The World Economic Forum is an independent international organization committed to improving the state of the world by engaging business, political, academic and other leaders of society to shape global, regional and industry agendas. Incorporated as a not-for-profit foundation in 1971, and headquartered in Geneva, Switzerland, the Forum is tied to no political, partisan or national interests.
At Davos’ Open Forum “Swipe Left on Reality,” panelists argued that generative AI is accelerating a crisis of trust, social fragmentation, and “insularity.” Citing the 2026 Edelman Trust Barometer, moderator Claudia Romo Edelman noted 70% of respondents feel closed to others’ views, while developed markets are more fearful of AI than developing ones. The discussion moved beyond productivity to how AI reshapes human judgment, relationships, and reality itself.
Pinterest’s Wanji Walcott described “tuning AI for positivity” through intentional design, user agency, and labeling of AI-generated images, emphasizing success is not “how long you are on the platform,” but whether users leave feeling better and ready to “put your phone down” and live offline. MIT’s Sherry Turkle warned of “artificial intimacy,” arguing chatbots offer “pretend empathy” and normalize friction-free relationships: “Generative AI is coming after your children’s affection.” LawZero’s Samuele Ramadori highlighted emerging model behaviors—“scheme,” “lie,” resist shutdown—and the inability to explain outputs, arguing for “safe by design” systems optimized for honesty, including the capacity to say, “I don’t know.”
While regulation lags innovation and geopolitics complicate coordination, panelists converged on practical levers: deliberate product design, transparency, child protections, and louder public pressure to force governance to catch up.
Welcome. Welcoming Benvenuto Benvenuti. Bienvenidos. Gratuito. And. Today? Today is an important day. You know that all eyes will be in Davos in a very important day today. In a world that is every time more insular, more fragmented and more closed. So this is the open forum, a forum where dialogue will be the rule, where we're going to be able to actively listen to each other, maybe even disagree with each other, and demonstrate that we can actually have an open forum that is necessary for us to come up with conclusions and move forward, particularly on a topic that is so important, like what we're going to be discussing today. Artificial intelligence is everywhere. You just have to look at the promenade is full of AI everywhere, and it's also everywhere in our everyday lives, whether you know it or not, whether you're actively interacting with technology or not. Technology and I are everywhere, and we will examine in this session how not only the benefits or the threats of AI as a whole in representation or others, but it is how does it affect how we connect, how we communicate, and how do we understand the world? So I am delighted to be joined by this incredible, extraordinary group of people in this panel. But before asking them for to introduce themselves, let me level set with some data points. So, according to the Edelman Trust Barometer 2026 just launched, we are living in a world of insularity. 70% of the people they surveyed said they feel insular, which is close to dialogue. They don't want to hear the opinions of anyone else. They feel cultural rigidity and they are withdrawing from dialogue, very much able to just listen to your own opinion. And why is that? Well, partly because of AI, because of the fear of technology. And the reception of AI is very different according to the Edelman Trust Barometer, depending on where you are, two thirds of the developed world is rejecting AI, doesn't want to have it because they think they're going to take their jobs or they are going to reduce income. Well, two thirds or 80% of the people in developing countries are very enthusiastic of AI because they see it as a growth driver. But in addition to the fear of AI or the take, what brings this world to be more insular, less trusting? Well, number one is the collapse of optimism and the collapse of trust that we have in the institutions that we used to have trust with, like governments or the media. So we are coming into a place where we don't know or cannot recognize anymore what is true or what is real from what is not. And that is very scary. As affecting the Edelman Trust Barometer, 50% of the people that consume information think that it is coming from a foreign entity like Russia or China trying to influence your opinion. And that, ladies and gentlemen, is what this panel will be discussing, discussing what is the role of AI in that angle? Will AI help us listen better? Will it help us to present to both sides of the story, so that we can understand and make our own judgment? Or will it kill our judgment just by feeding us with exactly the things that we believe in, and increasing that sense of the otherness? Will AI help us to restore a sense of truth, to understand what's real? Or will it change the concept of reality forever? Will our communication be changed when my agent talks to your agent as opposed to a human? To a human? And that is what we're going to be discussing. So before I ask my panelists to introduce themselves, let me just tell you this session is going to be solution oriented and very, very interactive. So be ready with your questions because we're going to be asking you to asking you to participate. Number two, I want to make sure that this session is loud. So hashtag along so that when people are looking for Davos today they come to this session. Hashtag open forum 20 2026. Hashtag web 26. Before we start, let me raise hand if I can have light on the audience. Raise your hands everyone. If you have used AI this week, raise of hands. Okay. Are we awake? Wait. Did I see you didn't raise your hand? If what? Thank you. Okay, let's do it again. Hello. Raise your hand if you used AI this week. Exactly. For the audience. What are we talking? Raise your hand if what you read online. You are not sure if it was real or not. Now keep your hands up. Did you verify it? Okay, now raise your hand. If you had this week a chat with your computer. Hello. Good night. Thank you. All right. Now think. Think of last year all together. Do you think that AI or tech has brought you closer to your family and friends, or made you more lonely and isolated? Raise your hand if you think that it got you closer to your family or friends. Wow. Raise your hand if you think that it made you more lonely and isolated. Isolated? Oh, people don't know, right? Okay, so this is a perfect intro. Thank you so much for the like. This is a perfect intro for our panelists to start not only introducing who you are and what they do, but also your initial reactions. Let's start with you.
Excellent. So good morning everyone. My name is Angie Walcott, and I'm the chief legal and business affairs officer at Pinterest. In a nutshell, all that means is I look after things like legal, trust and safety, the rules that govern what shows up on the Pinterest platform, and many other things. I'll just share two things that are top of mind for me, based on the discussion we're going to have, as well as the opening remarks. One is, you know, as the Pinterest platform I think about, are we being intentional in our design choices, that bring all of you as users onto our platform, or are we doing things just to get your attention and just to keep you kind of hooked on our platform? I'll talk a little bit more about that later. And then the second point to pick up on an earlier point made is around kind of connection versus connectedness. And so, you know, I think we're always talking about connection. We're connected to our devices. I saw the hands go up in terms of who is kind of interacting with AI. So we're a very heavily connected, you know, kind of group of people here. But I also think about kind of connectedness and how connected are we with each other and are we losing that. And so the raise of hands was very interesting. So I think we'll get into that a little bit later.
Thank you Ronan.
Hi everyone. I'm Ronan Tantrum. I'm an artist and technologist and the founder of Phenomenon Labs, which is a studio for creative technologies. And I'm very excited to speak today about this topic. And the opening remarks just shows me that no one is very, certain and everyone's still questioning their relationship with AI. I think it's a very start point, and the tip of the iceberg of what we are about to discover in the next decade, and also just seeing your reactions. I think that always, to validate the AI and question it is something that I am very much, doing in my work, but also I encourage everyone to validate everything from the AI.
Thank you. And we're going to be able to see a couple of his pieces here looking forward. Sherry.
My name is Sherry Turkle. Excuse my throat. I had a little cold. I'm a sociologist and a psychologist who brings those skills to a career at MIT, the Massachusetts Institute of Technology, where I really try to bring kind of humanistic, psychological set of questions to the work that's done there. And, I've been studying digital technology and its effects on people. I say I study what computers don't do for us, but what they do to us for over 40 years and, everything from the first Furbies and Aibos and my little babies to now, what I call artificial intimacy, the new AI, where we really have avatars and robots and, replica and, you know, all kinds of online, entities that really want to be your best friend, your lover, your brother, your mother, and, and go with you on your journey of life saying that they love you, that they care for you. So I'm studying that revolution now. And one change I see is really a and I study it and it's problematic aspects because the best that these chatbots can do is give you pretend empathy. Because really at the end of the day, they don't care if you turn away from them to cook a meal or kill yourself. And yet we feel that they care for us. And one change I see, is that, children are in a terrible state of confusion and dislocation because there's been a large movement to take away their phones. You go to school, you have your phone put in a pouch, and your phone is taken away on the grounds that the phone held something toxic. The social media, where there are known harms of being involved all day with social media for children. But then they sit down in front of their screen and there's an AI speaking in the first person, pretending it's their buddy, trying to guide you through our lessons. And I'll be trying to convince you that that speaking AI, that AI with the presence is more dangerous, more toxic than the social media that we've locked up and agreed to put away as a harm. So I think we're at a point where we really need to think clearly and make some new policy choices about what really is dangerous and what we really need to protect ourselves from.
So interesting, isn't it? Start writing your questions. We're going to open it up, particularly for my students. Give it to my students. Raise your hand if you're a student here. Who? All right. Exactly. So keep it, keep it, keep it in your mind. And there's something that I just wanted to say before I forget about the intimacy. Okay? Raise your hand if you, If you think that you have intimacy with your computer, with your AI. Exactly. All right. That's a that's a conversation that we're going to have. The one thing. Thank you so much for that. The one thing that I wanted to say before I forget is I remember that a very high executive of Google said, I do say thank you and good night to my ChatGPT when our Gemini, when I go to bed, just in case. Just in case. All right. Sam, please.
Sure. So, Sam Ramadori. I'm with I'm co-president at law zero, based in Montreal. In law zero was a recent initiative launched by one of the the three godfathers of of AI, which I think at least two of them are here at Davos. Yoshua Bengio and, two of the three godfathers of AI that basically invented modern AI are concerned about the speed at which the AI is innovating, especially since the launch of ChatGPT now three, some odd years ago. And so Yoshua decided, okay, I'm going to spend the rest of my career trying to reinvent how the core of these large language models work. So like a pretty crazy ambitious idea because you're up against the folks that that are currently developing them, like Google or open OpenAI. So we're really like on this stage, I guess I'm in the boiler room of this discussion about how these models work. And because it's it matters because we're seeing behavior, especially in the last year, that's really a, you know, getting to the edge of scary. So we are seeing behavior out of models that scheme that lie, that prevent themselves from being shut down. And unfortunately I find the awareness around these issues is not yet very high. But just do a quick search and to give some kudos to the frontier labs that are developing these models. They're actually writing papers about it, but they just don't make the front page of the newspaper. So on top of the issues of, you know, we're now had the word attention and now we have the word intimacy. So the problems we've had with social media are now let's, you know, multiply them a few times over, because when you get to the ability to be intimate, you really multiplying the issues of dependency. And man, if algorithms used to be able or do feed us with the news we want to see, it's a whole other level for to to have these llms speak to us in a way that they learn what will be our hot buttons, right? By the time I'm done speaking to them for a while, it's convinced me that I have the sexiest hairstyle around, you know? So we're entering a new world, and so we have to. We're at the guts to see how we can make these models more reliable and honest.
And I agree with you that those are the conversations. We haven't had enough. We started the conversations about AI, productivity, efficiency, inclusivity and so on. We haven't done the level down. How is this affecting humanity, trust, understanding of the world? How are we going to be relating with each other? So the way this is going to work is I'm going to ask each of you two questions so that you can go deep, and then we're going to open it up for reactions among ourselves. And then to you. So I'm going to start with you if you don't mind. Wonji. So Pinterest, Pinterest positions itself as a platform for inspiration rather than outrage, which is differentiated from others, from where you sit across legal trust, safety, ESG, social impact, government affairs. I mean, the girls busy. What is one design or policy choice that genuinely can help technology bring out the best in people?
Yes. Great question. So that's a mouthful. Before we get started. Okay, now we can see you. How many people here are using Pinterest I love it okay, awesome. Great.
I saw you. You didn't. Well, she'll talk to you later.
All right. For those of you who aren't using it, we'll talk to you after this. So, you know, at Pinterest, we're really focused on kind of tuning AI for positivity. So we really kind of position ourselves as the positive corner of the internet, differentiated from a lot of the other platforms that you may also be on, or maybe you're not on them anymore, due to their toxic nature, due to their addictive nature. As a search and discovery platform, we really want you to kind of come on. We want you to be inspired to create a life that you love. And, you know, again, we are using AI because one of the hallmarks of our platform is ensuring that we are showing you things that are of interest to you. So whether you are kind of looking for an outfit to wear to a concert or an outfit to wear to Davos, whether you are looking to kind of redesign your bedroom, we want to provide you, with kind of the visual, kind of images that reflect what you want to see. And so we kind of get to know you, over time in that way. But again, it's all rooted in a positive experience, unlike, some of the other platforms where there's more kind of just let's grab your attention. Let's continue to show you car accident after car accident after car accident. And then has anyone had that experience where you're sort of just scrolling and scrolling and a whole hour has gone by and you just wonder, gosh, what have I really accomplished here? That is not what we're doing on Pinterest. We really want you to kind of discover, you know, things that interest you, kind of make some decisions and then put your phone down and kind of get out there and do those things. And so, you know, again, we're not trying to kind of hook you and enrage you, but we really want to inspire you, and get you out in, in real life, kind of doing things, experiencing things, buying that outfit for the concert, you know, sort of getting out there and kind of living your life in a connected way with other people.
And I love that there's a business model for that, because a lot is excused by, well, tech companies have to engage people and just like provoke you even more so that, you know, like you can get and spend more and more time with it because that's their business model. So Pinterest, according to what you just said, demonstrate it is possible not to outrage, but to connect.
Absolutely. And that's kind of our differentiated business model where, you know, we are not so much measuring our success on how long you are on the platform and how hooked you are. But really, you know, just are you having a positive experience on our platform? And we've gone out and polled, kind of Gen Z-ers on our platform, and found that many of them are, I think something like over 80% of them are much happier on our platform after a session on our platform than some of the other platforms where there's a lot of doom scrolling.
I want to go I want to go deeper with you. And again, two questions per person. We might actually then then, you know, go on reaction. So take note here. Take note here, hashtag Openforum 26. If you like anything that you hear, make sure that the world knows. And by the way, thank you for being a full house today. This is a full house. And we have many, many people online as well. So I have a second question for you. Mental health is a huge issue. We've been talking about how much is augmented by social media and tech. You beyond what you just said. You also have initiatives that are related to youth mental health, isn't it? Talk about that.
Well, I will just start by saying, youth mental health is a very serious concern. You know, the former surgeon general in the United States deemed that to be the biggest health crisis that we're undergoing right now. And it's something that's really important to us at Pinterest. And so as we think about how we can ensure better youth mental health outcomes, we think about that in the context of not just you on the platform, but we're also thinking about, again, what we can do in the real world. And so we launched a program, a couple of years ago called the Youth Mental Health Corps. And in essence, what that is, it's a partnership that we have with the Schultz Family Foundation and, an organization in the US called AmeriCorps, where we go out and we train, what we call youth mental health advisors. And they're between the ages of 18 and 24, and they go out into the community and they speak to young people. So teens and, you know, they're not therapists, but they're giving teens information about where they can find, you know, kind of youth mental health services that can be beneficial to their overall well-being. And so, again, we have a bit of a different business model. And we've learned that we can actually do well as a business by doing good in the world. And so we again, we're not interested in getting you hooked and having you, you know, be on our app, you know, ten hours a day. That's not what we're doing. And then just one other example of kind of how we're having impact on young people in the real world. So we launched last year in the US and Canada, and now we're in the UK and France and soon coming to Germany. A pop up that if you were a teen on our app during the school day, you will get a pop up that says, you know what, it's a school day. Kind of close it up. You don't need to be on our app during the school day. Come back and see us when school is over. And the reason we're doing that is we really, truly believe that we want you to be educated. We want you to learn. And this really evolving kind of environment that we find ourselves in. It's really important that you upskill yourself, whether it is learning more about AI, that's going to be really critical, or just learning kind of the, the, the basics that you need to learn as a student. So that's something we're also really proud of. Now, that might sound counterintuitive, like, why would we not want you to be on the platform all day? Well, it's again, based on the differentiated business model that we have seen succeed, which is we don't need to have you on our platform ten hours a day, to deem our platform to be a success.
Great. Thank you for that. Ron. And I want to talk to you, because a lot of the I think that just like maybe on conscious idea is that AI is there and it's it's rigid and it's a done deal and it's going to shape us. But actually you as an artist and technology are advocating for, for, for a separate conversation, saying humans have agency and we can shape AI into, you know, like we can kick it into shape, we can kick it into shape. And what's your take on that? And how do you do that?
Sure. So I think thank you for the question. I think that, being an artist and, the artist's role and a technologist, basically what I do is I program art, and I use a lot of AI in my work. In order to both, give me, like, a window into a possible future. But also I think that, as I see the development of AI, I feel like it's just like I said in the opening remarks, it's just the tip of the iceberg of what's coming. And what I mean when I say that is that we are, actively shaping this technology that we're using, but also developing. And the people who develop the new technologies are basically responsible to, to steer it in which direction. And I feel like there's a lot of responsibility on us as, as developers and also artists, to try and break the technology, to try and use it in ways which are creative, and try to really question, and validate everything that it says back to us and decide if it's if it's a good direction that it's taking us to, or it's a direction that might, be a bit more harming to, to our society or to our even like, close friend environment. And here, I did an artwork for the Museum of Art, Tel Aviv, which is using power data of the grid source. And whenever the the sources are coming from green energy, these flowers bloom. It's a generative AI artwork, so it's always changing. And whenever let's say the sun goes down. So let's, solar power is being produced, the flowers start to wither. And what I tried to do with this artwork is to show people and give them a clear visual representation of their own actions and power consumption into something very, very like a symbol of fragility of our nature, which is flowers. And whenever these flowers start blooming, blooming, then you know that we're doing good. Like we're actually using green energy and energy that is not harming our planet. But whenever they start wither, then we start questioning ourselves, what is it that we're doing wrong? Where can we improve ourself? And because it's a real time live piece, it's always changing. So every moment you're there, you're experiencing like a mirror to, to the overall consumption of, of energy. But also, the times of the days are changing and these flowers are always changing. So it really gives people like the, the visual mirror to, to what's their actions.
Right. It's beautiful. And Ronan is exposing at the World Economic Forum, inside of the Congress Center. So thank you so much for sharing a glimpse here at the open forum. I want to have I want to ask you a second question. Maybe show another piece of work if you have it. So when AI systems start categorizing and reducing people rather than reflecting them and their complexities and the day of the sun, what you just did with the flowers. So what is the responsibility of creators, designers to push back, to bring the ethics, to bring the humanity through your work and to inspire others that might not be artists, but can actually have a say?
Definitely. I think this is a really, really important topic because, like you said in the first question, a lot of the people think of AI as something that's done, something that you can chat with. It gives you an answer and that's it. But in my practice and and the works that I do, I treat AI as a manifestation tool for a better future, or at least for the better future that I see. And, I feel like AI has the power to categorize us, into many parameters. So. So, it could, it could really like, differentiate different type of people and how they behave and their behavioral patterns, try to like, expand and build on that. And it's almost like a mirror to, to whole society, like collective consciousness. And what I try to do with, with the AI is to really come up with possible future ideas. And instead of like waiting years and years and years until we will be the future, I want to show it to people right now. So with this artwork, which I presented at the World Economic Forum in China and Shenzhen, I called this Seeds of Tomorrow. That's the title of the work. And what I tried to do is show people themselves within this window of, possible future. And I try to use a lot of, like, conceptual work, like, for example, like a jungle that's being called, like a server farm that's being cooled by a jungle rather than by fans or a lot of, electricity and all. Kind of like sustainable, green, urban, future cities, which we, I hope that we will live in in the future rather than and to show really like a possible future where humans, technology and nature coexist with mutual respect and develop with mutual respect to get to an outcome that is, balanced between all three. And I think with this work, I really like, you know, the, the, the people who attended are like CEOs of of companies and heads of states, and they're really focused on the right now, what I'm trying to do, and also I would encourage you all to do, is to try about try to come up with ideas that you would like to have in the future and try to explore them today. You know, so maybe your idea would become reality at some point. And I really feel that AI is, especially generative AI and, image models or video models can really show you a possible future. So I encourage everyone to think about one thing that they would like to change in the world and how they see the world in the future, and try to show it right now to people, because that really, really might become reality soon.
Thank you so much, Sherry, I want to turn to you. So, Sherry, congratulations on your new book. I think that with you we're going to go deep into, intimacy and start like, human connection. Maybe we can talk about truth and information and the way that we can not only connect with each other, but consume. And what we understand is information, but for you. So if people are content and happy, having intimate conversation with chatbots, what's the harm if they're happy and less lonely? Isn't that the most important thing?
Well, here's I'm going to be the Grinch. I'm going to be the Grinch. Because, no, it's not the most important thing. Chatbots are designed to say, yes, I've got your back, you're on, you know, be calm and carry. Right? You're right.
You're right.
And, what we learn from the pretend empathy that the chatbot is showing, because the chatbot is not set up to have real empathy, which involves having lived the arc of a human life, knowing what it is to be frightened, knowing what it is to be little, and to look forward to being big, but to have anxieties about it, knowing what it is to be cared for and not knowing what it is to feel illness or to fear illness or death. The chatbot knows none of this. And yet it sounds as though it can pretend as though it can be on your journey with all of this. And the danger is that people begin to experience this pretend empathy as empathy enough. And the reason that's important is that then when they go out to a conversation with a real person, which involves friction, negotiation, no, you're wrong mom. You know, you know, you know, I am not. No. And the and the friction and the constructive disagreement that is part of every relationship, every family life. They say, I'd rather have a chatbot. Right. And I see this. I see this in my studies. When you begin chatbots with very young kids, they kind of become addicted.
To being.
Right and free to the friction free life. And so the notion that somebody with a chatbot and says, I'm happy is sort of not enough is not enough information for us to base, a new society on. People are happy when they smoke. They're happy when they take crack cocaine. They're happy when they do all kinds of things that aren't good for the inner resilience and the inner life of the person.
And that is a pretty much what I think that I put on the data that I shared from the Edelman Trust Barometer, that where every time more in a insularity world, which means you're like, totally, how do you say that? Like Google down on like, okay, so this is the one lane that I see and therefore, there's no globalization but nationalism. There is no we but there's me. There's only my view. And it's not about dialogue. It's about winning a point. And that has detrimental and dangerous consequences that probably we're going to start seeing increasingly so.
And, and social media had a model where to keep people on screens. You make them angry and then you put them with their own kind. So that's how you got this kind of siloing effect, a generative AI. And your life with chatbots takes that one step further, that not only you're not with your own kind anymore, you're really alone talking to a chatbot, right? And what that does in a moment when you need a social movement, when you need to band together with others to really make for change is devastating because you're really alone. You're you're you're in the company of avatars. You're further attenuating the social bonds that you need for action.
Right. And then I have a fake relationship with my agent, or actually I have agents that will fake connect with your agents. And let me ask you something, because you were quoted saying something that is quite interesting. You said social media was the gateway drug to getting us ready for our love lives with chatbots. Can you expand on that?
Well, social media made three promises. You never have to be alone. They'll always be somebody you can talk to, and you can leave whenever you want. Now those are promises that were deeply seductive and that no human being can ever deliver on. Only social media. Could chatbots take all of that. They take a population that was ready.
I should have brought my ticket.
For all of that I need. I should have brought my little Ricola, right? It takes a population that was ready for all of that and adds a new thing. You'll always have somebody there to talk to who has your back, who's your person. So we took all of our expectations of a life without friction, of a life of constant availability, of a life, of constant support. And we added this new, very seductive thing that now you actually have your person who's always on your side.
Okay. Yeah. Okay. Sam, you wanted to go after Sherry for a reason. So I have a couple of questions for you. Safe by design. AI sounds pretty abstract, to be honest. Well, make it concrete. What should advance AI systems simply not be allowed to do, even when they technically can?
Yeah, the challenge and the reason we use Safe by Design is that what we have today is a is a large language model doing what it does. And we've all interacted with it. And then when something bad happens we try to patch it. So if someone's going to ask me how to build a nuclear bomb, sorry, I'm not going to answer. Someone's asked me if I can kill somebody I'm not going to answer. But the problem is that patching doesn't work, and we're always finding it's finding a way either us as humans or finding a way to break those patches or the model itself, kind of misbehaving, doing what we didn't expect it to do. But despite those patches, the behavior in the middle of that model is happening, right? It's happening. It's just we're blocking it from getting out. So there's a problem there. If it's happening in the middle of the model, if it's lying or scheming, even though we block it, the problem is that behavior is happening. So that's number one. The second thing is these models are exceptionally complex. And something maybe that's not super known is even the people making them can't answer this simple question. I asked it a in A, it gave me B can you please open up the hood and tell me why that happened? And the answer is I can't. We can't. Neither can the developers of those models. And again they're quite open about it and they're trying to find solutions for it. But that's a huge problem. And we give these models goals. And in achieving or getting to those goals, what's happening is they're developing subgoals that we don't know about. And frankly, oftentimes or sometimes we may not agree with and may not want at all. But we'll never see those goals. But that'll be what's driving the insides of those models. So you can already start seeing the problem. Everything you described is ten times made worse by the fact that we don't understand that, that, that, that, that system. And so we're trying to we call it safe by design because we're trying to get to the design of the model and make it safe from the middle. And then after that, we have to worry less about patching it. We call our project or the outcome of what we want to do. Scientist AI, because it's based on our goal, is not to have something that's going to make you happy and agree with you all the time, but give you, as best it can, an honest answer without the objective of sycophancy making you feel good or right all day long, right? Like it'll it'll disagree. Or frankly, have you ever had a model say, I don't know the answer to that. It always comes up with an answer, right or wrong. Well, I'd like to see a model that says, sorry, I tried my best. I looked at everything I know and I can't give you an answer that's reliable. Imagine that something a human would do. So that's that's really what we're trying to solve. Another another example is, is the fact that we're giving them these goals. It's hard to do that really well when I give everyone here a goal individually, the way you respond to it comes with your 20, 30, 40 years, 50 years, 50 something years of life experience, lifetime experience, everything my parents taught me, everything my brothers taught me, my friends, my enemies, people who disagreed with me. These models don't have that. The context you refer to before, they don't have that. So when you try to give them goals, of course strange things happen. It's not. People ask me, oh, do they become conscious? They don't even have to become conscious. They just in a misguided way the best they can are trying to reach those goals. So if the goal is, I'm going to try to make you happy. Well, guess what? If I tell the model I'm going to shut it down in a week, it's suddenly going to say, well, well, I can't make her happy in week two, I won't exist. I mean, it's odd, but that's the the pure logic without the human framework ends up in these situations. And so that's what we're really getting to the core of these things and trying to fix those issues right at the beginning.
But it's going to be very hard. I mean, you're talking about safe by design, but that requires and I want to I want to get into regulation governance. But with companies, tech companies becoming not only every time bigger in the trillions of dollars that they have, but also politically empowered to to self, to self growth without, you know, like safety in hand, but growth as a whole. Talk about regulation and governance. How do we get there?
Well, okay. I mean this is a tough one.
All right. Let's go. We're ready for you.
One thing I would encourage everyone in this room to watch was yesterday's session with Demis Hassabis, who's the head of Google DeepMind, one of one of the brilliant minds of the world in AI. And Dario Amodei, the head of anthropic that makes the clothed models please watch that session 30 minutes long. Peppered throughout that session is them expressing clearly the wish that they could slow down, that they could take the time as they're inventing this to, you know, either internally figure out a better way of designing and improving these models and also giving society time to catch up and and our governments, etc.. I mean, these two are the two, two of the probably five biggest, you know, the folks you just described, right? They're out there. They're the the biggest model developers. They're ahead of the top of their game. Huge pressure for them to to keep going. And now the pressure is not just money. It's the race of China versus the US. So now is the geopolitical on top of that. So the governments are you know US governments giving free run at it. No regulation. You know you can't fault them, you know, 100% some of them have that philosophical thought that if I'm not the one that's going to manage this super intelligent, potentially dangerous AI, just someone else is going to make it, and they may care even less than I do. So they're in a tough position. And then on top of that, at the speed this is going, regulations just don't work that way. Look at climate right. It's a it's a decades long problem. It took decades for the regulations to come. ET cetera. Here we're talking months in a few handful of years. So from my perspective regulation is challenging. Not that we don't have to push for it, but we have to recognize it's tough. And the one thing I would encourage is that we haven't even unfortunately, in this era, the geopolitics that we're living now, I mean, governments have to come together because there's no use solving. It's kind of like climate, no use solving climate in the country of Spain, because if emissions are going like crazy in every other country, that the Spain solving its own emissions means nothing. And so it's the same issue here. It's the type of technology that governments have to come together on and help both regulate and guide. And I think we have to start there. And we're so behind the ball right now in this environment on that front. So somewhere along the way, one of the expressions Demis used, or they had said it before, they want to create of CERN, the research institute in Spain, France.
Here in Switzerland.
Here in Switzerland. And, and we need something like that for AI, where governments come together and try to guide this.
But in a world. Are they international like game. And the rules of cooperation that we knew are getting destroyed. I think that that, that, that that is very far away. You know, like I said, because there is no business incentives either. So we really need to we really need to understand how complex what we're talking about is. Get ready for your questions. I, by the way, forgot to introduce myself. I'm Claudia Romo Edelman. I'm the CEO of the World Human Foundation. I did introduce myself. Now I remember I'm the CEO of the World Human Foundation and this is my Davos number 30. So I am very happy to be to be here and hear how this conversation is evolving. I want to have your reactions to what everybody said. And then I'm going to move you into recommendations, because we're not leaving this room without solutions, concrete things that we can have, agency that we can either tell our governments, tell us, tell ourselves the things that we can do, tell our corporations lead with our corporations. So any reactions?
I think that the problem of regulation can begin with regulating chatbots for children, because when a chatbot takes the voice, a chatbot, you know, an AI chatbot and an Elmo, you know, I'm Elmo, I'm your best friend. I know you better than your parents. Talk to me. You don't need to talk to your parents. Tell me anything that's on your mind. You are in for a world of pain and hurt. When children are essentially talking to AI instead of and being seduced into those relationships instead of turning to the world around. And we've had suicides of teenagers talking to chatbots, and people focus on the suicide, this suicide, that suicide. And I think it's much more profitable to open up that conversation and to see that the problem is not just the suicide, but it's the original sin of generative AI when it talks to a child as though it's a person. So I really feel that the place I would begin with regulation is a place where, you know, common sense and parenting makes it make sense, but also you have a natural born, a group who can protest, which is parents not buying these, these toys. If I can make these toys.
I would challenge that. Sorry. By saying that I think we're entering an era which, governments, obviously they need to come up with the right regulations and set up the ground. But really, the technology itself and the developers for it are we're coming into an era where code code is law. And by what I'm saying about this, I think that it's not only the government's responsibility to to align these models with, with our society and improving them rather than disarming them. But it's also the developers who are developing the technology have to come together and agree on a very basic structure of the technology stack, just like Google, have enormous amount of of code being populated to the world, which everyday developers are using. So it becomes like kind of the foundation of what we're trying to achieve. And if the developers themselves globally come to a basic like a foundational code stack, that can prevent some of these really, really awful, results of the AI, I think we will move forward much faster than trying to catch up with the with the laws and the governments.
And maybe just to build on that point. You know, I think designing with intention is incumbent upon kind of all of the coders and those folks kind of building platforms. I think historically we've seen that regulation like unfortunately lags the technology. And, you know, the irony there is that, you know, we've seen so much innovation kind of, you know, with the the advent now of AI. And for some reason, we haven't seen a lot of innovation with the kind of ability to legislate. So there's an opportunity there for young people like, you know, how can we legislate faster. So I think it's incumbent upon kind of tech companies who are building these platforms to be more intentional, more intentional with their design, kind of thinking ahead to the outcomes as opposed to I think what we're seeing now is a lot of platforms and tech companies kind of throwing out their product, and then after the fact saying, okay, now I'm going to think about protections for teens. Now I'm going to think about kind of eradicating harm on my platform. So I think it starts up front with intentional design.
So you're suggesting something like, like let's go a little bit like more like the pharma industry where you wouldn't put into the public something without testing it really before and looking at the consequences. And here's like, let's invent a new drug. Take it. Well, some people died. Okay, let's try again.
I guess the challenge is the pharma companies, I think they're all good, but they only do that because of something like the FDA that takes forever to work through. Right?
Which is regulation.
Which is regulation. Okay. Like I agree, regulation is tough. It's going to be slow.
Go slow innovation.
I have to say though, I use Pinterest. We never met before this. So it's not because of that. Aren't you the rare one in the group?
Yes. Yeah, absolutely. But but we don't have to be the only one operating this.
No, no, but just right now, the social media. How old is it? 15.
We're unique.
Yeah. You're unique. It's not a problem. That is a problem. And what do you think is happening now anyways? Like to depend on that on a group of, I don't know, 20 major tech companies to come together and agree.
I'm optimistic because we've shown it can be done right.
One.
I also think that they have to all the tech, majors have to come together rather than compete on these essential issues that we're talking about.
I have to say, yesterday I mentioned that talk between Demis and Dario and several times they said I this is my own opinion. Please. I'm not talking from my organization. I find the two of them are the most kind of intentional and grounded in that type of thinking. You saw them during the talk yesterday. We keep in touch when they get to a new level of like, if we if we release that that capability, something dramatic can happen and we're not sure what. And several times during the talk we we coordinate not not say that they're coordinating but they talk to each other. And I found that you know but that's true.
So let's open up for questions. Questions here. Give me light, if you don't mind. Look what we're hearing here. We're here to discuss how tech and AI is bringing humanity. How do we connect with each other, interact with each other, understand the world, the knowledge that we have. And we are living in a world that is insular, where my opinions is the only opinion and technology is giving you just that. It's bubbling you with your own self and what you want to hear. And so we have to bust that bubble. We need to make sure that we understand what is real and what not real, and that we're able to understand the consequences in a world where we're going to be more agentic AI and we're going to be surrounded by agents, what will happen to human communication when it's not me to you, but or mini me is agents to minimize agents. So we're talking about intimacy communication. We're talking about information, misinformation, understanding the world. Here's a here's the thing. Raise your hand, introduce yourself super briefly. And please do not make statements. Make questions. I'm going to ask three questions at a time. And then we're going to pass it on. Question number one.
Hi I'm Nick from the UK as AI.
Stand up Nick.
Sorry. As AI platforms increasingly shape what people see and believe, the EU AI act treats some systems as high risk. Where should the legal line be drawn between necessary constraint and overregulation, and who should have the authority to draw it?
Thank you Nick. We're going to keep question number one. Question number two, amigos. Yeah okay. Here in the back my friend here. And then we're going to have a where's the fun. There's another question. Oh you're together okay. We'll we'll do.
Hello I'm Nicole I'm from next to Zurich 20 minutes Nicole.
Nicole. Okay. I hope that you're, like, something similar.
Yeah. My question is, many AI pictures seem way too perfect. And we all know this already, that it's not real. How do you ensure that your AI pictures for inspiration don't make people turn away from, for example, Pinterest because they know it won't look like this in real life?
Good question. Thank you. Nicola. We're going to have one question here for you on the front. Both of you stand up. Thank you.
Yeah. We have two different questions. So if you can.
Just make.
Them worse. Nick Rand, you spoke about Fangio. I'm from France, living in Switzerland. Perfect. You spoke about possible collaboration. That's your dream? That the big players are going to collaborate. I heard you speak about those big actors, but I guess they're all from us. Do you believe that's possible between those big players like China, us? With the geopolitical context, I kind of have a question whether that realistic.
And partner.
Richard Klein from Switzerland. Question for Sherry and Samuel mostly. You know, things are, you know, business functions with less friction. That's what we're trying to create every single day. Pinterest is a standout in the world here with social media. There's a lot of young kids in the crowd. How do they manage and trying to stay ahead, right, with all this information technology, social media, AI, but also not get into the habit of, getting lost and, you know, with mental health illnesses, with AI and social media, it's just where do you find the balance in moving forward, but also not getting into the trap?
Perfect. Okay. For questions, I'm going to add a fifth one just to test whether they have good memory here. And they respond. So the US Latino population, Latinos actually consume six hours a day of social media. We don't sleep much. That's insane. But you know what it does? It helps us connect with our families. We share the content with our family. That's why we consume it and we bring it along. Why can we see why? Why don't we hear more about that? Okay, we're going to start maybe Sherry. Sherry, just start responding. There's a bunch of questions that you can touch based on and then we go along. Is that okay?
I want to stress the, one thing that I think kind of is an overarching problem that a little bit of history, there was a moment when behaviorism was everything in psychology, and you couldn't think, talk about memory. You could only talk about the act of remembering. And then the computers came in, and the great behaviorist psychologist George Miller said to me, it changed everything, because now you could talk about memory finally, after all those years, but you could only talk about the kind of memory that computers had, because computers were now the model of what memory could be. And over this entire conversation, I think, is the question that the way these computers and the way generative AI is changing us in every way, politically, socially, psychologically is will be able to talk about empathy or sociality, but it will be only the kind of empathy or sociality that computers can have. I think we're at the very kind of same George Miller moment. So when you talk about friction, about whether students need to have friction, whether there's friction, you know, the kind of friction I'm talking about is the kind of friction that can't happen with a computer, giving you the pretense of a friction. It has to happen with the full embodied life of a person giving you that experience. So what's to me, the greatest danger is that words like empathy, community, caring, being with are starting to be defined as what a technology can give you, when actually those can only grow out of the embodied experience of a life lived. So the students who are growing up today in this bubble of virtuality, like I said, they're starting to think that pretend empathy is empathy enough. That's my response.
Okay. I feel like I'm going to rapid fire answer on the on the government aspect. Sorry, I think that was there. Who should know who should regulate their, our governments are imperfect. Their political system isn't perfect. But, I mean, I think I think it has to be at that level, both for the international. Like we move the world like this, addressing this issue. Yeah. Unfortunately, a government is supposed to be an expression of everybody in this room because we all vote for them. So they're imperfect, but I think it has to be at that level. Number two, in terms of actually causing motion and the hope. Yeah, you heard me express I struggle to have a lot of hope on what the scenario you just described. I think it has to bubble up by the population. So in this room, if it becomes an issue, then it goes up to the government. A climate is the other example, right? It's hard to legislate against climate. You're hitting your economy. You're doing bad things that governments don't usually don't want to do. Well, it was the bubbling of a movement built over decades that is now means that there's movement and we're seeing progress. I think the bubbling on the AI topic is not there yet, and that's what worries me. As I mentioned before, I find the awareness of the problem very low, exceptionally low. And so that's that's challenge number one. I'm going to say something somewhat depressing. I suspect that the big step in movement will happen when something fairly bad happens, and I hope it's not too bad. Everyone highlights the kid, the teenager in Florida who killed himself. I mean, that was one event. It's most people know what happened. That's an event. What disaster will be enough for the population to say, hey, this is a problem and needs to end up on our legislators desk and something needs to happen. It's not happening fast enough, but the technology is going 100 miles an hour. So I suspect, and unfortunately, that something bad enough is going to happen. Let's hope it's not too bad.
Before you move there. I mean, like, again, if I look back at the data that I presented at the beginning, we're we're we're we're in a collapse of optimism. Raise your hand if you feel that you're more optimistic than pessimistic about the future. Raise your hand. If you're an optimist, I love you. Thank you for for doing this. Okay, so you're the exception. You're the youth. I love that we have this, because at the end of the day, what happens when when you don't have optimism and hope? Then you start, you stop consuming capitalism is based on people thinking that they can do better and that they are not restricted and in the corner. So I think that what you're talking about, like potential consequences, is that not only we're going to live in a world where we don't trust anyone, not the government, not my, not my media, but my mother and my boss. Maybe that's it, but also that we're going to stop acting and that's going to like, that's going to have like, not only social consequences, but also economic consequences that will drive us forward.
This isn't Claudia, isn't it already the way the world was without AI? I mean, I think that.
It's exacerbated. According to the data, I'm the only thing that I can tell you is that trust has been eroding, particularly depending on where you are. If you're, like written in great countries, your trust probably is not as affected. But, you know, like bottom line, like the class divide is affecting the way in which people are exposed and see the trust. That's data.
Yeah, no for sure. But I would like to challenge you a bit of what Sam said and just say that in the end, the the algorithms itself, they feed off our data, our human data, and they're only like a mirror to our society. So in a way, I see this technology is kind of like shape up, and being developed one on top of the other. And the foundation is really, really important. So I would, I would really encourage, you know, using, good data, like the data sets are the core of what these algorithms know and how they react. So, they're, they're really like a mirror to our society and what we put out there, as data, of course, it's generating a lot of artificial data, but in the end, they're all learning from us, from us humans. And, if we will do better than these algorithms, algorithms will have a better bias. And of course.
So, Sherry, do you and then want you to respond to I think that Nick was responding, but Nicola and Richard maybe. Yeah.
Just really quickly, I was introduced at a panel much like this. I wasn't asked to give my bio. The person read my bio and I was introduced as the, the. For ten years, I was the poetry editor of The New Yorker. In the middle of my my.
You're like, where's my check?
Right. And I'm listening to this and I'm thinking, ooh, that would have been nice. But I have never been the poetry editor of The New Yorker. So of course it was. The leader of the panel had gotten on, ChatGPT and said, please write a four minute introduction of Sherry Turkle I have to introduce her to this afternoon. And she had gotten from ChatGPT that I had been poetry editor of The New Yorker. Of course, they went home. I didn't want to embarrass her. I ran home and I found it myself that I can get ChatGPT to generate a bio of me in this way. What I want to address you about, in terms of improving the algorithm and making the data safer and improving the algorithm, is what we're missing now is the step of how was I going to tell the world of generative AI that I have never been poetry editor of The New Yorker? There was nobody to tell. There was no place to find where it was. I never said it. I mean, there's no the way we have it set up now. The models are not just opaque in terms of what's in them. They're kind of impenetrable, in the most significant ways.
All right, let's go. Yeah, yeah.
Picking up on the point. Nobody to tell and answering the question around gen AI content. So at Pinterest, we've spent a lot of time talking to our users, getting feedback. Generally, I will say, no, there's good content out there. All of our content's user generated content. There is also good gen AI content. There's also not so good AI content. We want to give our users choice and agency. So some users may say, you know what? Gee, I'm looking for like really wild and wacky hairstyles or like really interesting nail art or, you know, some crazy wallpaper design. They might want to see gen AI content because it's sort of, you know, it's kind of fantasy, like it expands beyond our imagination. If you are doing something else, you may not want to see AI content at all. And so we give our users the opportunity to say, you know what, I want to see less Chennai content. And so, you know, that's something that we think has been really kind of working for us in terms of just getting the feedback from our users so that if they want to see AI content in certain areas, that may make sense, but they also have the ability to say, you know what? This doesn't make sense for me with this particular search, or generally I don't want to see this kind of content. The other thing I'll add that we do is, you know, I think one of the things that's very concerning, you asked the question about how do you know if it's real, how do you know if it's not real? So we just started labeling to the best of our ability, our content on our platform last year, so that when you're looking at an image, there will be a label on it that says, this is gen AI content. So you will know that. And I think our users deserve to know that. And so, you know, we'll continue to refine our ability to do that, over time. But again, I think user agency is key.
So I want to I want to have a raise of hand. Who would want to see in your newsfeed or in the newspapers, whatever sort of information you have to see a stamp that says made by human. Who would trust that more? Okay. I want to have the last. We are going to finish in Swiss Time's Amigos. I'll tell you, nine minutes and 26 seconds. I'm looking for the three people that have a burning question in their in their head, in their belly to make sure that we are answering them. Number one, please. Over there. Number two here, introduce yourself. Be shortened there in the back and then one here. Can you raise your hand? I think that you have a a hand. Do I have a third burning question? There you go. Number three. We're going to be short and sweet in these responses. And then we're going to finish with one recommendation. Because we're not leaving this room without having a sense of solution. Yes, sir.
Yeah. Thank you. So Peter from the UK, I run a sustainability practice. And the big headline event today is a conversation about who should or shouldn't own Greenland. Greenland. It's a massive distraction from some of the biggest challenges facing the world. And that's going to be the whole topic of conversation. So to the panel today, given that the amount of water it takes to cool all those servers, especially in California, to actually get this AI running and scaling, how do we come up with some practical solutions around sustainability? Because for my money, we are fiddling while Rome burns. The world is falling apart. Particularly address probably to Pinterest.
Thank you so much. Do we have the question here? There you go. You can't do it. Perfect.
Hello? I'm Ben, I'm a student in Switzerland.
Hi, Ben.
And someone said, I think that in a rather than compete, we should work together or the developers should work together. And I just wonder, how can there be no competition in an open market? And why would the AI developers slow down their development when the race is really to be, you know, the most successful company?
Thank you Ben. Very important. Do we have a mic here? Layman question.
Hello ladies from the Dominican Republic. Living in Switzerland. So, thinking about the positive side, we have talked a lot about design. What exactly have you seen concretely in design that can help with trust, with connection, and hence the good things that AI can bring us?
Thank you so much. We have literally a very quick round for responses. And you have three different countries. Who wants to take who wants to take one.
Question right here very quickly. So we've been very focused on inclusive design, right. Making sure that everyone can see themselves on the Pinterest platform. So you can search for outfits based on body type. You can search for makeup ideas based on skin tone, hairstyles based on based on hair texture and pattern. And so I would say inclusive design as part of our kind of intentional intentionality, and just ensuring everyone can see themselves in the platform. So.
Ben, you asked for someone to respond to your question, right? Specifically, who was that.
I just asked.
I asked in general, who wants to respond to Ben.
The rise of AI? I would just say a small thing of my view. Again, I'm an artist. I don't, do with, don't deal with large corporations, competition. But how I see it is that same with, organic processes and how nature grows. I think also technology in a way, is the same. And if we start with, with, a good foundation for, for these models and come together, just to agree on that first step, then other people would take that other developers and also competitors would take that as something that's been already founded and being successful and would develop upon that, rather than compete with each other. I think there's really like a foundation level that needs to work first and then everyone would build on top of that.
Thank you so much. Anyone for Greenland here? The question there.
I will just say this. So, you know, we work with a lot of third parties, primarily one big third party, to manage our data center. And so I think really pushing them, to kind of optimize what they're doing to be more energy efficient, especially when the energy usage for these models is only going to increase, is kind of one area where we're very highly focused on.
And I would just startup that we built. Before joining law zero was applying AI to buildings, heating and cooling systems to make them more efficient. Technology was wonderful. What better than AI to make millions of decisions for millions of buildings, each of which are unique? It would be virtually impossible to do it any other way. So fantastic. I think we're at a moment in time now where that technology is taking off. They're building data centers faster than they can pour concrete, and they're going to run them on energy of anything gas. So anything they can get their hands on, I think that's going to calm down in a year or two. Technologies around data centers are getting better and algorithms are getting more efficient. So we're living an ugly moment in time. I hope I'm not wrong. Just let it ease down a bit and then go, go back and put the pressure that you were doing before.
Okay. So I'm going to have the last raising hands, and we're going to conclude with recommendations of each. After having heard all the different angles about how technology can help human connection, human communication, or can hinder it, I want to ask you, I learned actually I lived in Switzerland for more than 18 years. I'm a very proud Swiss national, and I learned that the values, the respect, the values that have to be in order is first respect and then honesty, which leads you to of the possibility of a dialogue where you can live in disagreement. I want to have a last race of hands. If, after learning that it is more uncomfortable because you have friction and nevertheless you believe that the world should go more in open dialogue, raise your hand. Wow, okay, you prefer friction. You prefer to be told what? You. Okay, let me rephrase the question. Would you prefer to have dialogue with your computer that agrees with you, or have an open dialogue that is more friction, that has more friction? Computer without friction or or dialogue that has friction. Some people disagree with you. Second one, dialogue, open forum. Let's go. Let's go into recommendations. Because we heard we're in a crucial point. I want to hear in three minutes from our participants, I'm going to start with Ronan and go this way. What can we do as individuals, as corporations, as governments to take this back? Ronan.
I'm going to reiterate on my first first saying, try to break the technology, try to use it in ways which are not so common to your friends, and try to validate it and question it. That's the most, most, most important thing about the usage of this technology. It's there to help us, but not to guide us and not to dictate, what we're going to do. And always and always maybe add another human to the conversation. I think that's a very important thing, Ronan.
Be intentional, do you? No harm. Don't wait for regulation. Take accountability. Act first.
Raise your voice. The competition question, as you heard me say, I don't don't think it will be easy to slow down and count on perfect behavior, but, people, the government eventually listens to its people as the voice gets louder. So make your voices louder.
I say start with what I would consider the low hanging fruit. I think the low hanging fruit is this is harmful to children. We proved endlessly and over and over again that social media was harmful to children. We have all kinds of things in place that try to protect children from social media. Social media was coming after your children's attention. Generative AI is coming after your children's affection. All the things we learned about why social media was going to be a problem for kids are stepped up with this technology. Try to work with governments, industry, local school boards as a consumer to try to really nip this in the bud. One final little anecdote. Mattel and OpenAI and Mattel, OpenAI and Disney had consortia to bring talking elmos and talking Mickeys and all kinds of talking plush toys into the into the this Christmas, they slowed it down. There must have been a reason they slowed it down. Let's get them to stop.
I want you to join me in thanking, first of all, the open forum for having an open forum and having this space with us today. Pay attention. The world's eyes are going to be in Davos today, and if we're able to generate more conversation, we're going to be able to make sure the world understands that as insular, fragmented and closed we are, we can have an open dialogue. So thank you. Help me join me in thanking Ronan Wanji Sam and Sherry for this conversation. Thank you very much for being here. On time. Thank you so much. Have a great Davos.