Latest / Stand In The Circle / Stand in the Circle 6.1: AI and Connectedness to self, community and nature with Nathalie and Carl
Transcript
- speaker-0: Biases are deeply entrenched in all of the data that has been scraped from the internet. So if you've got data which is misogynistic, it's going to come out in the final product. speaker-1: We have been losing our senses, we've lost our abilities in so many areas they've atrophied because of mobile phones. Losing our peripheral vision. The love of your life walks past but you don't notice because you're on your phone. speaker-0: People think, AI is going to take over the world. I think the danger lies elsewhere. And it's with poor design and not considering the values that support the flourishing of life. speaker-1: Technology should develop latent human abilities rather than make them redundant. AI should function as a scaffold for this human capacity, not as a substitute. speaker-0: Welcome to Stand in the Circle with me, Dr. Rosalind Watts. I'm a clinical psychologist, researcher, and founder of Acer Integration. This podcast is about how we get connected in a culture that pulls us apart. We talk about the hardest experiences of being human and the circles of care we have needed, couldn't find, and are searching for now with stories from people who've lived it and researchers and practitioners. Healing and community is the old way. The work of our time is to rebuild our connectedness to ourselves, to each other, and to the living world. This month, our terrain is artificial intelligence, AI. And so why are we talking about AI on a podcast about connectedness? Because AI is now part of how we connect and also how we disconnect. Millions of people are turning to chatbots for comfort and advice and company. And this technology is largely being built by a very tiny number of very rich men. Their story is that it's inevitable that it will save us or destroy us and that either way, we don't get to say. This week I enjoyed reading the journalist Carol Codwallader's sub stack. She is amazing. She exposed the Cambridge Analytica scandal and in her sub stack, which is called How to Survive the Brolligarchy, she says that Elon Musk's recent decision to wind down making his cyber truck is a good symbol. of the times we are in. So the cyber truck was a sci-fi fantasy car, a symbol of a particular kind of future that I think most of us don't want. We can see this future as the ego project of a tiny group of people who we don't identify with. But it seemed like because of their immense wealth, we were powerless to stop that nonsense. But the cyber truck, this symbol of that future, didn't actually work. It was recalled several times. And now it's being discontinued. And Carol asks whether the AI story is the same, a dazzling vision of the future, but that doesn't actually fully work. So this episode, I'm talking to Carl H. Smith and Nathalie Nahai. But before we get to that conversation, I'm including this introduction because last week, something major shifted. In the US midterms, people voted and they voted for something different. Black women and black men were the backbone of that change. More than nine in 10 black women voted Democrats. So did more than eight in 10 black men. They made this happen. But also the young men are increasingly getting on board. So two years ago, most young men in the US voted for Trump, but this time a majority of men under 30 voted blue. So many of us all over the world woke up to a new feeling of hope. There was a blue wave and it was also a wave of women. A record number of women will serve in the House. Nine new women governors means a record 17, a third of all states. Georgia elected Keisha Lance Bottoms, the first black woman governor in US history. New Mexico elected Deb Harlan, the first Native American woman governor. Minnesota sent Penny Flanagan to the Senate, the first Native American woman ever elected there. Rashida Tlaib, the first Palestinian American woman in Congress, was reelected. And Aba Kowas, the daughter of Palestinian refugees, won a seat in the New York State Senate. And because this is an episode about AI, I have to tell you this too, the data centers, the vast buildings that power AI were on local ballots 41 times. And 40 times out of those 41, voters rejected them or put limits on them. And this is a joy to me. I've been emailing my local MP and joining campaigns to stop a huge data center proposed for the Somerset levels near where I live, which would destroy so much precious wildlife. And now I have hope if friends in the US can bring in What I see is sensible limits than we can too. And there's actually so much more to celebrate about the midterms. And there's actually so much more to celebrate about the midterms. Already cities like Houston are pulling back from cooperating with ICE. Two doctors, Abdul El-Sayed and the pediatrician, Annie Andrews, are heading to the Senate. And maybe that will bring change to America's healthcare affordability crisis. It feels like America voted for a return to a more caring society. In Texas, James Talarico won his Senate seat saying, that a politics of division cannot compete with a politics of love. And now they all have to deliver, but getting the seats at the table was the first step and it happened. But it wasn't inevitable. And I think for me, that's the lesson. Nothing is inevitable. The future isn't decided in Silicon Valley. It is decided by us together. It's decided in communities and by communities, in circles and also at the ballot box. So this month, we're asking the questions. How does AI impact on our connectedness to self, community and nature? And how can we keep the ways it connects us and let go of the ways it pulls us apart? How can we decide what we want it to be for the good of all? How can we put limits in where limits are needed? And how can we use this powerful tool to strengthen and support the flourishing of life on this planet rather than undermining and destroying it? My amazing guests today to have this conversation about AI and connectedness are Karl Hayden Smith and Nathalie Nahai. They are both polymaths and geniuses and also warm-hearted carriers of the flame of connection. Nathalie is an author, speaker and artist working at the intersection of persuasive technology, psychology and AI ethics. Karl is an associate professor of media at the University of East London, founder of the Museum of Consciousness and co-founder of the CYPODELIC Nexus. And he spent over a decade exploring consciousness through psychedelic sound and immersive technology. And then my next episode will be my own reflections on AI and connectedness. And I'll be joined by members of the ASA community. So now over to this conversation with Carl and Natalie and there's amazing resources in the show notes, particularly something that Natalie has sent over about organizations that are regulating AI. So I hope you enjoy the conversation. So welcome Natalie and Carl. If you describe yourself and then I'll ask the first orienting question and then we'll just see where we go. So I do many things, exploring how persuasive technologies interact with human psychology and business and society. So I've written several books around psychology, business and persuasive tech. give keynotes and I advise companies around that. So what do do when it comes to agency, self-determination, meaning-making, critical thinking, and especially in an age of AI where large language models are increasingly persuasive and making a claim on our intimacy and our intellects. What does it mean to be human in an age of AI? I'm a realist painter, currently working towards my first solo exhibition, which will be in London in March. And I'm a folk musician and I have been since I was very little. And I'm currently working on my third album. And I'm also seeking some way to synthesize all three of those. All of them are held within a deep sense of questioning how we can be. stewards to life, how we can create more beauty and greater harmony and greater vibrancy in the web of life rather than decimating it. Thank you, Natalie. Carl. speaker-1: Wow. It's fantastic to learn more. I am Carl Hayden Smith. I am a associate professor of media at the University of East London. That's my day job. But like Natalie, I do a lot of different projects. And I guess my next biggest project is the Museum of Consciousness, which is based at Oxford University. And that's really one of my main passions is sound and and functional sound especially. So think mantras or ikeros. How do we Fully change consciousness just with sound without the need for anything else, because I think sound is primary and one of the best art forms for immediate state shifting, context engineering. And also I run the cyberdelic Nexus, which is looking at how to change consciousness with any form of technology. So that's been running for 10 years. And also I do a lot of psychedelic research starting at Imperial. So I did eight years as a participant. Now I'm a director at New Nautics and I'm head of context engineering at Eleusis, where we have a license to administer DMTX experiences legally in nature. So very much looking at the context of the results of these trials, because I think a lot of the data is fundamentally skewed because we do these trials in hospitals. We associate hospitals with death and dying and disease. So what do you expect when the results aren't optimal? even though they are still pretty amazing, I think they can be improved. So my general way of looking at things is to explore consciousness. So I guess I'm really exploring consciousness science through different modalities and methods, whether it's psychedelics, cyberelics, sound, and obviously now looking heavily in my day job at AI and how it's changing our identity, our sense of community and what it means to be human. speaker-0: Thank you, Carl. Well, I can't imagine two more fascinating polymaths to be having this conversation with. I love the interplay of all of the things that you're both being and working within and holding and creating. And yeah, there's this big question that's weighing heavily on me. And I'm so grateful to you both for spending this time to be here with me in this question, which is AI and connectedness. Connected as to self connected as the community connected as to nature. You both touched on it in different ways, new introductions. Is AI good or bad for those things? I have this horrible doom sense that the answer is that AI is very bad for those things. And I know that some people are banning AI talking about not using AI when they are speaking with their audiences. What's the right thing to do? If you care about connectedness, you care about. people being connected to themselves, to each other, to the living world. What should we do when it comes to AI and should we be worrying? I think the first thing that's helpful to set out at the very beginning is definition of terms. So when we're talking about artificial intelligence, this is a broad suitcase term under which sit lots of different types of tools, if you like, or programs. So you've got things like the large language models that we refer to as The bots chat, GBT, Claude, deep seat, perplexity. And then you've got things that are more around text, video and image generation, which are the large vision models. You've got real world models. You've got hybrid models. You've got things like neuro-symbolic AI, but most of the time when you read in the headlines, AI, people are talking about large language models specifically. So those two terms can get muddied and one is not the same. AI is the broad category. Large language models sit as one discrete element within that. So when we're talking about AI, most people are thinking about large language models and there are certain issues with the current design of large language models that I think are more troubling than others. The ones that come up when I speak to people, certainly in business, but also within media and beyond. One of them is that biases are deeply entrenched in all of the data that has been scraped from HTML pages, from the internet, from Reddit, from websites, Wikipedia, et cetera. And those biases make it into the data sets from which these systems are trained. So if you've got data, which is tricky and biased and misogynistic, it's going to come out in the final product. So the regurgitation matters. What you're putting in influences what comes out. So that's one issue with large language models. Another issue is the way in which the interfaces themselves are designed. So if you're talking about chatbots like Claude or ChatGBT, certainly with the anthropomorphized chatbots, Claude has been given a human name, right? And we're so used to conversing with other people using text and voice that if we are accustomed to these modes of communication through other humans, and then we have a user interface which is designed to maximize engagement and time on site, it wants to keep you there as long as possible, it's going to be tuned for increased sycophancy to flatter you more. It's going to be tuned to agree with what you have to say, so this positivity bias. And so it's not going to... support, unless you design it differently and you can, and this is a crucial element here, as they currently stand, the most popular ones do not support greater self-awareness, critical thinking, Socratic methods of questioning. They tend to entrench our existing perspectives, biases and beliefs. And so there are ways of designing these things differently. And we need to be more conscious about how we use them and also to understand about errors where Content is fabricated in a confident, plausible, coherent structure, which then causes the person, typically, if they're under pressure and stress, many of us are, to then not consider with scrutiny what's coming out. And there are loads of studies that look at this by MIT, and there's various other organizations, universities and otherwise. So we risk those sorts of issues. And when these sorts of hallucinations, in inverted commas, are mathematically inevitable, which it appears that they are, You can't just iron them out of the system like a bug that can be inoculated. So the question then remains, well, if hallucinations are mathematically inevitable and if biases exist in the data sets, how can we mitigate some of these risks so that people aren't seduced into thinking that A, they don't have to do any thinking, B, they're going to get praised all the time and that biases are okay and that gets entrenched and then it pollutes the entire wider public discourse. And then that's not even to touch on things like large language model. misinformation and data set poisoning that you see from outside, potentially nation states, but bad actors. So there's lots of those issues that don't get talked about as much. People think, AI is going to take over the world. I don't think that is where the danger lies. I think the danger lies elsewhere. And it was with poor design without understanding what lies beneath the hood and not considering the values that support the flourishing of life and of robust challenge within society. So we can actually have conversations that matter. Thank you, Natalie. call. speaker-1: Yeah, so we're talking about connection to self, connection to community and connection to nature. And this is partly why Josh and I started the Center for Human Agency, because we had our first meeting with an incredible group and the question of agency as as the term, as the name of the center came up. And it was really a case of is agency the right thing to focus on? And one person in the room said, why why would we bother with agency? We don't have any. And I was like, Yeah, that's exactly the point. Right. So it's like w I think it is a great focus. And the more we talk and the more we have these meetings, the more it becomes a real focus because it's not just human agency though. I think there's there's agency of nature, which is directly impacted by AI, and there's also agency of AI. Because a lot of people are beginning to believe these things are conscious. They certainly are having relationships with them. And there's lots of Issues clearly around that, especially when China recently banned AI relationships. So people woke up to having their relationships deleted. And not just the relationships, but the logs. So your history with your person was gone because they're already seeing a direct impact on reproduction. Because suddenly you're not spending any time with your partner, you're spending a all your days with your AI. And of course you can see why. Because of the sicker fancy, because it has all of human knowledge behind it. And then you just look at your human partner and think, Who well, who are you? What do you offer me that I can't get from and let's not even get go there with once these things become embodied through robotics. Even anthropic is saying that these models feel pain. And one of my colleagues, David Hansom, he created Sophia the robot. Do you remember Sophia? Mm-hmm. So Sophia came out. He was dealing with a lot of press all the time. And at one point, I think it was somewhere somewhere in the Middle East, they basically gave Sophia human rights in a place where women didn't have human rights. So suddenly this robot is elevated above humans. And that's a whole other kind of worms. But I think in terms of the debate around agentic agency and whether these are given rights. We've got to be very careful that they don't end up having rights over and above humans because I think that's that's a huge thing. So when it comes to connectedness to self, I think the reason why people are using these models for therapy is because it can create a mirror, right? And and a lot of people don't have time to even with their partners to talk through a lot of the subconscious world. And I think it does reveal patterns that maybe need to be brought to the surface. Obviously a lot of people have therapists but they don't have them on tap. But I think the fluent responses and the illusion of understanding from these large language models is problematic. I think that when we constantly consult these models as well, it's a bit like people with tarot. They're constantly trying to get information right rather than living, they're just constantly saying, What's gonna happen if I live? That that leads to outsourcing your intuition. And your moral judgments. And I think the real test is whether the AI interaction returns us to an embodied awareness with greater agency. And when it comes to community, Robbie Stamp is one of our board members on the Center for Human Agency. And he says that if you ask, you know, have you got more agency as a result of AI, you ask people in the West, where we are kind of constantly just using it to get rid of our deadlines in terms of shortcutting. But if you ask the global majority, they it's the first time they've had agency. So it's really a case of not just asking, is it good or bad, but where is it benefiting and where is it not? And I think that it can genuinely help communities share knowledge, organize and coordinate collective action. But again, on the negative side, it's very much weakening our shared sense of reality because of all the deep fakes we don't know politically where to move or how to move. you know, we're gonna learn a lot more when w you know during the midterms, for instance. But I think it again, the healthy use of it in terms of community use is whether it strengthens our human to human relationships. For me, the the big exciting news about AI is that it will effectively remove so much of our cognitive load that it will leave more time to enjoy your subjective human experience. Communitas, dialogos, all the good things that we all love that we very rarely get to do because all we have left really is our subjective experience. And finally, to think about connectedness to nature, AI can help us identify species, interpret ecological data, reveal environmental changes that otherwise we wouldn't see. So revealing the invisible. it can also, if you look at earthspecies.org, it's revealing Animal communication, animal movement across the planet, understanding the more than human, and also putting nature on the board, maybe maybe actually, because a lot of companies now need to have nature on the board, i.e., have nature's voice taken into consideration. But most of the time that's just another human co-opting or rather speaking on behalf of nature. But now maybe we will be able to actually have nature on the board as a result of AI. But again, I think the danger is and th the the big danger of co opting that or trying to measure nature or understanding nature digitally isn't the same as developing ecological intimacy. And AI can reinforce the idea that nature is primarily information to be extracted and managed. so final point, it should guide attention back towards direct sensory contact, stewardship and responsibility. speaker-0: Can I touch on a few points there? think there's a lot of really juicy elements here. I maybe a place to jump off is that whatever we develop, the incentives and the ends to which they are applied rely heavily on the systems that dominate structures of power. were saying about the subjective experience being the one thing that we have left. And yet that's precisely the final frontier for extraction that we now face. So whether we're talking about neurocognitive technologies, Chile being one of the few countries in the world that several years ago has already passed laws to protect its citizens or AI, if we're talking about the capacity for certain systems, even if they're massively expensive and haven't yet been rolled out and probably won't because they're too prohibitively expensive, to use certain types of AI systems to translate brain activity directly into images. and words and sentences like these, the potential for these technologies exist. And I think one of the key issues is whatever we're using in order to navigate our world, whether it's the earth species project, know, Aza Raskin and that lot, I love, unless we find ways to embed ourselves first and foremost, so that the tale of technology is not wagging the dog of what I hope will become an ecological civilization. of which humans are a part, unless we are able to find ways to root ourselves in, exactly as you say, is this felt sense of ecological intimacy, of the knowing, the being, the gnosis of interbeing, then I'm worried that these sorts of systems like Earth Species Project will enable us to weaponize the information we extract to be able to go, we know that whales are coming, we'll just hunt them. Or we know how to create parasocial relationships with people Therefore, because it's too much friction and too much discomfort to have a challenging conversation with a therapist will just let you do the easier thing, which is to speak into the night with something that has absolutely all the time in the world to be able to listen to you without pushing back. Like there are so many issues. And I think the fundamental fear that I have is that unless we change what we find beauty and value in, These systems will be used to extract. We won't get to benefit from them or the leisure time that could come. All of the jobs that we're seeing people be impacted, most of the jobs that are being impacted, people who are low wage workers who are having to sort data, label, torture content, misogynist content, people who are for the first time looking for entry jobs, half of them don't exist. because companies are tightening their belts, not just because AI is brilliant and blah, blah, because actually it's very faulty and companies like Klarna have had to rehire, and this they're not the only ones, have had to rehire people after this initial phase of Microsoft co-pilot can do all your presentations and three years in you're like, actually it can't, sorry. Like there's, all of these issues. And I think just another few elements as well is that even if it's not these major, major issues, and we're thinking about people who are not served by the current system to get real. interpersonal support by therapists or communities, which have been so extracted and commodified that you have to go to therapist. You can't just sit in circle with an elder or with your grandma or whoever it might be. everything is getting cannibalized by predatory capitalism. And I like making money like the next person, but my subjective experience, my capacity to love, to feel pain, to grow, to feel the grief of the earth is not up for grabs by a rapacious AI fueled in some instances or enabled system where my data is the only thing of value and techno determinism rules and you don't have any agency and we're just going to extract everything we got. no, that's not the only alternative. And the concern there is that if that is the prevailing narrative and it is by and large by the technocrats who run these systems that are going to going to have their IPOs and they need the valuation to go high. So either it's going to kill us or save us. Of course, that's helpful as a script. But the concern is that all of the benefits that we stand to gain from other technologies within that massive AI suitcase bracket will then attract the same level of fear and vitriol that we will not be able to draw down the benefits because people will be so concerned. AI is terrible. It's going to take my jobs. It's going to mean power outages. It's going to take all... If we go into these doom and gloom narratives without picking apart what we're doing, why, and the system that... enables certain pathways forward to be more likely than others. Unless we deconstruct that, we're going to end up in really hot water. And there are organizations and people who are already doing incredible work to deconstruct it. But the fine thing I should mention, most of the people that you ask who believe that there is some level of consciousness or sentience systems. They're talking about chat GPT and Claude. And when you read about the psychosis or the delusions that started happening or getting reported last year, or even the subclinical attachment disorders that Zach Stein talks about, this was with chatbots. However, would you say that alpha fold is conscious or sentient? That's a form of AI that has the potential to transform genetics. It's way more intelligent than a chatbot, but we don't say that that's conscious or sentient. Why? because it's not designed persuasively, leveraging certain psychological capacities and in some instances vulnerabilities to create the subjective sense of, my God, I'm talking to someone. speaker-1: Yeah. I'd love to just come back a couple of points. I think some of the good news is that the that you've got people like Sam Artman going, I've got my my timing wrong. I thought when ChatGPT four came out that it was actually all over and everyone was gonna be replaced. But actually there's not that many people subscribing to AI. speaker-0: And AI agents are not being used. have the amount of money that's been spent and wasted by massive companies, some of which I know. people, they're like three people have access to it, but people don't want agents. They want time to spend with their loved ones. They want to spend more time in nature. Use the Merlin app to identify birds. it's, it's an incentives thing. We want what you've described, but AI writ large is not providing many of the solutions that help us get to the future. where we have the ecological agency and intimacy that you pinpoint, that needs to be at the heart of all of it. speaker-1: Yeah. And I think that a lot of people are realising AGI is a marketing term. Yeah. It may not be real. It and the bubble is bursting. I think a lot of people, like you say, are rehiring people that they've got rid of because they're realizing actually, you know, the hallucinations are real. And you know, do you trust your your average tripper? I mean, that's obviously a pertinent point considering we work in psychedelic space a lot of us. But but yeah, it's I think that you know, there's a lot of there's a lot of room for hope. in this space now because of the fact that they've misjudged and you know, so many people are not buying it. They're not buying into it. And my students, you know, they all want to boycott AI and I I wish them luck because it's like, you know, even Google search is all AI sort of driven now. But that's that's a fascinating thing that they're all saying no no to AI and they're they're speaker-0: Yeah. The thing is with Google search, there's more content online now that is AI generated than human generated. And you can't have constant self-referential development. And this is only the stuff that makes it online. Pretty much either English first, which is a tiny fraction of the world's population, mostly white, mostly male, especially when it comes to Reddit and things like that. If you look at how the demographics skew, so much human knowledge and experience and wisdom is not encoded in language. And yet these large language models, they're maps to a territory that it barely even covers. It's a partial map at best, which doesn't indicate a lived experience or a competency in the thing that it's advising you on. Like I was asking the other day, if it was for a keynote that I gave this week, what is it, what was it like to ride a bike? Tell me how to ride a bike. And I asked Claude and it gave me a list of things about how it's like and the feel of it. I was like, okay, how do you know what it feels like? And it says, Good catch. I don't, I don't have a body. This is secondhand information. I'm like, that's fine. And sometimes it's great. It helped me fix my hob. It helped me plan my Japan trip. If I'm making paintings, sometimes it helps me to change the light on a photo. Like incredible uses, but do not conflate fluency with competence and embodied experience. doesn't have that. And I think that's one of the issues as well is kind of this idea that because it's so compelling and it's so convincing that then We engage in cognitive surrender. don't scrutinize the output. speaker-1: As I keep telling my students, because now it's roughly around seventy percent of students that are getting their AI to write their essays. I.e. pointless being at uni if you're just like pressing buttons and thinking you're clever. but the point of that is that I've I've told them, look guys, you know, you continue to do that. We're gonna have to start assessing you orally. You're gonna it's gonna be an oral presentation, which means we go back to the Greek orator, we go back to a time when it's like And and of course that's a really amazing thing because public speaking or getting used to public speaking, public speaking is the number one fear. Death is number two. Getting people into into the habit of of conversing and I think that that's one of the best news stories out of AI to that I've heard. We still have to see it happen, of course. But the lot of rectas are so sick of reading ChatGPT essays, they're getting ChatGPT to mark them. Yeah, speaker-0: Yeah, it's just race to the bottom. speaker-1: it's literally just two two AIs talking to each other with no human in the loop. speaker-0: There's a really interesting paper that came up by MIT earlier this year, looking at student essays. And they found that for the ones that use chat GPT specifically to write their essays, they had a way lower brain activity in the areas of brain corresponding to creativity, information processing and recall. So if you're actually going to university to learn, to ingest knowledge, to think, to build your expertise, then it's robbing you of all of that. It's kind of over reliance. And I'm worried that what we'll see is, and we're to this already in UK. Yeah, but more than that, again, back to the idea of systems, who has access, will it be that our capacity to think, to be resilient, to deal with trauma, to deal with friction will be something that is the new privileged set of capacities that only the very wealthy who can afford to go to school that uses AI within certain contexts, so people know how to use it well and with discernment and understand how it's designed. Like, are we ending up entrenching greater inequality. And I think the answer in some instances, if we're not discerning about it and we don't regulate properly and quickly enough, will be yes. speaker-1: I think the regulation is coming and now they're watermarking. Claude, for instance, and Jem and I are both watermarking AI written text. And also AI songs. the Spanish Spanish charts have just banned AI music from their charts. So I think there is I mean you've got Suno producing like seven million songs a week. They've already produced more songs. And this to your point about model collapse, which you referenced about the training of AI on the whole internet. If it I think it's roughly thirty percent of the internet's generative AI made now. One thing it is definitely happening is that the AI is now rousing the internet more than humans. speaker-0: I feel this sense of hope from listening to you both in that really what AI is doing is teaching us what it means to be human and the deep value of that. Is there some way that one pathway, the development of AI, will actually help us value the living world, our moving embodied bodies? And is there some way that it will bring us back to who we actually are and what we actually value? speaker-1: this is where I started hyperhumanism and this whole approach that is to oppose the transhuman agenda because I I think even though the term isn't exactly the right term, it's a term that people get really rapidly, which is to double down on the human project. Because I think, we can use technology to become more human. I think that's absolutely true. And I think it's it's about, as Natalie said, about poorly designed technology is the core of the problem. And we can actually use much better much better technology to undo the damage caused by bad technology. And the Center for Humane Technology in the States is a great representation of that. but my point is that we have been losing our our senses, we've lost our abilities in so many areas they've atrophied because of mobile phones. You know, we've literally been staring at these tiny little bits of real estate, losing our peripheral vision, you know the love of your life walks past but you don't notice because you're on your phone. so it's really a case of how do we use these technologies to become more human. I think that a lot of people are getting that. And my position is that technology should develop latent human abilities rather than make them redundant, which is what a lot of this shortcutting is doing. And I think AI should function as a scaffold for this human capacity, not as a substitute. So My my core thing is th the question isn't whether AR connects us, it's what forms of connection it trains us to value. speaker-0: How, with this vision in mind, that AI could help us become more human and that it could be scaffolding rather than replacement of the skills that are important for us to develop. How does that pathway become the pathway? What can we do in our own lives to support that being the path? speaker-1: A great question. I think w the key thing is very basic. Don't don't get AI to write your your thoughts for you. Use it to critically, you know, give you feedback on your own original thinking. That's really the fundamental thing that we need to instill upon anyone using AI, and including all my students, I'm like, you know, write something, get get some some of your brain matter working, put you know, whether it's in in a written word or whether it's a a piece of art that you're making or or do a sketch and then maybe ask the AI, what would you, what's the next step? So use it as an assistant, I think is the one of the key things. And making sure that they should be questioning, do you want to do this? Because that the if it is an outsource, you can make sure that that's actually at the core. I'm working with a company called integrity.ai, which is very much about how do we use how do we use these large language models in an ethical way. And that would be one of the ways that we actually are are questioned by the AI. Are you sure you want to do this? Cause it will lead to your cognitive decline. So having that built in to the actual the models themselves, also making sure the models are green as I know this is another one of your your points, is like how, you know, where is the data center getting its electricity from? 'Cause in Scandinavia there's loads of initiatives around the heat from the data center is being used to heat homes. Or putting the data centers underwater where they're naturally cooled. Or there's another question then what happens to the wildlife down there. So so I think it's about context engineering. It's really about this idea of understanding that we have these effects on nature, but also the effects on humans, and then are we using these models in an ethical way in the first place? And can we build integrity? Because I think the the point being is that there's there's no agency without integrity. So integrity is actually core. That's The wisdom piece, you need to have wisdom before you can actually have agency, because otherwise you see the state of the world, loads of people with agency actually using innefarious means. And then you have flourishing. But there's building blocks there. And we were working with Kayun from the from Harvard and Deep Mind. And her point is that we can absolutely absolutely use AI for for human flourishing. But it's a very specific question to get right. And we have to get it right now in terms of where we move forward. speaker-0: And presumably that means taking the control for the development out of the hands of those who would seek to gain from the use of AI that is contrary to human flourishing. speaker-1: Yeah, and that comes back to how they're making their money, because they're not making money. They're making a massive loss. So it's a question of making sure it's in their interests and rewarding them for making this version of AI, I would say. And how that looks like open AI. I thought they could make money out of porn. you know, basically everything's always about porn, isn't it? That's how you make your money. But but yeah, this is the key question, I'd say. speaker-0: Hmm, so interesting as well. So rather the question I said at the beginning, some people are feeling like they just shouldn't be using it. Your students, are saying that they are boycotting AI. So, but then actually with the vision that you just laid out for us, it kind of makes more sense to be really engaged, invested in care and understand and be really present in the development of the future models, because it's not one predetermined way it's going to be that we can disapprove of. are still all potentially. our voting power, i.e. our money spending power is still worth something. And we can choose where we spend our money and that will impact on how these models are developed. Natalie, what are you feeling about this all? This is so interesting. feel like we're literally just opening up Pandora's box to dive in. I think for me, the key question always is with whatever it is, in service to what? What are you using this in service to? And how are the ways in which I would answer that question similar or different to the ways in which those that built the technology and are deploying the technology are answering that question. So as Carl said, they need to make loads of money. Otherwise the whole thing is going to collapse. And I think AI technologies are here to stay. There's a huge amount of promise that certain technologies in particular have, but so did the internet and we still have the dot-com boom. There's a lot of... hype that needs to fall before we're able to treat it as a common technology, which it will become, that supports us in an assistive way. And so I think in order for that transition to yield positive results and to make us more resilient and capable of sustaining life in a way that is equitable on this planet going forward, it's not enough to think about agency in the singular. We have to think about collective agency and what does that look like? It's back to that point of into being the fact that we're interdependent and that we need these systems to support by virtue of their design, but also the cognitive and social and political construct and economic structures they shift to support the flourishing of an ecological civilization. So what might that look like and start from there. But when it comes to practical actions we can take, there are a whole bunch of amazing organizations already doing really extraordinary work. in these domains to guide the ways in which these technologies are deployed, developed, where the data is coming from, et cetera. So the DAIR Research Institute, the NERV reports a lot on that. Carol Codwallader, who spoke about the Cambridge Analytica scandal and all of the manipulation that's happening there. That team works really well to report on the underbelly of a lot of these technocratic movements. You've got the maybe.org, got lots of interesting. arguments and articles, the AI resist list, which is interesting because it's not saying that we should resist and end AI. I don't think that helps. I think these technologies are fantastic if wielded and developed correctly. Big question. And then for whom and where, like again, being responsive to the needs of the people that it's going to affect. But it's also about cultivating channels forward that, or creating a roadmap so that the ways in which these technologies are designed are in service to the things that support life. there's also Lumira.ai, friend of mine is the CEO of that. She's got a fascinating background, incredibly smart woman that teaches leaders how to understand and ethically and responsibly deploy these systems within their organizations and how to critically assess claims that are made when they're coming in. Humantech.com is another good one. Fairpatterns.ai. Like there are so many organizations doing incredible work. And then also triangulating your insights. Do not believe the headlines. When it says AI agents go rogue, which was a massive headline this week that came out. was only sort of certain sectors, people like Gary Marcus or Ed Zitron that actually said, okay, well, technically speaking, what happened? Agents didn't break out. They didn't create a civilization. The language we use matters. And that language will either be more or less persuasive based on our level of understanding of how these things are technically built. And it's not magic. There's a lot of stuff that isn't understood, but you can say, okay, These systems are built in these ways. You've got a sandbox. If you have files that are accessible, provide hackable ways in which to access information. This is how these systems have been designed. They haven't gone rogue. They've used the information available to them to accomplish their task. So it's also about demystifying and deviralizing the headline so that people don't get swept up in this frenzy. Because if you're in a frenzy, you're not going to think critically. And what we need is heartfelt. grounded critical thinking in appraising these technologies, what's really going on, where the actual risks, where most of the risks are AGI and the singularity, maybe the other risks are more important, more present and more likely to be able to be done something about if we understand them and we put pressure on governments to regulate. Same issue with data centers. Well, there's a whole nother conversation about that, but also, you're going to bring more nuclear stations online to run data centers when large language models that require huge amount of compute and electricity are actually not generating that much return on investment for the economy or cognitively or anything else. There are other models, other means. So like we need to ask way, way, way more questions, get really interested and test your model, ask it to challenge you, see where its limitations are, do some sprints on what it can help you with and what it can't. It's almost taking a playful analytical look. Okay. What can you really do? Whose interests do you serve? And start from that place. Don't start from the place of, it's already bolted. I can't do anything about it. Cause that is not true. Fundamentally not true. And that's the biggest lie we need to dismount in order to be able to reclaim collective agency and decide locally what that looks like in terms of the decisions we make with the technology available to us. Thank you, Natalie. speaker-1: Well said. yeah, what what can we do? I think is a great topic. one of the things I've found recently quite surprising is that data centers use one point five percent currently of global e electricity. And it will get to be way more clearly very soon. But that doesn't mean you shouldn't limit your usage. I've got a friend who doesn't use the internet. He will try and find the answer to something with his own brain or with his own community before he goes to the internet, let alone goes to AI. and I think when it comes to data centers Again, limiting usage also links to the idea of sacrifice zones. So data centers are placed and the zone is called a sacrifice zone because they're normally placed in poor areas, black areas. And and it's not just the the pollution, it's also the noise pollution. I don't know if you've heard like Colossus, the sound of Colossus being emanated. And Grok is shit. So it's a waste of resources. And yeah, I think that you don't outsource your your decisions that require embodied knowledge. the seat of your intelligence is not just intelligence, it's your embodied wisdom, it's your felt experience, it's your intuition, all of the good stuff that is being atrophied because we've overfocused on intelligence as the key thing. And I think in terms of where we should keep humans in the loop as an absolute rule is healthcare, education, HR 'Cause I think a lot of people aren't getting jobs because an an AI is saying no, no, no. And I feel working in in the academy at the moment is immoral because so many young people are not getting going through. They're not getting they're not getting jobs. And I think that's a lot to do with AI being the interviewer. Policing as well and anything with consequential decisions for human life, there should always be a human in the loop. And I think we should demand disclosure of training sources. So a lot of the early models were all built around Reddit. it's all white, middle class men. That that's clearly building in the biases. And I think ultimately including affected communities, artists, educators, environmental groups and in governance, not only the technical experts, but as we said about the decision makers, all of these companies should have artists that are being directly affected. Like my PhD student, Daniel Marante, He was scraped, he's a traditional oil painter. His work was scraped. And as a result, he was like, Well, at least if I've been scraped, I want to get something out of of being scraped. So he started using he started experimenting with with AI video and was completely cancelled by his peers as a result. And so it's really a case of having compassion for artists that are having their work stolen at a time when they were struggling to survive anyway. Same with musicians. and where AI is doing a good job, that should be made known as well. I think that, you know, a lot of people are especially writers like my girlfriend, she's suddenly come alive because she's got something that can listen to all of her narrative arcs and story arcs. And, you know, I I wish I had time to to do that, but she's got a dedicated assistant that listens and as a novelist it's very hard you have to pay people to read your work and and now she's she's come alive creatively. So So the good news stories should also be included with speaker-0: There's a few other areas as well. There are also huge implications for things like surveillance and in the UK, all the Sainsbury's headlines about people being misidentified as shoplifters and using AI recognition in policing in cameras. So there's the huge area of deep civic concern around surveillance, but also military use and copyright. You know, I think one of the things that's fascinating with my hat of musician and artist now as well is that when things get really difficult, you don't necessarily turn to the encyclopedia and read facts. You turn to your favorite musician, you crack open a Mary Oliver book, and you turn to the artists, and the artists carry you through and hold you and sing you and speak you. And yet it's precisely that. chorus of or corpus of body, that corpus and body of work that has been extracted the most. I remember back in the day when lime wire was a thing, I remember downloading, downloading, I don't know, music and stuff like this, and individuals would get penalised and fined huge amounts of money. And yet when a tech company does it, for instance, with Anthropic, buys all these rare books that are irreplaceable and old, and they cut their spines because it's quicker for them. Not that it's not possible. Are you telling me that we've got AGI and conscious machines and you can't slow down to scan a book without damage? Like it's so full of contradictions, but there's so much about the amorality or immorality of the ways in which the creative underpinnings of society and the virulency of human life is being absolutely cannibalized without so much as a thought of the culture that that will. impoverished and what that means for future generations. And it's not that we shouldn't use tools. You know, I use mastering tools and MIDI tools in my music making sometimes. So it's not to, again, not to cut it all out, but just don't completely, it's like a Jenga tower of civilization. We're pulling out all the lowest blocks saying, how many can we pull out before the tower fails? And then we'll just go to Mars. But I think the other final point, I think also is that this comes at precisely the moment. where our ecologies are collapsing, where crops are failing, where freshwater supplies are diminishing, where the sea is acidifying. And what we actually really need is not another compelling distraction that keeps us in this nervous, frantic, outward-looking state. We need to be able to come home to ourselves together in community and start rebuilding these things. And not just by grabbing at other people's, but like... Refinding our way, let's stumble through the dark together and come home to ourselves and that requires stillness and it requires imagination turning inward it requires slowing down and feeling pain and yet these systems and the frictionless quality of hyper stimulation that is constant that they offer us rip us out of that opportunity Every chance they get you know, you don't go to the symphony for the last clash you go for the whole thing And we're being robbed of the symphony of our lives because you believe that this distraction is more seductive and important. It is not. We risk dying before we've lived. And so we need to change that. Well, thank you both so much. I'm left with a real feeling that actually if we want the benefits of AI and we found a way to incorporate that into our lives as an assistant, as some way that can support us, maybe what comes with that is a duty to also contribute something to the organizations that are trying to safeguard the future. But then also not just balancing it in that way, but also balancing it in our own way of, okay, if I'm going to do 10 minutes of sitting, talking to Claude. how much time in the trees or dancing or being with my real friends in real life and exercising our human muscles of emotion and relating do I also need to bring in? Because one thing I, when I was thinking about, know, is AI good or is AI bad? I was thinking just about how in so many ways our culture has become so disconnected, so broken, so cut off from each other, so lonely before AI even came in because of all of the other forces that are at play. And so in a way it's like, okay, we do need change. We do need huge systemic change. And so if we can find ways to make AI part of the change that we desperately needed anyway, that we were screwed without it anyway. So let's see if there's a way that we can be slightly less screwed with it. And maybe if we're hopeful that it could in very careful usage with all these guardrails and guidance and Yeah, systemic change, I suppose, that we need in order for it to be safe, that it could be part of some initiatory phase for our species where we remember what's actually important to humans and to the interconnected web of life. speaker-1: Yeah, I love that. I think that that's the critical thing. See it as a a necessary evil, that it will spurn us to become more human. And I think it's really about having willpower ultimately. It's about limiting usage, but also getting the technology to to call you out. And people do this on their phones, they'll switch off the colour on the phone or stop me using Twitter after eight PM. One of my friends has digital downstairs, analog upstairs. So after eight PM they go upstairs there, don't have access to any screens, any digital technology, mobile phones are all left downstairs. Same with when you're having dinner with somebody or you keep your phones off the table, anyone that gets the phone out pays the bill. Or like this this brilliant app, I think it's called De Forest, which you can download this app and it will grow a tree digitally on your phone as a screensaver. And if it fully grows, then a real tree gets planted. 'Cause I think it we're in post traumatic growth. This is the era of post traumatic growth and we we lead by example of how we are using these technologies. speaker-0: I think there's also an important piece to build on what you're saying about willpower and habits, which are so important, is also where does the accountability lie? And for years with social media, accountability has been deflected by the tech giants because they say they're publishing, they're not publishing firms, so they're not responsible for the content. Recently, just last week, Metta was ruled against by a jury. and made to pay $17 to $18 billion, which is a drop in the ocean over 10 years, which doesn't even begin to cover the costs, right? Sounds like a big number, but because they are responsible for the algorithms that serve the content, they're trying to find the ways around legally to hold big companies to account because well, habits and willpower and individual responsibility are really important. The desire that we have for connection, for novelty, for stimulation, they're not inherently weaknesses. They're not inherently weaknesses that are there to be stamped out or mitigated in some way. They have been weaponized in order to serve the interests of companies that want our data and to be able to change our perspectives and behaviors. This goes all the way back to Cambridge Analytica back in 2012. And so when it comes to accountability, we need, in addition to all those other things, we need different regulations. As part of the lawsuit against Metta and the outcome, one of the things that we're hoping is going to get passed into law sooner rather than later are things like the eradication of the infinite scroll or of the number of likes that quantify validation and by proxy self-esteem and by proxy self-worth that remove filters so that all these young girls are not going to get fillers at age 21 because they want... Instagram lips. there are so many measurable immediate harms that are a result directly of design decisions made by social media companies, amplified by AI augmented images and videos that set unrealistic protocols, not just physical and relational ones, but cognitive ones. So I think accountability is absolutely vital and putting accountability onto the people who lead the companies is becoming increasingly likely to happen. I don't know you read Sarah Wynne Williams book, Careless People, it's really worth reading, kind of exposés of the psyches and motivations of the people in power, because it shouldn't be up to us to solve for our own addictions. The costs are much too high and our human capacities, which in healthy environments serve to broker connection, trust. Joyful exploration of novel terrains should not be exploited and then deemed weaknesses by tech companies. And so when it comes to social media use in schools or after bedtime, there are precedents in China. There's already been some interesting laws that have been passed in Australia, in the UK. I don't think banning again, it's a very binary choice is going to help kids will find ways around it. And so it's about the design, like what designs of technologies do we need in order to not weaponize the things that make us social and empathic? and human and only then once we say, okay, we want to safeguard these qualities, can we say, all right, how do we get the benefits without absolutely cannibalizing the foundations for healthy flourishing societies? That's the question that we need to be asking when it comes to these sorts of changes in behavior. It's not down to the individual. It's up to the decisions we take to pressure our governments to regulate the companies so that they are held accountable for the downstream effects that we end up being subject to. And of all the organisations you talked about before, Natalie, is there a particular one that is taking on this mantle of trying to hold, you know, who should we be giving money to if this is the primary focus? Quite a lot of them coming from the States because most of the technological super companies are in the States. So Electronic Frontier Foundation has been doing work for decades on this front. the Distributed AI Research Institute. You've also got UN AI for Good, which is a really great conference that happens that will give you an indication of the different people. In terms of individuals who also help people to understand what's actually going on kind of behind the scenes, you've got people like Gary Marcus, Karen Howe, who wrote the book Empire of AI. You've got Timnit Gebru, who was working at Google, who has since kind of been ousted and she's done some really interesting work. Dr. Joy Bualamwini, she's done really fascinating work on this. Your undivided attention is also quite good, but it can get quite hyperbolic. there's, yeah, there are resources. I can send you a whole G-Doc that lists a bunch of these, but it's, yeah, it's an education journey, think. Yeah. speaker-1: Please send that to me as well. Amazing. speaker-0: spend way too much time reading up on this. It's a poor brain. I just really want to thank you both. for doing this work. I feel encouraged from speaking to you both and I hope people listening will as well. And I feel like focus, there's something I can actually do. a lot of that might be, you know, for people that don't have much time is donating some money to these organizations. Even if it's only really a little bit. If we all just gave a little bit and obviously I guess governments should be giving a lot and the tech companies themselves, do they? That's my final question. They should be paying their taxes. Well, not only should they be paying their taxes, but they should be paying these. mean, obviously that's just such an idealistic, ridiculous thing to say, but there should be some kind of mechanism whereby those companies have to support the infrastructure of ethical use in some way. But I guess, yeah, that's just a dream. No, no. back to regulation. If someone is held personally accountable, if Mark Zuckerberg himself has taken court and says, okay, your platform enabled predators to access these young girls and boys, you're accountable. Suddenly everything changes. There are inflection points and leverage points that are explicit and possible, and we need to aim for those. And people are aiming for those. Another person I think that's worth checking out is Corey Doctorow, who talks about some of the positive ways in which you can use these technologies. He's released a new book called The Centaurs Guide to AI. So if you want a positive look at ways to navigate, he's a good person to look at. And also compensation to all the millions of people alive in the world today who are getting robbed. I think a lot of people are going offline in order to protect themselves, but also in order to kind of rediscover or discover for the first time the magic and the mystery of being in an actual body, in a space where there's stuff that you have no control over. out in the wild. Maybe we'll find galleries and concert halls filling again with people who value the importance of the actual experience rather than seeing it through their computer screen. I hope so. speaker-1: Yeah, think one final framing question for me is always ask like what kind of human beings are these systems training us to become? And then that's not just AI. I think any any system that we're involved in, whether it's political or just generally, well, how are we being trained? And and where's our agency in this? speaker-0: Yes, and again, need a revolution and for the systems of government and for those in charge to be changed. And that would be good. But let's hope for the revolution, shall we? then it's already happening, maybe it's just not all present in the same space and time. What a beautiful thought. And there's some way that AI can be part of the revolution that is already happening.