Can AI redefine what it means to be human?

Hi everybody. Thank you so much for joining us today. We're excited to welcome you to AI and the Art of Being Human. Today we're joined by Dr. Andrew Maynard, Professor at Arizona State University and author of AI and the Art of Being Human for a conversation on how AI is reshaping how we think, work, create, and connect. The discussion will be moderated by Dr. Mark Esposito, Chief Economist at MicroOne. We're so glad to have you here and are excited to get started. So please feel free to use the Q&A on the bottom navigation bar and Mark will do his best to answer those and ask them throughout the session. I'll pass it off to Mark.

Great. Thanks so much, Sydney, and welcome to everyone who's watching us right live. And Andrew, particularly pleasure to see you again, this time on Zoom. As a context, Andrew and I were colleagues when I was at Arizona State University, and I always been fascinated with the work you do, Andrew. I think you've been pioneering conversation that now are becoming this surging way before anybody could even talk about this. So really happy that we could finish our series with you. So Andrew, I love for you to share a little bit about your exposure to the technology that these days are so, I would say, vocally transforming our life to the point everybody talks about it. I think when we were meeting already before, a lot of your work was already deep into the intersectionality of technology, but you never forgotten, I think, a human dimension around that.

Yeah.

So just, just share with people who don't know your work well how you got to where you are in terms of your thinking.

And I know it's difficult because it's a long, convoluted story, but the short version is I'm a physicist. I trained as a physicist because that was the only thing I could do. I was really, really bad at everything else and spent many years studying airborne particles, including nanometer-sized particles. But that was my entry point into nanotechnology. And so, goodness me, 20, 25 years ago now, I found a flip in my career where I moved from the lab to starting to work with many different stakeholders through the US government on thinking about how you develop nanotechnology at the time for social good while working out where the downsides are and navigating them. And that got me really interested in this intersection between cutting-edge technologies and society and the question of how do you actually steer these technologies to a point where they truly do help us flourish as a species and create better futures while avoiding some really big pitfalls. And then so nanotech was the gateway drug, if you like. But over the last 20-plus years, I've expanded that out to look at the intersection between many, many different technology trends. So I work really at the intersection of technologies that don't even have a name necessarily, and yet are poised to transform society. And of course, over the last 5 to 10 years, artificial intelligence has been front and center of that, however you define AI and whatever type of AI platform you're looking at.

So, Andrew, if, if kind of, if I understand correctly, one, one of the, the purposes you have is really to advance this, this coexistence where technology and, and human development, they kind of, uh, they can flourish together. Uh, maybe, maybe it's a big question to start, uh, early in our webinar. Do you find that we're getting closer to a better configuration of that, or do you find that it would probably have more fragmentation in this very same ambition than when maybe you were prior to this current wave.

It is complicated. So I will tell you my professional intuition response, which is we're getting further away. And I say that because much of my career over the last 20 or so years has been putting humans at the center and asking, what does human flourishing mean? What does it mean to be the best version of yourself in your within the society you're in. And so that has been my driving question. Then the secondary question has been, how does technology enhance or inhibit that? And you can only address that question if you approach it holistically. You don't just sort of say, we're gonna build a technology and assume everything's gonna be great. You've gotta look at that social connection because at the end of the day, how humans behave and think and believe and react changes our technological future. And there, I think we've actually taken a step back. We now see a global society that is so fixated on going as fast as possible with technologies, we tend not to even engage with the question around, is this a good idea or is it a bad idea or can we do it better? So from that perspective, I actually think that we're in a worse place. But there is a caveat. If I step outside my sort of own head and look at the world, it is undeniable that we are developing technologies that improve people's lives. So now you have a paradox. We are doing things that can massively improve the world we live in. At the same time, we're ignoring the things are things that we shouldn't be ignoring in order to make sure that that actually works truly to human benefit?

That's, um, the— I think the fact that you mentioned this, Andrew, as a paradox, I, I think we, we kind of feel it, we can sense it. It may be some of the, the tensions that we're trying to navigate these days. Um, there's also been, I think, with the rise of more dependency on technologies, I think there has been more conversation about AI for goods. I think several tech labs. You know, the, the, the session I facilitated last week was with two fellows which were two professors of University of Washington. They're both fellows at the Microsoft AI Economy Institute. So there's been a lot, I think, of spaces where the interlinkages between research, industry, and, and, and AI as a force for good has, has institutionalized somehow. Do you see this to be on the surface of what we really need, or do you think they are simply kind of pioneering triggers that might scale a much wider distribution of this level of narrative. It's hard for me sometimes to assess them, whether they are virtuous or whether they're just compliant.

And I think it's complex. So one of the places that I start from is the understanding that there are very few people out there who actively want to make the world a worse place. So I work with entrepreneurs and founders and others, and almost to a person, they're motivated to do what they want to do because they want to improve people's lives. Occasionally, I come across somebody who is honest and say, I just want to make a bucketload of money. But most people genuinely want to make the world a better place. So you've got that intrinsic motivation. People are not motivated to make dangerous technologies. They're motivated to make technologies that improve lives. But here's the challenge. It is great if you want to do that. But if you do not understand how to do it, if your view is so narrow that the toolkit that you have is not the toolkit you need, that's when mistakes happen. And I think we're at a point now with very, very fast-moving technologies where there's a disconnect between that desire to build technologies for good and that understanding of how you need to do that, who you need to talk to, the toolkit you need to build, the things you need to be aware of. And I think part of that comes down to a certain degree of technological hubris. I mean, and I know this, I'm a physicist because this is how we're trained. We're trained to believe that we have all the answers, so we don't even need to talk to anybody else. Else. And as a technologist or an engineer, you tend to believe that you know how to fix the world. You don't necessarily need to have that humility to ask somebody else how to do it. And so I think this is part of the challenge we face.

Do you find, do you find that, um, the, the fact that this technology is coming from the AI labs at, you know, allow me the term exponential rate, right? Although, uh, I think it's, is a borrow of, is borrowed sometimes incorrectly. And what organization can do at best is some form of incremental adoption. Do you Do you find that the spread of technology development is what is causing this divide, or do you find that no matter how much we're going to engage, we're looking at really a structural disparity that is difficult to reconcile?

I think partially it's a structural disparity, but the speed of things at the moment is just exacerbating that. I mean, so artificial intelligence is such an interesting example because the speed of development and implementation is just phenomenal. It is the fastest and most transformative technology transition I've seen in my professional lifetime. So that is really revealing the cracks, but the cracks have been there for a long time. And it's very interesting to see if you go back 25 years, 30 years, when people were really focused on nanotechnology as being the technology that was going to transform the world. There was a really deep focus on looking beyond the technology to understanding those social interactions. So the philosophy of what we're doing, the ethics of what we're doing. And that was very much driven by an understanding on the back of genetically modified foods and other gene technologies that if we get it wrong, society is going to reject this technology, so we'll never see its benefits. So with nanotechnology, there was this very deep concern about getting it right by talking to different stakeholder groups. We seem to have lost that now. AI is going so fast that the mantra here is, If we stop to talk to people, it'll slow us down too much and we'll lose the race, we'll lose the imperative, we'll lose the ability. So that is, I think, where these cracks are appearing. And the incentive structure is such that there are more incentives to go fast and possibly break things than there are to go slow and get things right.

If you remember when we started the adoption of the web, the web was equally transforming in the sense that it was not just a technology that was enhancing, but it was an infrastructure structure. Yes. And it took time before I think it, it became part of our sense of normal. Uh, eventually it worked by distributing the access. And I think today, uh, we can generally ask the question, who owns the internet? And, and I think one part, part of the answer will be, well, it's a distributed technology in many ways. It's decentralized, right?

Yeah.

Now, some, some argue that especially with the event of the, the more service-based AI with generative AI and the gen AI, we are in a phase of infrastructural penetration again. So our concerns of now might be perceived with the zeitgeist of now, not necessarily with the fact that this is just a transition. I can see why people will bring the parallelism there. I do sometimes wonder whether we can even contextually compare it. What's your take? Are we in the midst of a transformation that is structural, but we could expect eventually that we'll reach some form of consolidation?

Yeah.

Or do you find that the premise from the web is very different?

I think there are some similarities and some differences. So I think this idea of a move towards distributed technology and that leading to infrastructure, which is net positive, is a really important one. We definitely saw that with the web, the internet, we've seen it with other technologies. And I think we're beginning to see it with AI. So there, I think there are very, very clear parallels. And sort of one of the parallels is if you go back to sort of the dot-com boom, for instance, we saw this sort of roughness of the transition. It sort of went up and down. There were crashes, there were benefits, but things began to even out. And I think that we're on a track to see that with artificial intelligence as well. But here are the differences. One of the differences is you can't always go fast and reach equilibrium in a closed system. And society on this earth is a closed system. And anybody that studies systems engineering and complex systems knows that at some point, there's a transition between a predictable linear phase and a highly unpredictable chaotic phase. So the question is, even if this worked with the web, will it still work with a technology as transformative as AI, even if it follows a similar trajectory? And I think that the answer there is we simply do not know. But this deeply worries me. The second thing which is different is that this AI isn't just infrastructure. AI isn't just something that builds stuff around us that we live in. It is the first technology that gets into our heads, into our minds in a way that nothing else can do, simply because it's designed to mimic and emulate how we think, how we, how we sort of communicate, how we perceive the world, how we build our mindset. And I think that that ability for it to get inside our heads is transforming things in a way that we haven't seen with previous technologies.

Interesting. So it's almost like the promiscuity, right? That will be difficult for us to eventually discern.

Right, right.

To some degree. Yeah. And es, de, fr, pt, it.

No, no, no, no. I think you're right. And what intrigues me, and this is me putting my academic hat on, you can see how this could be disastrous or it could be transformative and just lead to a completely changed and wonderful society. And I'm absolutely fascinated to see where it'll go. The one thing I think that is clear to me is that this is a transition that won't be like previous transitions simply because of human interaction?

Um, so first of all, when you were talking about closed system, you reminded me a lot about the narrative that I think has been, I think, correctly pushed on the planetary boundaries and how any, any kind of aspiration to pursue like an infinite growth kind of clashes against the physics, right, of, of the planet. So I, I think we, we find, um, so like resemblance in that level of conversation. Uh, but specific to, I think, this technologies, the fact that we are kind of moving away from a linear modeling to what will likely look much more entropic in many ways, right? Isn't that nonetheless a sense of evolution? So are we simply evolving from an engineering, I think, scientific, pseudoscientific approach that came from Taylorism into a society that is recognizing nonlinearity, decentralization? So almost like a way of emphasizing complexity as norms rather than, than the problem. So, so do you see this to be somehow an evolution of some sort?

I do actually. I think that's a very good way of putting it. And I would, in my work, I articulate that as saying this is a transition that we simply can't stop. We're absolutely in a transition phase. And if we want to thrive as a species, we've got to learn how to live on the other side of that transition. You can't just sort of stop the transition for many, many different reasons. So if we do end up in a different phase, if we do end up in a different social and technological sort of dynamic. It's up to us to work out how to navigate that, how to thrive in a very different way of seeing the world and interacting with the world.

Interesting. Andrew, I want to start also move our conversation towards the book that you have just released. I think if I understand then the motivation behind, you thought about the book AI and the Art of Being Human. Right. As, as not necessarily a prescription, but as a reflection on maybe the opportunities for us to engage in this roadmap. I wouldn't imagine you writing a book that is about, you know, the, the determinism of failure, right, of the human species. I would imagine the book is providing—

it's a positive space for it.

It's a positive book, right? So, so if I try to understand what was motivating you to write the book when you're coming nonetheless from, from several inflection point that gives you, I think, legitimate preoccupation, right?

Yeah.

What is the story behind the book?

Sure. So first of all, I have to say this was a collaboration between myself and a venture capitalist, Jeff Abbott, which in itself makes it really interesting. So you've got me as an academic coming from a point where I ask questions, and part of that questioning is to be critical about where things are going and ask what-if questions. And then Jeff is somebody who funds AI startups. And yet intriguingly, he also found himself asking questions about where we're going as a society. And when we came together, we realized that there is a real need for thinking about what are we trying to achieve in the future? What does a flourishing future mean? What is the role of humans and our own sort of self-identity in this? And we both realized that whether you are sort of looking more holistically at that connection between society and the future, and technology, or whether you're actively investing in companies that are trying to accelerate this, at the heart of this is people and what it means to actually thrive as an individual, to flourish as humanity. And out of that, there's a very natural question, which is, as we develop these incredibly powerful and transformative technologies, and in the case of AI, a technology which is very much a mirror into our sort of inner self, how do we ensure that we thrive no matter where we're coming from within this new era? So within that, we set out to write a book that would help anybody that picked it up understand from their own personal perspective and background and story, how they can chart their own pathway towards thriving in an era of AI. So it was designed as a very practical book. We started off thinking, this is the book that's needed for the startup community and founders because all over the world we were finding AI founders that are both building AI companies, but also asking these questions. They want to know, what does this mean? What does it mean for me? What does it mean for people that use my product? So they were engaging with these quite sort of deep philosophical questions. And we felt there needs to be something for these. But then we also realized, I talked to my students, they're asking the same questions. My colleagues, they're asking the same questions. My parents and my friends. So it needed to be a book that was very broad. And Jeff, coming from a business world, wanted it to feel like a sort of a management sort of a guide. So he wanted something practical, something with tools in, something with sort of do this now exercises. So what we ended up with is a hybrid book that is very practical. It has 21 tools for thriving in an age of AI that are designed to help individuals sort of do this. But all that is wrapped up in storytelling that helps people connect with these ideas.

Fantastic. I have a— I mean, I remember some time in my own teaching talking about AI as an extension of our brain because I think I want people to feel that we own the conversations that are about AI. They're not necessarily decoupled from us. But I love the fact that you phrase it as a mirror. So if you think of this as a mirror, right, and you think about the fact that a mirror might push a person to some level of deeper introspection in terms of who am I, so identity, Purpose, value. What does the mirror is revealing about us? Right. That I think is one of the things that came to my mind as I was listening to you.

So many things, I think. And of course, many people have used this analogy. We just sort of jumped on the back of a much larger dialogue here. But I find the analogy of a mirror deeply fascinating. So you think of a sort of a physical mirror, just the thing you look into, and think about what it can tell you. It can tell you some really constructive stuff. You can actually sort of understand sort of what is working for you and what is not working just physically and how to adapt that, even if it's sort of what clothes you're putting on or how you're sort of straightening your tie. It gives you very, very practical information, which can both reveal flaws and open up possibilities. So a mirror is very much a two-way thing. It can tell you things you don't want to know, but tell you things that are really helpful. Now you take that analogy and look inside your head. Now, if you think of AI, first of all, and we're talking about generative AI here, large language models, They've been trained on a vast corpus of human experience as expressed in writing. So this is a corpus that reflects us intimately as society and as humans. It gets deep into our psyche. It's not just on the surface. It gets into our psychology, how we think, how we behave, how we believe, what we think of ourselves. So this is a mirror that is able to reflect all of those back to us. So when we look into it, it reflects something which which is deeply personal to us, actually gets to the essence of our self-identity. And in just the same way as a physical mirror tells us sometimes things that we want to see and things that we don't want to see, AI can do the same. But it can also begin to lead us in different directions. So now you think about the physical mirror as something which maybe doesn't reflect reality, but reflects an altered reality. So imagine you're looking in a mirror and what you see facing out of, um, towards you is somebody who looks really ugly and sort of really off-putting. That will change how you think about yourself. It will change your behavior. On the other hand, if you like sort of Snow White, if you look in the mirror and you see an absolutely wonderful, beautiful person, that's going to change how you behave as well. Now extend that to the, the realm of your psyche and imagine what happens when AI reflects these different versions of ourselves and what we could back to us. It has the power to positively sort of change who we are, what we do, how we think, but it also has the power to completely sort of take us in a very dark direction as well. And this is one of the challenges. How do we actually navigate that?

And Andrew, I'm fascinated, fascinated with what you said about the book, uh, together with your colleague and, and Jeff, right? And the fact that it's both— I think there's a point of inquiry as an academic. We want to ask questions, but there's also a practical, uh, side of it. And you mentioned about 21 tools, right? Tools. And I also just collect the fact that you mentioned there's a storytelling behind this. So how would, for example, a story or a tool help us with this conversation about the mirror? And the reason why I'm asking is where we get blurred vision on unequivocal truth, because you could argue the mirror is a reflection also of our darker side, right? Yes. Who help us to navigate that? Because I think what society has to some degree lost is the sense of collective narrative that was defined in a sort of like purpose that was not necessarily defined by me, but it was societally agreed on. So how do you think we can help in this case by providing these tools? What could be an example of that?

So great question. And let me sort of unpack it a little bit to answer it because I don't think it's a clear black and white answer. The first thing is narrative is so important. Storytelling is so important to us individually helping ourselves understand who we are and how we actually fit in with society around us. But it's also important to society as a whole. It's how we find our collective identity. So you start off from that perspective and realize that this is a very powerful means to help us understand the world around us and how we're going to navigate But then you add that to the fact that generative AI at the moment and foundation models are so new that we don't have cases that we can sort of say, we can look at this person and we can see sort of what worked and what didn't work, and we can write that up into a case study. They simply don't exist. Which means that one of the most powerful ways of helping people understand how to navigate this age of AI is to tell fictional stories very much like we've done in the past with myths and fables, but stories where they can see themselves in those stories. They can say, that person has experienced the same sort of opportunities and challenges and feelings that I am. I understand that. And through watching the story, I can see sort of how they moved forward. So this is the approach we took. We, everything is based on fictional stories. And we actually worked very closely and very intentionally with AI to develop these. But out of those stories, which are all based on very specific characters facing very specific challenges, we build out these tools, which in the character's perspective, help them actually navigate the challenge they were facing, but in a way that anybody that's reading this can actually see that working out for themselves. So as an example of that, the book opens with a story around the idea of AI as a mirror. And it's a founder that suddenly realizes the AI that she's working with seems to know her better than she knows herself, and to a quite startling degree. And it first of all takes her off guard thinking, so sort of as my personality, as my identity leaking into the world. But then she begins to understand how she can actually lean into that and use this, use that mirror to actually enhance what she is doing, find her purpose with the company that she's building, and have the impact that she's looking for. And it's one example out of many where we then begin to build a tool. In this case, it's just sort of the framing of AI as a mirror and how you actually use that as a concept to make decisions.

It's fascinating. Well, first of all, I think, you know, considering that this is not far from the holidays, I think it just will be a great gift, right, to be honest. And I don't say this because—

Well, I think it's a great read. It's funny, the story aspect, we've had some pushback. I was saying before we started, we launched this at a business school in Lisbon. And of course, a business school, very formal sort of pedagogy, methodology, sort of case study and the person who was moderating said, I picked this up and I started reading this story. I thought, what is this? Are we going to have to cancel this event? Because this is not the sort of thing that I want to be talking about in a business school. And then he said he read a few more pages and he got it. He understood it. But it is an unusual way of approaching things.

[Speaker:GUY_SPIER] Absolutely. I know there will be questions coming already. Some are piling up. I might try to maybe use some this time between the two of us for a few more minutes and then we can go to the questions. Sometime back, and I change a little bit on the focus to something else, right? The World Economic Forum was featuring you for having been a professor that was integrating artificial intelligence in your line of work in the university, right? I do remember when I think you were pivoting with ChatGPT early on when there were even universities considering going back to paper, you were actually going ahead and trying this out. So, so tell us why the forum kind of show like your work, what they're trying to represent with your specific work that, that you have been doing around this. And the reason why I'm asking is maybe to get to a more personal level on how you've been using this technology yourself as an educator, just, just for people to, to, to also learn that there is a space that we can carve for ourselves where this doesn't have to become an antagonizing technology, but the technology of some form of expansion or elevation. Right.

Yeah. Which is really important. And I'll say upfront that I'm conflicted here. I use artificial intelligence a lot. At the same time, I'm increasingly aware of sort of where the limitations and the dangers are. But with teaching, this actually goes back a couple of years now, leading up to when we created this video with the World Economic Forum. And I've worked with WEF for many years. So that's partially the segue into this. But as ChatGPT was released back in 2022, I realized very, very quickly that our students needed to be comfortable with this new emerging technology. It was very clear from the get-go that this was going to be a game changer, that whether you decided to use it or not use it professionally, you had to be aware of how to use it and how to extract value from it. And at the same time, I realized we weren't teaching our students any of this, that there was no course that students could take that gave them a nuanced understanding of how to leverage the emerging capabilities around large language models. So I set out to create a very, very simple course on prompt engineering. It was a one-credit course, an online course, and it was designed over 6 modules to take students who knew nothing about this technology to the point where they could make smart decisions on how they used it. And it wasn't what you consider to be a normal sort of progression. So I actually started off with a number of exercises that challenged students to think about what could possibly go wrong with the technology. Effectively, I broke their trust in it to start with. But from that point, that point of realization about what the limitations were, we were able to build up an understanding of how to use it in very sophisticated ways. And so this is very much leaning into it. In a spirit of exploration as well. So the whole course was based on exercises that they carried out in ChatGPT. Okay. And so everything they did, they had to be engaging in a deep and sophisticated way with the technology. And alongside that, they were encouraged to think about how they would use this in their personal professional lives. So that's really sort of what the World Economic Forum focused on, and this idea of this is a technology that's not going away. People have to know how to leverage it and extract value from it. So we can't afford to shy away from it. We've got to give people those complex and sophisticated skills. I would say I taught it twice and then I stopped. And I stopped because it was apparent that the technology is moving so fast that the course could never keep up with it. So my advice now is always if you want to learn how to maximize your use of these tools, sit down and use the tool as your tutor. And one of this still blows my mind, you take something like ChatGPT or Anthropic's Claude or Gemini or anything, you can sit down and open a window and say, I know nothing, please teach me. And it will teach you. And I still can't get over the fact that it is a really good teacher.

Well, Andrew, about this, which is fascinating, right? There is, I think, not too old papers on the GenAI wall that I think it was a, it was a field experiment that was trying to say, well, if you have too much distance with the, with your field of expertise, the model will not be able to, to reconcile the distance. So there was the idea that proximity matters within, I think, the ability that we have to build capabilities. And this is really where I'd like you to share your thoughts on this. Do you think that the story should be about capabilities like the entire Jaguar Frontier story that has been going around for a while? Or do you think that it's more about responsibility rather than capability. And, and I'm asking because I am myself conflicted sometimes where, uh, myself coming from economics talks about capability because it's something that we can functionally represent. Mm-hmm. Um, but then there is a side of the story to say, well, maybe it's not capabilities where I should really place my bet.

Yep. I know. I, I think that's exactly right. And I, I think it's something of a hybrid, but responsibility comes first. And, and so the, the, the, this idea area of expertise is a very simple one where large language models tell you what is beautiful, not necessarily what is truthful, which means that they are compelling, they are persuasive, they are very easy to believe. Now, if you're working with them in a domain that you know a lot about, you can bring in your critical reasoning and you can say, that's right, that's not right, I need to tweak this. If it's an area you know nothing about, you do not have that ability. So when you see something that's beautiful and persuasive, cohesive, you don't know where to challenge it or how to challenge it. And that requires a certain degree of both responsibility to realize you have to bring some critical thinking to the table, but also a set of tools that allow you to navigate that. And I must confess, this is something that worries me a lot. And I've actually started talking about this in terms of the possibility of AI being a cognitive Trojan horse. Not because I think that the technology has got unbelievable potential, but what we're effectively doing is if we're working with it in an area that we know very little, we're actually inviting it into our heads and it's telling us beautiful stuff. And we simply do not have the time or energy to challenge everything. So we end up just believing the beautiful stuff. And at what point does that change the way we think in ways that are not helpful? And I have no answer there. But I think it's one of those things which puts the onus on responsibility rather than expertise.

And the capabilities. Yes. Andrew, look, this is fascinating. I mean, it's mind-blowing. I think that we could You speak for 35 minutes and it, it felt like just, uh, uh, the first sip of coffee, right, that we had. Look, there's a few questions that came through and there will be more coming, right? I think it's fair that, you know, we've been having more than 100 people connected all the time, uh, during the conversation. So I'm extremely humbled by the fact that people decide to spend their time with us. Look, there's a question and Sydney helped me to, to paraphrase it from Ben, um, where I think he's talking about how are we gonna really build trust, um, in an accessible or equitable AI, especially in time where everything seems to be decentralized. On the contrary, we, we see conglomerates of power that are equally coupled by geopolitical interests. Uh, we see data center being largely driven by, by US interests and not distributed as much. Um, and we only have a handful of AI labs capable to really deploy this technology. So how do you reconcile access, fairness, trust, uh, decentralization when somehow But now the leverage seems to be showing in the other direction.

So I did— this is very facetious. My sort of cynical quip was going to be, we're not going to have to worry about it because AI is going to persuade us to trust it no matter how much we resist. But if I was going to be serious, I— so I think there is a very, very serious challenge there. And the challenge is both where the power lies and where the political motivation lies. And that gets to the point where where trust becomes a very, very serious issue. But I think there is a little glimmer of light as well, and it does come down to the decentralization. So yes, the, the model training and model development at the moment lies in the hands of relatively few companies. But what is coming out of that is open-source models and even sort of small stripped-down models which can be independently developed and deployed. So we've now got the ability for individual communities around the around the world with relatively low levels of expertise and equipment to build AI models that suit their purposes and that are trained to reflect and respond to their worldviews and their beliefs and their environment. So that in itself, if we actually get to that point where you have this proliferation of, of small localized models, that is one way to begin to get to that point where people trust the models that they're using. But it still doesn't solve the problem of everybody using something really big from a major company where there may be sort of alternative motives behind how these models behave. Or even if there aren't alternative motives, the fact that these models are not perfect, they're not truth machines, they're beauty machines. And there I think we've gotta think very differently about the nature of those relationships we have with the technology. The trust question, which is such an important question I don't think there are any clear ways forward yet. There are plenty of people that feel as if they do have answers about around how you build trustworthy AI. I think we've gotta do a lot more work here and maybe even reframe the question and the conversation around the nature of trust and what we are truly trying to achieve.

Yeah, great. Look, there's a question from Pam that I'll try to massage a little bit to fit what you just shared now, right? So is there any example where you think that by using AI, we truly scale in, in, in a net positive so that that is becoming a clear victory where we can say, look, not necessarily that we can replicate this because I think we do have replication dilemmas in a lot of this conversation, but something that we can say in this specific field, we have been able to really overcome the tension and we have been able to use the technology for advancement of the human species. Is there anything that sometimes you use as a way to showcase what we have done that can inspire other fields to eventually follow?

So I think it's too early for clear case studies, and everything is nuanced and double-edged. But if you look at 2 or 3 areas, I think we're finding really interesting sort of breakthroughs. And this one I'll start with is learning and education. Highly complex area with a lot of stuff that's unresolved, but the bare bones facts are are that if you want to learn about something that you know nothing about at the moment, you can get an overview, a high-level understanding within minutes or hours using AI that you could never previously achieve. Which means that we're seeing a democratization of access to knowledge and understanding, flawed as it is, which far surpasses anything we've experienced before. And I'm sure that if you were just to stop people in the streets and talk to them about this, they would come up with examples of this has worked. I do it all the time. I can be in a meeting where we're talking about something, sort of a committee looking at sort of the next big technology coming out. And I can be putting questions into AI and have a far greater understanding of the field within a matter of minutes than I could ever get out of that committee of experts. So that, it's unbelievably transformative. It comes with incredible dangers, but I think it's absolutely transformative in learning and education.

Another area— Oh, please, please, Matthew. No, go ahead.

No, no, I was gonna say, I mean, and there are 2 or 3 areas like this. The other one I would put in actually out of many is actually research and discovery. Highly contentious at the moment because there are people that will claim that nothing original can come out of AI. But what we're discovering is the combination of artificial intelligence and human ingenuity is a matchmaker. In heaven because the AI can create connections, serendipitous connections between bodies of knowledge and understanding that a human never could because we just can't hold all that in our heads. So it can throw out possibilities. And then if a human can weed through those and pick out the possibilities that are truly transformative, that then accelerates research and discovery. We're already seeing this. So that, I think, is an absolute game changer.

That's great, because I think there was a question from Gage, and I hope I pronounced it incorrectly, about whether the dependency on statistical model is going to encourage conformity and will eventually suppress creativity and divergent thinking. And I think it's great because I think if you observe it just in terms of the function of the model, there is a statistical model that generates the inference, but is the— in the combinatorial power, I think that we can create divergence, right? Which is exactly what you said, right?

And of course it comes down to responsibility as well. Yes, absolutely. These statistical models could lead to a convergence and a homogenization of ideas. But used rightly, they open up possibilities amazingly. And I think it's a case of working out how we use these in transformative ways rather than just regress to the mean use.

Yeah, I have a question. I'll try to kind of push a bit Alexander's question, considering some of this part already being covered by you, Andrew. And I'll try to phrase it like this. We could argue, and I think agree, that knowledge has become very cheap, to some degree very affordable. So in a period of our time, in our life, where knowledge is becoming inexpensive, should we shift to imagination? Should we spend more time in imagining our future? And I'm asking this because I find that in the past, imagination was almost like esoteric. It wasn't taken seriously. But we also seen the country that were able to imagine a future that was not statistical, Mexico, our country, they have been able to change, leapfrog in their evolution. So how do you see AI helping us with this? And not necessarily in terms of AI is too early, but maybe back to the conversation of the mirror, that introspection might help us to reconcile roles or things that we should be doing more because other things have been intermediated to the technology. Yeah.

I think that that's a fantastic question and perspective. And it gets right at the heart of what it means to be human. So just as an example of that, going back to the book and the tools, one of the tools we have is very simply called the human capability spectrum. And it looks at a spectrum of sort of where humans bring something unique to the table versus AI. And on the left, it has the things that you do that an AI can do. And that's quite a lot of stuff. It's a lot of transactional stuff. It's a lot of mechanical stuff. In the middle, you've got relationships, the things that you do that depend on human-to-human relationships. And then at the far right, it's got the transcendent stuff, the stuff that you do, the stuff that defines you that is uniquely human just because it's embedded in that nature of being human. And the point of the tool is working out sort of where the things that you do lie on there so you can put energy and effort into the things that really define you and let the machines do what the machines are good at. And I would say in that transcendent category comes imagination. Um, it is one of the, the superpowers of humanity, is one of the things that has enabled us to get to where we are and enables us to move forward. That ability to both imagine different futures, um, but for us individually as well as society and our community, but also plan a way of getting there. And that is something that a machine can't do. A machine can reflect what has happened in the past and our thinking in the past, and it can catalyze our imagination, but it can't do the heavy lift of the imagination for us. That is something uniquely human because it is our future. So I think that if you extrapolate from there this idea of focusing on imagination and creativity, I think there are incredible possibilities there and imposibilities. That say that we can enhance that, we can augment it with AI if we use it smartly.

Yeah, you know, Andrew, I was watching a series produced by BBC called Humans, and at one point in time, as they were looking at humans' remainings and understanding how some fossils were designed to be stone that were basically crushed on purpose because of rituals, is the fact that humans could imagine something that they can't see, right? This abstraction that we have. And I think it's interesting that after, I think, decades and hundreds of years of industrialization where we were becoming the intermediaries of the functions, we can now intermediate this to the technology where we can go back to basically the abstraction of ideas, if I understand your spirit, right, as well correctly.

Yeah. And it's absolutely a yes and. So I'll just— this will be very brief. I'll tell you one of the stories that came out of the book. And this is an interesting story because this story was actually created via Anthropics Claude. So the story was AI-generated, but I think it was profound. And it's a story about, totally made up, an Amsterdam painter who puts a series of paintings into an exhibition and it doesn't win a prize. The thing that wins a prize is a piece that was generated by artificial intelligence, technically beautiful. And the whole story is about this person sort of introspectively examining what this means to them. And he has an old student to come along says pretty much, you must be bummed because an AI beat you. And he said, no, I'm not. He said, I look at that and it's technically better than anything I could make. And he goes back and looks at his paintings and realizes that they're technically competent, but they've got no soul. And so he decides there and then he, he needs to go back to the drawing board and work out what he's trying to do. And the next day he clears his schedule, goes into his studio, gets his easel out and his paints out, puts a silk blindfold on, um, and starts painting blindfolded. And takes the blindfold off, and what he has in front of him is a mess. Total mess. But he realizes it's a human mess, and it's something that a machine would never do. Not because it can't, because it wouldn't see the point in it. And from there, he begins this whole practice of painting blindfolded because it allows him to express something that is innately human inside of him, which a machine wouldn't do because it wouldn't see the point in doing that. And it gets to this idea of the machine stimulating a new way of seeing the world and imagining the world and being creative, not to replace the machine, but stimulated by the machine.

Wow.

And the fact that this was actually written by an AI, I still can't believe it.

That's exactly, I think the part that is by itself, right, worth the mention. Look, there, and I'm aware of time, I'm going to try to wrap it up. There's two questions I'll try to merge together, one from somebody that is anonymous, another one from Sanjay. And I'll phrase it like this. A lot of, I think, the preoccupation where people are working, like in your case in the US, like in my case in the US, now in the UAE, it's like we are exposed to the access to the technology. And therefore, the follow-up to that is, how do I use it in the right way? And therefore, a lot of this conversation are coming from a point of permissibility. Ability of the access that we have. But there is, there is a data, uh, equity problem in the Global South. There is an increasing cost of energy, um, and there are many places where AI is not perceived to be a reconciliatory technology, and the country is exacerbating more. Uh, do you see this to be just a matter of, of time before we can see this eventually more evenly distributed, or do you find that we're really in the presence of almost like a chronic disparity that cannot easily be addressed because of the concentration of capital that are not easy to replicate in the Global South.

I think there are huge challenges here, and I think there are questions where there are easy answers, but I don't think they're the right answers. So for instance, take the question of equity. Absolutely right that there is a concentration of power and ability in the Global North. The critical question is, why does that matter? How does it affect things? Because it isn't simply a case that if somebody doesn't have it, they are disadvantaged, and if they do have it, they're advantaged. It's more nuanced and complex than that. And I think we have to engage with that and we have to work out sort of where the true challenges are and how you actually even them out. But more than anything, I think that the complexity and the nuance of the environment at the moment is we need open conversation. Conversations. People need to be free to have these conversations that are focused on finding pathways forward where they are given permission to ask the difficult questions. The people that think that we should be rejecting AI, they should be as much as a part of the conversation as the people that say we should be going as fast as possible. The people that say they are deeply worried about the energy and the environmental impacts of this technology need to be as much of the conversation as those people that say AI is gonna solve that in the future, so we gotta go fast. And so on and so forth. The pathways forward are so complex and convoluted that we have to open up the ability for everybody with a stake in the game to be at the table. And that includes everybody from sort of people in the Global South to the Global North, to even within somewhere like the States, communities that really don't have access and understanding to those that have access. And so, and this is a huge challenge because this is not what's happening. And one of the things I would say here, and this is a little personal from my perspective, one of the few places that this sort of dialogue can be stimulated and happen is at a university where we do bring together different disciplines, different stakeholders, different missions, and we can create those safe environments. Now, whether that is happening or not is a different question. I don't think it's happening anywhere near the scale that it needs to happen. But this is one of the things that I think is essential for us to be able to move through this. I don't think we'll get to a better future just by accident here.

Absolutely, Andrew. Look, I think while you were talking, I might ask a question, but I think, I'm not sure, Professor Andrew has kind of covered that in the participation required, where it's also about defining that engagement in a way that is creating equity by the opportunity for specific voices to be heard across the spectrum. And then I appreciate that. Look, we started, Andrew, with Ben, who asked our first question, and he kind of asked a question that is perfect for me to close. And he basically said, Professor Andrew, did you use AI to write the book? If so, how and where? And in the ideation of the book, proofreading, structuring, finding evidence? I think it's a fascinating question because we would never be asked this question, Andrew, 5 years ago, right? I know, I know.

And so the answer is a little embarrassing, and it's a little embarrassing because part of my craft as an academic is a writer. I consider my skills a as a writer, very, very precious. And I'm writing increasingly these days about how I don't use artificial intelligence in that. But this book was written very, very heavily with AI. So I can tell you very briefly the process, and it was a very detailed process. So we started off with Geoff, my co-author, and I spending a lot of time with OpenAI's ChatGPT, having conversations about what we believed, what we felt, what we wanted to achieve. That led to hundreds pages of conversations. We used those again, working with AI to come up with a plan for the book and a structure for the book. We then switched over to Anthropic's Claude and came up with a very, very detailed workflow for how we would actually work with AI. This took a couple of months to sort of come up with this workflow. Then once we got it, every single chapter, the first draft of it was generated by Claude. And we chose Claude very, very specifically because it had a writing style and a cadence and a nuance. We actually trained it on our own voices, but it had that sort of style that deeply resonated with us. So yes, every single aspect of the book was AI-generated apart from two things. And the two things that were not AI-generated were the COVID of the book. I was very adamant about this. I needed a cover illustration that had nothing to do with artificial intelligence, that reflected the challenge you what we did. And then we have the final bit of the book, the afterword, the personal bit where we talk about why we did it and the process. I was insistent that we wrote that as well without the aid of AI.

Wow, fantastic. Andrew, look, I'll bring it to a close. Your question keeps on coming, but I think we have achieved a lot in the time that we had. I guess more people, and I saw the message from Sylvie, who was a friend that felt passionate about your work, and she's adding your book to her list of the holiday readings.

Oh, great.

So first of all, Andrew, thanks so much for having spent time with me. There is nothing like really engaging in an open and genuine conversation about where we are. And people like you are the kind of voices that help us to navigate what sometimes feels to be complex and not easy necessarily to know what's the first step of the degree of inquiry that we all need to eventually own to some degree. Second, look, I think AI and the Art of Being Human is at the center of what we should be definitely engaging more at university communities and organizations. So I invite all of you to really read Andrew's work. And finally, I appreciate the friendship with you, Andrew. I'm happy that when we met at ASU, it was the beginning of an opportunity to keep on following each other's work. And extremely proud that we could do this together as our final series prior to the holidays. Holidays. I wish you great time as you're— we're trying to decompress a little bit more towards the end of the year. And I know January, the madness starts again. But thank you again for being here with us.

This has been an absolute pleasure. Thank you, Mark.

All the best. Bye, Andrew. Bye-bye.

Bye-bye.