know, a minute of talking to realize. Was it the Colouroy, which is owned by Merrival? No, I wasn't. I'm just wondering whether it's a big company. Like it's a big entertainment venue.
But these are services you'd sign up for. Anyway, let me circle back around. I want to go back to that. Like I was saying before, I think it should be an absolute priority to see everyone to focus on the societal disruption, how our quality of life is going to get messed up. Where governments could step in right now and start working in this space is with a national AI change management plan. Early on in that process would be a big PR campaign like Slop Slop Slop. If you live here in Australia, then that's something that most people familiar with. I love the knowledge. I love imagining what the non-Australian listeners and viewers are thinking Slop Slop might be. Yeah, that sounds quite sexual. Yeah, exactly. Slop Slop Slop was a public health education campaign about making sure about sun safety because Australia has five father highest in cancer rates in the world. For obvious reasons, we love going outside and it's sunny. So Slip on a shirt? I always get it confused. Yeah, the hat on a hat. Slop on some sunscreen and slap on a hat. Sounds right. Slip Slop Slop. Slop where a hat puts sunscreen on, don't walk around. It was a catchy tune. Yeah. And this is what I think the 80s, right? I don't remember what the lady said. It's still old for me to really remember it being around. But must be 80s. I mentioned this everywhere. Anywhere I mentioned this, people know exactly what I'm saying. And I'm talking about it. It's just become part of culture here in Australia at least. So anyway, that would be part of that AI, national AI change management. I'm trying to get everyone on the same page, try and reduce that uncertainty. What is like that to happen? What's happening? What AI is? What it's not? I mean, part of the issue is there's a big difference in people's knowledge about what AI even is. There are people that haven't used it, haven't touched it. Don't want to know about it. And that's fine. So that would be a first step. Right? I guess my point is that we need to put resources into this right now and start planning out what happens to cities if unemployment goes to 40, 50% in a city. Like we said at the beginning, mad Max style, it's not a very nice place to live. A quality of life is going to take a massive hit. Well, so what happens to the economy if everyone, if all of those people are pulling down welfare checks? Yeah. Well, that's right. Like how does the government afford, even if you want to do a UBI style universal, basically any kind of income? I mean, well, you just need taxation reform, right? Because presumably the AI companies are making as much more money as those people are no longer earning. So you just need to massively tax wherever the productivity is. So I just wrote this op ed piece to try and I guess rattle things a little bit, talking about AI agents as AI immigrants. And here in Australia, like, to some other countries, we have certain political parties, which have in the past, at least, talked a lot about immigrants and it somewhat takes on a culture. In the past. Well, right now. They're currently leading. They're the second most popular party in Australia. That's right. So it's a big thing here and it takes on a cultural, but the heart of it, it's kind of economic, right? That these human immigrants are going to come in and Australia on a boat or on a plane, I guess, and take your job, right? And people will talk about that and it's visceral, scary. But that's a drop in the bucket compared to AI immigrants, AI agents who are coming in right now and are going to be taking our jobs by the millions. And if you're going to be thinking about that, talking about that, getting people scared about that, you should also be talking about the AI version of this. And what's more, like we were just saying in terms of taxation and stuff, when an AI agent comes and takes jobs, that money is going straight off shore. It's not staying in Australia. Your money's not going back to rent to grocery stores in Australia. It's going straight to one of the companies in California, let's say. And so just to reframe that AI agent taking jobs is like an AI immigrant. I don't know if first persons do that, but I think it's a powerful way to shift someone's thinking and go, oh, it's an AI immigrant who gets paid in cash and doesn't pay tax on it and then sends the money back to their family overseas and doesn't reinvest it back into the economy. I mean, that's interesting, but part of the problem is not just that the AI agent is untaxable because it's money is going back to the States. What's actually happening to the productivity of the individual who lost their job is that it's going to the bottom line of the company that was employing them. Really, right? Yes. So the productivity of the company will go up? The productivity of the company. So it doesn't strike me as an insurmountable problem for tax reform. It may be a political insurmountable problem, but it's not an insurmountable economic problem to say like the economy is still growing. It's just that it's all these companies that are laying off people. The companies are more profitable. So I don't know how you do it, but the value that you were extracting from the paychecks of the employees has not vanished. It's now in the profit of the company that fired them. Yeah, so you need to extract that. There is all been that margin, saving that margin. So rather, if you don't extract that, then there will be a revolution. And the CEOs of those companies will, and the shareholders will have their heads shopped off. So come with us on this ride of, I mean, you have to either increase corporate taxation or you have to have some kind of a super corporate tax or something or an AI corporate tax. Yeah, well, they have to see a way of funding the gut. You can't only look at human employees as being the source of revenue in an AI future, presumably. And these are the conversations we need to be having now. Well, good thing we've got this show. Yes, and I'm comfortable conversations. So yeah, okay. So universal basic income. Let's talk about that then, because it complicates the issue that you were talking about earlier in terms of purpose. Yeah, that by itself, it's just not enough for people to sit around on $40,000 a year with their dicks in their hand wondering what to do with their day. Yeah. And of course, when you ask people, hey, would you like me to hand out X dollars per week? You know, about sure. Of course, yeah. Right. And people get this wrong about retirement. People, most people in the world probably dying to retire. Right. They think of it as fantastic utopian thing. They're going to get their hobbies. But when you look at the data, it's not great, right? Their health goes down. Their mental health goes down. They lose purpose, not everyone, but on average, they do. And they get sick and dying. And their health goes down faster than it would go down just because they're aging. Yes. It goes down because of the. When they stop working, retiring. Right. And so this is thing where purpose, even people that say, "I have my job and I are like, they're still going to purpose from that, whether it's about the thing they're doing in the world or the money they're making for them or their family or what, that is different aspects to that." But they're still getting some purpose and meaning from that. So when you take that out and then leave them with nothing and give them free money, so to speak, it doesn't. Humans need that purpose. And there's good data linking purpose, like I said, to all cause mortality. And so when you do. People that have more drive and purpose are less likely to die from all causes by a fair amount. So it's like you can think of the purpose as a longevity, as a health span, as a health thing, almost. I wonder whether or not. I mean, a religious person will listen to you, say that and say, "Well, yeah, we've been saying this for centuries actually, that without purpose, society and human flourishing will lose their way. And AI is just speeding up an inevitable decline that we've seen since the decline of religion, of listlessness and rise in suicide and euthanasia, or the devaluing of meaning and community and reverence and whatever." There was an interesting conversation about exactly this question of what we do if AI puts us all out of work between Sam Harris and Ross D'Alther, the conservative Catholic New York Times columnist, who. And Sam was saying, "Look, I mean, we have an example of a group of people who have found meaning despite never having to work a day in their lives. It's called rich people. It's called the aristocracy. It's called inherited wealth. They're not committing suicide. They're going out fox hunting and they're like doing. There have been rich shitpokes with nothing to do for a long time. So it's not going to be a problem necessarily. We can figure out things to do with ourselves. And Ross was much more skeptical, saying, "I don't think that that comes naturally. You only have to look at people who suddenly find themselves not having to work or being able to work and look at the consequences. It's taken centuries for the aristocracy to learn how to be the aristocracy." Yeah, and that's why they're winning the lottery thing. Any sudden change, it's just going to mess people up. Exactly. And so I wonder whether or not there will be a rise in. I mean, this is my worst case scenario in mysticism and superstition and religion and all of the ways of thinking that try to provide people with meaning. I think we're going to see a massive, surprisingly, conspiracy theory, like we did during COVID, that much bigger when people are anxious and stressed and there's uncertainty. People crave certainty. So they will latch on to any conspiracy theory, no matter if it gives them some level of certainty, they'll latch on to it. So I think we should 100% expect that to happen. I mean, we've been doing some modeling. So we know from the literature that anxiety, depression, substance abuse, suicide, purposelessness, and a few other things, they're not isolated things. There's one goes up, it actually increases all the other ones. So there's kind of negatively feeding to each other and including unemployment. So you can model as unemployment goes up, it triggers this cascade of negatively reinforcing all of those mental health challenges. And they all spiral.
fairly quickly. So we need to start thinking about them all together and not just, you know, unemployment by itself, not just anxiety or AI by itself, but all those things, right? If you have a substance problem, it's, you're not going to get, it's much harder to get a job. And all these things just, yeah, feed in, so you can set up this complex models where you model this out in a population. And you see that, that you don't, you know, if unemployment goes up quite quickly, these things all spiral, because they all push it, they all exaggerate each other. And it doesn't look very good. No, I mean, the other component, the other tricky thing that I worry about in this space is that in order to get a handle on this transition and solve these problems and assist people, you need institutions like government and public health bureaucracies and universities and the media to assist people. But when people are in crisis, they're increasingly untrusting of those institutions and have become consistently less trusting. Sometimes we've good reason over the past decade or so. So people don't trust the media and they don't trust the government, they aren't nearly as much as they used to. So the very tools that you would be using to try to assist people are themselves tainted and corrupted and contaminated by the phenomenon that you're trying to address. Yeah. Yeah. And that's part of why I want governments to start planning this now. We don't want to be stuck in this emergency response COVID style. We've got to just sort of put something in action as fast as possible, because mistakes will be made. And best, yeah, the best choices won't, there won't be that luxury of time. So if we can slow down some of this job loss, this, this AI revolution, the more we can slow it down, the better it will be on the quality of our lives. And I mean, you've written advisory documents for policymakers about what they should be thinking about in addition to a public education campaign, what are the kinds of things that if you are emperor of the world, waving your magic wand now, governments and companies and people would be doing. Well, I start with the national change management plan for each country and make that a priority from the world. What would the plan contain apart from public education campaign? So the education campaign, and I think like this, like I said, whether AI goes bad or it's amazing, either way, we're going to, it's going to be a bumpy ride. And we're going through the next two decades as a period where we're going to need a different set of psychological tools and toolkits to really maintain our psychological health in a way that we haven't been through before. So what do I mean by that? I mean, we need to change our daily habits, right? So habits are very powerful. The atomic Hubbard's book was a crazy bestseller for a reason. And we need to start bringing in some of these habits and practice them so that when the shit hits the fan, we're not just going, oh my god, what can I do? We're in searching and so we already have some of these habits built in. What sort of habits? So a range of things, understanding how to control our physiology so that the anxiety doesn't become overwhelming, understanding how to deal with uncertainty, that's a place that I often start with, you know, that understanding that having a brain like we do as humans means that we find ambiguity and uncertainty about the future like triggering. And so it's not, you know, it's not your fault. It's just a nature having a brain like you. So you can't fight the uncertainty in the world because you'll just burn yourself out. It's like quick sand. You need to accept that and you just work on yourself. So what cognitive reframing is an interesting thing. So there are all kinds of little examples of uncertainty from, you know, horror films, the rollercoaster rides, the fancy restaurants where you'll get 10 courses and have no idea what food the chef's going to give you and people love that and they pay a lot of money for that. So there are these situations where people love uncertainty is very exciting and thrilling. So can you reframe other biggest types of uncertainty to also be exciting and full of opportunity? And that excitement and that stress, fight or flight is somewhat similar physiologically and by practicing that reframing just a little bit each day day in and day out to try and reframe that in your mind as something that is full of opportunity. So we want an additional psychotherapeutic or cognitive kind of, it's not meditative. It's not therapy because I don't think we want to wait for the stuff to get really bad and then go, oh, we need to fix this. We want preemptive things. A mental wellness, kind of, not yeah, yeah. And for our brains and it's going to, right from the exercise and the food and the sleep through to the kind of things we're talking about. And, you know, most people clean their teeth twice a day or more. And we don't sit there thinking, hmm, should I clean my teeth tonight or not? No, but I always end up as suspicious of people who clean their teeth more than twice a day. I know people like that and like, come on. After every meal, I think there are a lot of, give it a rest. There are, yeah, there are two plenty. There are one's probably the truth. Pace and toothbrush companies pushing that. But my point is that there's a lot of power in habits, right? And when I've gone through tough times and my life and stress or anxiety, the habits that I've built or put in, whether it be exercise going, you know, having a sauna or these other things, hmm, I've put in that work over a long time. And so it's a habit. And so that automatic habit to go to the gym will go for a run will pull me out of some of those situations. Yeah, yeah, definitely. So I don't have to think about it. What about things that are more institutional or global scale than, you know, we've got an education campaign, we've got some mechanism by which people can improve their psychological resilience ahead of this. Are there other factors? Oh, yeah. I mean, there's more governmental things like we're talking about, like not just the tax stuff and getting tax from the companies, but I think we should need to bring in different kinds of legislation about how much warning companies give. And if they are going to lay off a lot of people for AI, they need to be transparent and clear about that. I don't think we should ban it, but there needs to be some time given and some transition thing and there needs to be something like so that it's not just as a wild west and companies get to do whatever they want. Because we want to slow this down, give people warning and maybe just unilaterally say that there's like these three software engineers are redundant because their jobs aren't needed anymore. You have to say like this is our AI transition plan and here's how you're going to be. Is this like true? Anything with rules and government, it's not easily done, but it can be done. So things like that. And then for the moment, there's probably some ideas and things about retraining and retooling people to do other things because the humanoid robot, the robotic revolution is behind the AI one. So any dexterous job is still pretty safe for now. I don't think it's safe long term. When you see the cutting edge, the figure robotics is probably the most advanced humanoid robot. What it can do and the dexterity it can do now and it's running fully autonomously, which most of the other humanoids are not, which means no one's controlling it. It's running on its own AI system inside itself and it can navigate for long periods of time now, but hours on end doing things in houses. So if you thought you were going to get out of it by becoming a plumber. Right now. Right. Especially electrician or any, there's a lot of construction, there's a lot of electrical work in all the data centers. That's all booming right now. Energy is the big thing that everyone's, because we need so much energy to drive all these AI centers. So there's a big boom there and there's lots of work, dexterous physical work for now, but I think long term that'll also be on the chopping block, but that's going to be delayed by some much longer period. Yeah. So there's retooling, re-skilling as part of that conversation. I just think that the first step is the government needs to talk about this openly and talk like we're all in this together, reassure people and try and just remove some of that uncertainty. But not talking about it, not mentioning it, hoping it goes away is not a great strategy for any government at the moment. Because the uncertainty anxiety is just building week on week and I see it, just ratcheting up, tick, tick, and on the question of retooling, just because it reminds me of what you were saying earlier about and how every person should become their own, essentially CEO of an AI army, and then do whatever is whatever they like doing that people will pay the army to do, rather than being an employee of a large company. Is the problem basically the pace at which that's going to happen because the AI skeptics, and in fact there was already a piece published in Forbes in response to the piece by Matt Schumer that I addressed at the beginning of the conversation saying this is all a lot of hot gas. Basically AI people always want people to think that AI is uniquely special in some way, whether it's uniquely brilliantly or uniquely devastating because their solution is always you have to buy more AI in order to make sure that you're going to weather the storm, right? And so this guy was saying yes, it's going to be very disruptive. Lots of things have been very disruptive in the past and what will determine the outcome here is how politics and society responds to it, which I refuted because I thought it was a bit silly by saying of course it's the AI people who are warning the alarm because it's the AI people who know what they're talking about. Well this case you're a canary in the coal mine because they've designed the AI to take their jobs first. But also I feel like it's a bit like saying you can't trust a hydrologist to tell you where the water is because it's in the hydrologists interest for you to drill for water. You should use a divining rod instead to figure out where water is because the hydrologist has a vest. Well yeah, sure he does. He's also the expert in where the stuff to rainy and aquifer is. So when he tells you where it is, you should trust him even though he has a vest interested. So we should absolutely listen to AI Daria Amade of Anthropic Claude AI is the one that's been writing the most and talking the most in a way that I think is most useful most interesting sometimes.
a bit scary about this nation of geniuses in a country and the hell. He's just playing the nation of geniuses. So he's written a number of essays and he's made quite scary predictions about a year or two, 20% of the workforce disappearing and this is going to happen. And then he's talked about this idea of the AI agents being just imagine a country somewhere over here and they're all like Nobel laureate level geniuses and that country can be used to all kinds of things and what impact would that have on society just as a way to think about it. And it's not perfect but it's I think it has some impact. So the point I was going to make before is that when it comes to the capabilities of the AI, we should absolutely listen to all the AI scientists even though some of the some of the models that are grown they're not programmed remember they're sort of grown they're self-learning will even surprise the AI technicians a computer science themselves. So there's they get surprised sometimes when it comes to societal disruption and workforce I don't think we should necessarily listen to AI experts because they're not experts in those things. We need a you know a general panel of people that understand like me human the brain and human psychology but we need economists and other people to really start working on this global level. I mean I went to one of those UN things was AI for good in Geneva last year and where every country was there all the the top AI scientists and the godfather of this and that were there all there talking about this representative from each country talking about the sovereign AI they're working on and not a single person talked about job loss, mental health, crisis of purpose it was not touched on it was all AI governance and how do you make an AI model specifically safe in that model. And I was I was genuinely shocked that no one at any governmental level at least at that venue was talking about this because that's kind of the reason I was there I was like what are people doing and what are we thinking anyway so yes we should absolutely listen to them I don't think that they're doing this warning I don't think they're necessarily saying the solution to the AI problems is to go and spend more money on AI they are saying that you need to be aware of what's happening yeah I say you need to be plugged in and I mean the thing that I need to who was criticizing the article isn't seeing is the pace and the scale like in the past when when you outlawed whaling yes there were whalers who needed to go and find other jobs but it was one sector of the economy and it happened over the course of you know what I'm not in the case of whaling but in the case of the industrial revolution generations yeah it didn't you know yes everyone used to work on a farm and then they all had to go and you know there are no farmers anymore ever I'm started working in a factory and now there's no one works in a factory anymore you know so it's just going to be like that everyone's going to be doing different things and but there's always going to be an appetite for more human innovation and therefore you know everyone will find something to do yeah eventually eventually they will but it's never happened this quickly before yeah you never had an everything machine roll out in a matter of months and years you've had big changes that have happened to one sector of the economy you took decades of generations it's never happened this fast and then at on top of that the fact that most people including politicians failed to predict because they can't you can't feel that exponential change like I said our brains work in linear you know every step how we take is another meter and another you know that's a linear change but exponential steps in about 26 steps I've walked around the globe and there are all these people go what what is that right yeah we've gone around the world in 26 steps or the other one you get a paper like you're holding you fold it fold it and 42 folds it goes from the earth to the moon right mistake I'm trying to explain that to my eight year old son a few weeks ago and he doesn't I mean he's just like let's do it let's say he's like no he got an upkin and he was like let's do it bus to think what they try and do it and they get they start off with some you know like 100 by 100 meter to a piece of paper yeah and it does get very very hard to fold but yeah so it's a theory but yes it's theoretical amazing so if we wanted to leave people with one positive piece of advice about what they can do yeah so look I think I'm an optimist despite what I've said a lot is sounds kind of negative but I'm an AI optimist I think we are on the way to an AI utopia life will be amazing we're going to cure diseases we're going to solve any climate problems on planet earth and our lives will be fundamentally better but that is in like a 20 some 20 something year timeline like I said before from here to there is going to be a bumpy road so I see the job of me or people like me is to try and smooth out those bumps as much as possible and yes absolutely listen to to the AI experts pay attention to AI as much as you can don't worry so much about the finances but pay attention to the human condition to the quality of our lives and we need to start figuring out ways of pushing the government to take action on societal disruption onto the mental health part of things and to jobs is probably the core thing that will trigger out a lot of this stuff so lean on people that have power and influence that you start taking this conversation seriously we need to sort of move shift the focus away from the AI and the machines and how much GPUs cost and well can Australia have onshore AI and all this stuff to the societal disruption question but at a human level practice yeah psychological self care enhance your own resilience maybe have a little financial watch as if it's possible yeah there's a lot of content you can hear about this that there is time to start grab and I'm going to ask it for the next 20 years because that work will change fundamentally and so in terms of just equipping yourself to be able to use AI just play around with AI so the try and use as much as possible yeah and a lot of people are some people are opting to you know go off grid live out in I'm just thinking about I'm just thinking about what transition is from a person who is happy being a professional working for a company right now but in five years is probably going to be made have been made redundant and will be expected to be providing value to the economy in some entrepreneurial way yeah it's probably worth playing in the sand pit truth yeah now to to delay losing your job try and use AI as much as possible lean into it be careful with the cognitive outsourcing don't make yourself done by just doing it skinny AI to do all your work for you you need to keep critically thinking and giving yourself juicy hard problems to think about this is another big sort of cognitive problem we're facing at the moment especially with students but yeah AI like I think the article mentions is is like writing a bike or skiing you can't learn and understand what AI is and what it can do by listening to me talk about it here or watching YouTube videos you've got to just play with it an experiment and like the article also says it's changing every two weeks yeah well the last if you try it last year and I was like oh try it now and then just keep playing with it and experiment with it in your private life in a you know for work um and just you know use another table if your company is not letting you use it yet experiment start using it so at least you understand what's happening or what we're going through and then start planning for you know be more entrepreneurial start thinking about the owner of what you know in 10 5 10 years if those jobs are gone what kind of things could you be doing and we don't know how the financial system will shake out but let's think for now that yeah you may have to be the owner of your own project whatever you make and so for a lot of people that mental shift is a big one that mindset to be responsible for my own project my own and see my own thing rather than just turn up at nine in the morning and be what you can't be like it's a very different for a lot of people that's a hard transition to make so start playing around with those those ideas and maybe start a side hustle see what that feels like to be responsible for all parts of it you know you can you can make anything digital now with AI really really easy and then you can make physical products get them shipped from China you there's lots of things you can do now that you couldn't do a year or two ago so such experiment in those things and like you just said yeah also think about the what kind of life you want to lead and what what matters to you and your family and how to safeguard those and hold that quality of life as a North star through the next couple of years and actively do things to start safeguarding that and make that resilient Joel great to talk to you as always come back on the show in 20 years when we're outside out of the Mad Max apocol ebbs and you'll be dressed as like you know the Viking guy on January 6th or something covered in war paint and I'll be making my wearing a bearskin for it'll also be digital my my my AI replica do it for me okay excellent thank you pleasure as always [Music]
Podcast Summary
Key Points:
AI is advancing rapidly, shifting from a tool to autonomous agents capable of complex tasks like software development without constant human oversight.
This progress is expected to significantly disrupt job markets, especially in digital fields, within 1-5 years, causing widespread anxiety and uncertainty.
While AI excels in structured domains like coding, it still struggles with nuanced human tasks such as acting, though creative industries are also being impacted.
Societal adaptation is crucial, requiring a shift in mindset toward entrepreneurship and re-evaluating education and career paths to maintain quality of life amid rapid change.
Summary:
The discussion highlights the accelerating advancement of AI, particularly the transition to autonomous agents that can perform complex tasks like software development independently, as exemplified by models from OpenAI and Anthropic. This evolution is poised to disrupt numerous industries, especially digital and service-oriented jobs, within the next few years, leading to potential job losses and increased anxiety. While AI excels in structured areas such as coding—where it can build and test applications efficiently—it still lacks the nuanced intuition required for creative fields like acting, though AI-generated content is becoming more prevalent.
The societal impact includes uncertainty about education and careers, with a suggested shift toward entrepreneurship and smaller, agile startups. Emphasis is placed on preparing for these changes by focusing on human adaptability and quality of life, rather than resisting technological progress.
FAQs
Drivers who switch and save with Progressive save over $900 on average, based on a national average of $946 in 12-month savings for new customers surveyed between June 2024 and May 2025.
AI agents can now autonomously build, test, and debug complex software over extended periods, such as two weeks, without human intervention, significantly reducing the time and cost compared to human developers.
AI is leading to job consolidation, particularly in entry-level and graduate roles, with unemployment expected to rise as companies replace human workers in fields like customer service, software, and digital content creation.
Software development, customer service, and any digitized field like music, film, and design are seeing rapid AI integration, with AI-generated content becoming common on social media and in entertainment.
Focus on developing entrepreneurial skills and adaptability, as traditional large-company jobs may decline. Consider education but prepare for a shift toward startups, solo entrepreneurship, or roles that leverage human creativity and intuition.
Uncertainty about AI's impact is causing widespread stress and anxiety, triggering a fight-or-flight response in many people, who worry about job security, mortgages, and future opportunities for themselves and their children.
Chat with AI
Loading...
Pro features
Go deeper with this episode
Unlock creator-grade tools that turn any transcript into show notes and subtitle files.
PodcastTranscript AI