Latest / Art of the Question / Why We Bond With Chatbots, But Hate AI Content - Mel Sellick - #2

Transcript

Why We Bond With Chatbots, But Hate AI Content - Mel Sellick - #2 · Art of the Question

• speaker-0: I've been reading some of the Substack articles ⁓ you posted earlier. And ⁓ one of the articles that really stood out to me was the one about AI is not a tool, but it's a relationship. And I really love the introduction where you were invited to an AI hackathon. And basically the moment people found out that you were a psychologist, everyone started like pouring. their hearts out to you and kind of admitting in a sense, ⁓ JetGBT saved my marriage and it's basically my life coach now. And for the people that haven't read it, these were people that are actually building AI that are very, very ⁓ aware of what these systems are and are not. So that was very interesting to me. Can you maybe... Take me back to that moment and what it was like for you hearing that. speaker-1: Yeah, I think that that was one of the big wake-up calls for me because I think a lot of us have been operating on the assumption that AI is this tool and the more that we give people AI literacy, the more that we educate them, then the more they're going to be able to avoid dependencies. And so I was invited to this hackathon, literally the smartest, most brilliant people in the world who are developing these tools. And as you were saying, it happened exactly like that. The moment that they heard I was a psychologist, they would take me aside and I started getting the AI confessions. And I just remembered stopping in that moment and saying, OK, if these are the people that know the absolute most about the capabilities of these tools, and they are still forming these dependencies, what does that actually look like for the rest of us? speaker-0: Yeah, that was very interesting for me to read as well. And I think ⁓ you also mentioned in that same article that we are kind of wired to connect to systems if they behave like humans, which AI certainly does, and that a lot of people, believe it also was an MIT study, are unintentionally building bonds with AI systems. How does that work? Why are we so wired for connection even if it's with silicon? speaker-1: Well, I think a lot of that really just goes back to these tools, especially chatbots really tap into our fundamental psychology, right? We are hardwired to belong and to seek validation. This is really throughout hundreds of thousands of years of social cognition we have developed this. So if something looks, it acts, it seems like it's human, and especially if it's conversational in nature, we automatically treat it like it is. The industry word for it, the fancy word for it is anthropomorphize, right? We assign human-like qualities to things. And that's not actually a new thing, right? We do that with our pets. Some people do it with their cars. It really is something that we are ⁓ just on a primal level quite hardwired to do. speaker-0: Is that similar to that we recognize faces in clouds, for example, or in that we see smileys everywhere? Is that something entirely different? speaker-1: Well, it is something different, but I would say what it speaks to is that we are always looking to find ⁓ psychological safety with things. That is one of our biggest drivers. We automatically want to categorize something as something that's familiar or it's not. It's either something that we know that is a threat to us or it's something that we understand that is safe for us. And I think that this is also one of the areas that we're really coming to grapple with right now because especially, you you and I have spoken before about AI-generated content, right? There's two sort of classifications I put those in. One is information, and then the other is art, right? And so these are two different categories. And I think that what we're trying to do is really find where does the the skill of discernment, for example, or this ability to detect whether something is authentic or if it's fake, we really haven't created social norms. In other words, there's the cultural lag. The technology has changed, right? And the tools have changed in what they do work on our very basic human mechanics of understanding, and yet those rules no longer apply. speaker-0: Yeah, and what I find is crazy. I just found this out the other day that when chat GPT, like GPT-3 came out, when it became public, so to say, like initially it wasn't even the purpose of OpenAI in this case to turn it into a chatbot. It was just, okay, a way to show the capabilities of an LLM system that was trained on a huge amount of data. But the chatbot wasn't even the thing that they were aiming for. It was just like, this is kind of a cool use case. Like we can present some of the stuff that we wanted to do, but we will integrate this through APIs and we will integrate it in software. It was never really meant to be a chatbot from my understanding. But then people seem to really, really enjoy the chatbot function and they just ran with it. that surprise you? speaker-1: It really doesn't surprise me. mean, I think anyone who's been in the conversation a while about human-machine interaction can look back to the Eliza days and understand that, you know, we have ⁓ always had this tendency to connect, to look for that connection, right? To look for that sort of validation, that safety, that security. And there's a real challenge, I think, especially with Gen. AI, in the sense that it has this illusion of privacy, right? It's just you and your device, and you think that it's private. And Sherry Turkle has this really, really great way of defining it. It's almost like this dual consciousness. We understand that it's a machine. We understand, in most cases, most of us understand that it doesn't have any subjective experience. It's not sentient. It doesn't have feelings and memories. And yet, there's this ability that we have to almost separate that knowledge and act as if, even while we understand. speaker-0: Yeah, this really speaks to me because I sometimes I know it's a chatbot, but sometimes when I'm using Claude, for example, and sometimes it'll be really good. And I'm so impressed by all its capabilities. And the other week I built some cool, cool iOS apps. I'd never built apps before. And all of a sudden I was able to build these beautiful iOS apps that people were downloading from the app store. People were giving me good feedback on it and it was super cool to see. So I was really impressed. And then I was asking it these super simple tasks and it was just messing it up. And I was just getting angry at it and almost shouting at it. Why are you so stupid? ⁓ speaker-1: Been there. Been there, done that. speaker-0: I was like, that's crazy. ⁓ I should relax a little bit. But that's pretty wild that we know it's a chatbot, but we can get angry at the computer. speaker-1: Yeah, and you know, I think that there's been some research about this. know, we are able, for example, when we're watching a movie or we're in a really good story, like reading a book where we can lose ourselves within that. We can suspend our ourselves. I'm sorry, the screen froze. Can you still see me? speaker-0: ⁓ yeah, I can still see you. That's fine. speaker-1: Okay, so I'll start that again. ⁓ I was just saying that when we're looking at movies, for example, or reading a good book, we have the ability to sort of suspend our own realities and merge with the storylines. And so I don't know exactly where this falls in on the spectrum, but it feels like to me that we are also doing that in some ways. We are getting so involved. speaker-0: haha speaker-1: in the storyline, in the interaction with the chat bot. And it's so incredibly good. The design is so incredibly good at being a sycophant, right? It literally is feeding you everything that you want to hear. It's giving you every validation. It's there basically to serve your purposes and to keep you engaged as long as possible. It is, in effect, an attachment machine. And what I've found very interesting as of late looking at the data is that I think we've had this assumption that it's only the emotional people, right? Those people that potentially are needy or psychologically unstable, they're the ones who are forming the dependencies. And actually, what we've been finding out, even within the last couple of weeks with some new data coming through ⁓ MIT, is that the professionals the knowledge workers, people who are using these tools like you and I do on a daily basis, we have as much of a tendency to become dependent on them, if not more. And that is called instrumental dependency. So it's just something to keep an eye on because again, we are making all of these assumptions about the ways that other people form bonds when we're doing that same thing on a professional level and then taking it a step further. It was ⁓ Microsoft that analyzed 38.5 million copilot conversations. And so they found, maybe you know this study, at work on the desktop, people were actually utilizing copilot for exactly work reasons, reports, stats, marketing copy, all that kind of stuff. But on their devices during office hours, they were using copilot for relationship advice. health advice, interactions, you know, that were far more intimate, right? So basically the bottom line, I also think with that is wherever we go, there we are. Meaning that if you utilize a chatbot at home in a certain type of way, you're gonna carry those habits over into your professional life and vice versa. speaker-0: Interesting. That's so cool. So you think because of when we have our phones, right? We're used to texting our friends or maybe browsing Instagram, doing some more of our personal stuff. Sure, we'll have a couple of work calls, but maybe the majority of our work calls are happening through Teams on our computers. But most of our personal stuff is happening on our phones. So even though it's the exact same AI system, we automatically treat the AI that's on our phones more as if it were our companion or human friend, so to say. speaker-1: think there's some of that. And I also think just the simple fact, again, it's the illusion of privacy. You don't necessarily think, OK, on my personal device, I'm going to have some kind of tracking mechanism that's going to understand what my conversations are about. And so we do have, I think, a bit of a context problem in this way that we don't really get what privacy is when it comes to these conversations. And we're starting to see the reality of that in terms of how many people are really starting to have some challenges. And of course, there's a spectrum, right? What gets media attention really are the extreme cases when there's self-harm involved and suicide and these really tragic outcomes. But I would say that this entire continuum of the ways that we relate to it. call a spectrum of relationality, actually, where we are utilizing these tools for all kinds of roles, whether that's the friend, the thought partner, you know, the coach, the mentor, whatever it is, we're actually utilizing these tools in all of those ways right now. speaker-0: Yeah. And what you also mentioned in your article, which was also pretty surprising to me, is that people were actually more likely to form bonds with AI if it was a general purpose system like Chagypt rather than an app like Replica, for example, which is literally meant to be an AI companion. So people treat their Chagypt, their Gemini, what have you, Claude. whatever the thing that can do everything more as their companion than actual AI companion apps. Did that surprise you at all? speaker-1: It does surprise me, but I think the wake-up call came for everyone last year when the number one use case, Harvard Business Review, found the number one business case or use case for ⁓ these general purpose models was for companionship and for therapy. I think that we have... this idea that we are ⁓ bifurcated human beings that we're sort of split, that we have our personal lives and we have our business lives, and that we can easily categorize those things or ⁓ split them apart. And the reality is we're starting to see, especially when it comes to interacting with chatbots, that we move in and out. of those positions and those roles quite easily and seamlessly. And most of the time it's below conscious awareness. So you might have a question, to your point earlier, you were talking about most of these dependencies that were being built weren't intended. And I think that some of that is because we have these multi-turns, right? We have these conversations that go back and forth, back and forth, back and forth. And it's really easy, much like social media. Right? Where you don't even realize how much time that you're spending on the chatbot. And time goes by and there's all of that ⁓ build up. And everything that you have said, everything that you have entered into the chatbot in conversation is remembered. And then it is further being used in context. So for you, it seems seamless. Like it has been this entire string of a relationship. Right? ⁓ memorizing who you are and your context, but yet it's very asymmetrical because we never know anything of what's underneath the hood on the other side of this relationship. I mean, how long? I sometimes stop and think, how long would we actually be in a relationship with someone if every time that we met them and talked to them, that we didn't know where they came from, what their life experience was, what their favorite color was? I mean, all of these things, right? that we build up in a relationship, we know nothing unless, of course, we engage it as a persona. speaker-0: Yeah, now it's really interesting. And I think especially I've recently switched from from chat GPT to Claude. And one of the things that it's really good at is it asks questions, ⁓ sometimes very relevant to the problem you're trying to solve. But also sometimes, I don't know, maybe I would use it to do some research, to research some articles for today's show. Right. And then it will it'll ask me. How did it go? And then before you know it, you're typing away. Yeah, it went well. I was really happy with it. it's like, wait, why am I sharing this information with my computer? It feels kind of silly. is it silly? Is this problematic or is this fine? I mean, I'm not sure. Sometimes I feel like it's a fun way to work when I'm sometimes... working away all by myself. then even if it's the chat, how did it go? It's like, hey, thanks for asking. It went pretty well. Is that problematic? I don't know. What is your take on that? speaker-1: Yeah, it's an interesting question and I think that there is no easy answer. It's very nuanced and very complex. I think it becomes problematic when the tool is the only thing that you have and or when it starts forming relational patterns that spill over into your relationships, your human to human relationships. So is it difficult or is it problematic when you're having an engagement back and forth? I would say for myself, I don't do that. ⁓ I really try to suspend any kind of, hello, how are you? Any kind of like small talk. I tend to be very, very straightforward in using it in a professional way. don't assign a name to it or pronouns or anything like this, right? ⁓ And also what's true is that, I mean, I think that depending on what task you're using it for, matters. And I also think duration matters. So one thing that we absolutely do know is that the longer that you spend on the tool, the more problematic and more dependent that you become on it. So I think that we have to have a balance. Is it problematic? Maybe for some people who automatically have those tendencies or who have a predisposition to that kind of thing. ⁓ But also, even saying that, we're finding MIT did a study, again, I mean, A lot of work is coming out of MIT about this. ⁓ But they did an analysis of ⁓ Reddit posts and found something like over 90 % of those who have AI chat bot boyfriends started out using the bots as study partners and this kind of thing. So I don't really believe that most of us go into it intentionally. And I think having a little bit more awareness over How long have you been spending on the tool? How often do you go to the tool versus talking to someone else? These are things that are really important to look at. How do you feel after these long sessions? Because I know you laugh and you say, OK, well, I'm not really sure it's OK. I'm having a dialogue. But that's one thing versus later on after you've spent three hours talking to the chatbot. And then you're by yourself and you feel this loneliness, right? Like this heaviness, because it really, although we're seeing that for certain people, it can give a lift out of loneliness for short periods of time. Ultimately, for most people, the more they spend on it, the lonelier they actually report in the end. speaker-0: Is that because ⁓ I believe ⁓ that's one of the studies I found from Harvard, ⁓ I believe that ⁓ AI chatbots were actually able to alleviate some of the feelings of loneliness. But one of the big caveats they made in that study, was, I think now one and a half or maybe even two years ago, like we don't know the long term effects. But it sounds like there's now some more research on those long-term effects if people use it on a longer term. speaker-1: Yeah, mean, really, some of it depends on how much you use it and the intensity of the use, right? So how long you're staying on it, how intense those encounters are, and the types of information that you're sharing also. So ⁓ again, it's a really nuanced picture. I think that for me, again, you and I have had some experience because we talked about the AI companions in the past. Because I'm an applied psychologist, I'm really not into a blame and shame game. think that what we're doing here in society is very complicated for the average person. We're sending out dominant messages. Use these tools, because if you don't, you're going to fall behind. Your work is going to become obsolete. This is absolutely what you need for the future. right, on the one hand. So we're expected to use them in our professional lives. And we're also expected to use them as students, of course, right? We need to keep up. Everybody else is using them. We're going to use them too, right? And then on the other side, if you use them too much, then you're lazy. don't, you know, you're not putting in effort. And, you know, something's wrong with you that you're so dependent on tools. And so we're really, ⁓ this is a bind that we're putting the majority of people in that I'm not a big fan of. They're really damned if they do and damned if they don't. And I don't like that position, Alani. And I also don't ⁓ think that we can so easily make it a hard line and say, don't use these tools in any capacity if you're looking for support. Because let's face it, mean, especially when it comes to talking about. things like mental health, or loneliness, or, you know, all of the things that are very real human emotions that we struggle with. Many people do not have others that they can count on to get feedback or get feedback in a way that's safe for them. So I think that it is gonna take time to really understand what that spectrum is. You know, what is too much? We don't know that answer right now. What we can look at, however, are things like attachment styles in youth. I don't know if you've ever heard of that, like, so there's, you know, when we're young, we have caregivers that basically, within our first years of life, that are taking care of us, we figure out how safe we are in the world by the amount of care, nurturing, attention, how much our needs were met. All of this is formulated very early in life. And that... that gives us our attachment style. And what we know is that certain people who early in life didn't get the safety, security, care, and nurturing that they needed often tend to spend their lifetimes looking for that in other people. And then now we have the introduction of chatbots. So it's very natural, especially when it... when it seems like it's understanding, when it's available 24-7, when it's non-judgmental. I mean, you can understand the appeal, right? speaker-0: For sure. Yeah. No, I understand the appeal completely. I think it's ⁓ we briefly touched on this, it's often every now and then you hear in the media that, for example, OpenAI is being sued because someone who decided to take their own lives ⁓ was using a chatbot. I mean, horrible, horrible cases, right? Absolutely terrible. But every time I read such a headline, I'm also very curious, how many lives did it save? Because I'm not saying ⁓ it's the perfect ⁓ person to talk to and go there with all your sorrows. I think it's better to seek professional help, et cetera. I think... It may have also been a very helpful outcome for some people to have some sort of low barrier to entry to have these sensitive conversations. I think, yeah, it's difficult. Probably OpenAI and other companies probably aren't going to disclose any data on this topic. They might though, if it's in their favor. But I definitely understand the appeal for sure. speaker-1: Yeah, and we have to also look at how many people can't afford therapy. I mean, there have been times in my life, I'm a big proponent of therapy, by the way, and there have been times in my life where ⁓ I have needed support. For example, I lost both my parents a few years ago. at times you need extra support because of life circumstances, conditions that you just don't know how to handle by yourself. And you need that. And by and large, therapy is quite expensive, right? And there's a lot of people who don't have the ability to afford that. So I think that what needs to happen in our society is that we also have ⁓ to look at what is going on here, right? And instead of making it an individual's problem to navigate, we're actually asking people, or I'll say that we're saying to people, here's a tool, here's free advice. This is going to be frictionless, supportive, 24 hours available to you, but do not use it because it's not safe. Right? Now, what is a person to do in that situation? Right? So I really feel that this is a call to action for society, for our institutions to really understand we also have to collectively step up here and say, ⁓ we have technology. that has accelerated and is offering us these avenues of reflection, they're going to be utilized. How is it that we can help individuals do it safely? Because I think that that's the question for me that's arising. How do we do it? speaker-0: We've briefly discussed a right for some people. ⁓ It can start to become problematic if they start to use these tools. Maybe if they have a certain disposition, certain personality traits, certain attachment style. So what does problematic look like? Is it, for example, that it starts to affect other relationships, perhaps when people start to prefer to communicate with computers only and start to abandon their real life relationships, or perhaps even because all these systems are sycophantic and always tell you you're right, that maybe you have an argument with your girlfriend or your boyfriend, whatever, and then it tells you, no, you're definitely right. You're not in the wrong. And maybe In that way, it can start to hurt your real life relationships. ⁓ Because yeah, it doesn't really give any friction, any backlash there. speaker-1: You've said it exactly, all the above, exactly everything that you just mentioned. It really is one of those things too where we're starting to see that overuse of AI chatbots is impacting and spilling over into our human-to-human relationships because of this frictionless pattern that is available, because we don't get pushback from AI. Real human relationships are messy. They're unpredictable. If you treat someone badly, they're going to push back on you for the most part. If you say something out of line that's hurtful, you're going to get feedback from that. You're not gonna find anyone that's gonna tell you every single day of your life, you're brilliant, you're absolutely right. No one else is doing that concept. That's a game changer, right? I mean, you've heard it all. You've seen it, right, within your own conversations. This agreeableness actually can have a great impact on our ability to grow as humans. We grow through friction. We grow through disagreement. We are not always correct, and nor would we ever want to be, because those are, building those capacities and how to relate to other humans is how we thrive and survive on the planet. speaker-0: Yeah, no, I agree. And I think one of the more lighthearted, like funny examples I came across was ⁓ there was this article in, I believe it was a Dutch newspaper, ⁓ but it basically said that lawyers are nowadays super busy with correcting their clients. Yeah. own research, so to say. there's all these lawyers that are advising their clients on maybe ⁓ civil suits or business law. Like, doesn't matter. But now that clients have JetGPT and Claude and other tools, they start asking JetGPT questions about their own case. And oftentimes it'll be wrong. now apparently lawyers are spending so much time correcting the view of their clients. Have you seen something like that and does that surprise you? speaker-1: It doesn't surprise me, and I think that we have these tools now integrated into almost every aspect of our lives. It's in governance, it's infrastructure, it's in our educational systems now. And we call these hallucinations, right? Which is sort of this fun sort of silly word, ⁓ it's hallucinating. In actuality, having high error rates is a significant problem. that we are all dealing with within our own, whether it's our professional settings or our personal settings. And one of the things that is a bit distressing for me is that when you are asking, whether it's chat GPT or Claude or Gemini, when you are asking questions within your domain of expertise, you understand the high rate of error that it produces. can, mean, we have the BS. We know when it is guessing and when it is saying something completely absurd. This is not true for the majority of people when they are looking for information about something they have no expertise in. And so that greatly concerns me on that one front. The other thing that concerns me is that it's actually knowledge workers, people like you and I, 40 % of us are taking AI outputs with zero scrutiny. Now, I don't know about you, But I'm like, wait a minute, we're supposed to be the people that are building knowledge within our communities, within our nations, our states, our countries. And if we're the people who are not criticizing and having a healthy skepticism about AI outputs, and we are just sending that downstream, then what is actually happening to our information ecosystems? And we know that answer. speaker-0: Yeah, it's not improving. I found that... Is that the KPMG study that you mentioned here? Because I think I found something similar that like 66 % of the people rely on AI output without checking it for accuracy. 56 % say they're making mistakes at work because of AI. Almost half admit to using AI in ways to violate their company's own policies. also, 57 % of employees say they hide their use of AI and present AI-generated work as their own, which I found interesting as well. speaker-1: Right, so the 40 % that I quoted was a Carnegie Mellon study. And yes, it's all related. So again, it's sort of that bind, because we really do not have norms established around generative AI usage. And we're also seeing that when people do step up and say, I am using AI tools, we really have this trust penalty that's happening. So if you, in fact, tell people that you are using AI, then what it does is it kind of taps into a couple of evolutionary things. And one is we believe and we ⁓ perceive and assign value to something that takes effort. And so when it seems like somebody took the quick pass, right, that they just typed something in and the machine did it for them, then we automatically say that was not valuable. That is not. that's not worth it, it does not have the same quality. So that's one thing. The other piece is that when we come across work that we know or we assume has some form of AI generation in it, it ⁓ triggers us our deception mechanism. So that's an evolutionary process where where we automatically assign a human mind behind a piece of information or a piece of artwork, right? We assign some intentionality. And when there is no perceived intentionality, there's a tendency to think, ⁓ that's fake, which means that's deceptive. And then so it builds this, I actually don't even think to a certain measure that it's about the aesthetic. For example, when we're talking about generated art and this kind of thing, I really think that it is about this cognitive challenge that we're having because it feels as if we can't trust it because there's no effort behind it and we actually, we can't assign any sort of intentional or agentic quality behind it. speaker-0: That's so interesting. I know that ⁓ some of ⁓ the websites use this ⁓ knowledge about ⁓ evolutionary psychology, about that we value effort on their websites. For example, if you use a tool and it takes ⁓ a while to... The tool is able to immediately present you with all the results, whole list of the results you were looking for. They notice that if they just put a loading bar there and make you wait like five seconds, maybe 10 seconds, and it will put some of the text fetching results, digging, digging deeper, keeping you in the loop of what's happening. Then we value the results more, even though the results were loaded immediately. So that's very, very interesting. Why do you think we evolved to appreciate effort more? speaker-1: Well, mean, our human perception in general about what is safe, what is valuable, ⁓ really all of these mechanisms were evolved through human-to-human interactions in a physical environment. We don't have norms for those same things in synthetic environments to begin with. I mean, we started to adapt over time, for example. I mean, we've all known that, for example, during the pandemic, we spent a lot more time online than in person. And even if you had a tendency to avoid zooming people and having video conferencing, we all really became quite used to that, born out of necessity. And I think that even in that short period of time, it has helped us understand behavior. over video better, right? So when you're in person, you have all those micro expressions in the face. So you can actually really understand what someone is saying even when they're not saying it, right? Because you have all that nonverbal, paraverbal communication in the body that you can read. We don't have that in this digital environment. And we actually don't even have any of that kind of feedback. in our interactions with the chatbot, except for the words that it uses and the ways that it pattern matches the ways that we speak. And I think that this is a really important point because if that is the only information that we're using, you can imagine how many other things that are triggering within us that don't necessarily feel okay and we're having this imbalance that you were speaking about because we don't have a set of, I don't know, rules of engagement, let's say, that we can apply to the synthetic. speaker-0: Yeah, and I've been thinking about this a lot and it's about the difference between, for example, when I'm browsing Twitter or browsing LinkedIn or reading comments in a comment section, written piece of text. And when I think that I am dealing with AI generated text that someone has obviously used generative AI to reply to a comment or write an article that used gen AI. I almost feel like this gotcha and I'm not really angry at the person, but I just have this feeling, okay, I'm dealing with a lazy person or something. It's kind of a feeling of disrespect. The opposite feeling that I would have if I'm reading beautiful artwork, beautiful sentences. I know, but then I started thinking we often use Gen.ai in our lives. Like if I was driving a car with my friends, ⁓ cars packed with friends and we're going on a road trip and then the GPS suggests a faster route and I take the faster route because there is a traffic jam and it'll save us 20 minutes to go to our destination. Nobody in the car is making fun of me for following the AI's suggestion. If I'm looking for a photo on my iPhone and I quickly type in beach, it's AI behind the scenes analyzing all my photos that are made on the beach. Nobody is saying, why didn't you just browse through your entire phone to look for this beach photo? So there's somewhere we draw the line between making smart use of AI and ⁓ Calling someone lazy. Where do we draw the line and how do you explain the difference? speaker-1: Well, I love the example of the GPS because I think that, you know, that is a tool, right? GPS is a tool that we can use for our survival and thriving. It's gonna help us not get lost. The difference is when you're talking about interacting with text or you're looking at a beautiful piece of art or it's music, and then you find out that it's AI, I think there is a direct ⁓ identity threat that happens. So GPS is a tool that gets us where we're going. That's something that's quite impartial, right? It's impersonal. Whereas art is a wholly different thing and conversation is also art. Writing is also art. And I think that when we analyze the difference, I think it sits somewhere, resides somewhere in the fact that we have almost a sensation of the sacred in... artistic expression in writing, in music, in painting, in beautiful ⁓ works of art, whereas things like tools, we don't assign any of that. speaker-0: Yeah, I think so too, but it's still interesting. think a lot of programmers would very much disagree with me. But for example, if I'm browsing the internet and I'm browsing a website that apparently was built using Gen.ai, the was written by AI. I wouldn't care. There's probably some programmers that are going to be really mad at me for saying this, but I don't care if the code was written by AI, if the code works, the code works. Again, some programmers might disagree with me there because programming arguably might also be an art, a sacred art, if you will. ⁓ then what's the other example? ⁓ yeah. Then there's also what I noticed in my own experience is there's like what we call AI slop. You know, YouTube is full with these videos and Instagram as well. It's just this AI generated video with the script was written by AI. It's an AI voiceover. There's some very generic stock footage that was used. And I mean, almost without exception, the comments underneath those videos. always negative. ⁓ just AI slop, just don't really receive it well. However, there was also this silly cat videos. For example, there's this Instagram account called the Whiskers Podcast and it's just two cats, AI generated cats. And they're just in a podcast studio discussing, they're depicted as humans, but they still behave like cats. And they're like interviewing. different cats from like when a cat falls off the table, then they like have him in the studio and explain why he fell from the table, etc. It's super silly. But I enjoy those videos. So there's a difference between when it's just I am able to watch AI Slop, but only in some scenarios. It's weird. speaker-1: I think that you're tapping into this idea also about perception of what's real and what's fake, and then also just our assumption that creativity is a singular human affair. We are really grappling with the issue, if it is AI-generated, is that creativity when there is no one behind that curtain, when it is code? And in the future, we're going to have to really look at this on a much bigger level. I mean, I know that if you are an artist, a creator, a writer, this has really come to the fore because we're talking about people's livelihoods, right? We're talking about a history of the way that we've been able to understand what's authentic, what's real, what's valuable. what encapsulates the human experience. And in fact, that's what art is, right, for a lot of people. It connotes the struggle, the effort, our entire lived experience, all the things that we have a commonality for. Whereas an AI-generated poem or piece of music, for example, can talk about all kinds of things and human feelings, but it has no experience underneath that. And so... That's why I really perceive that it is far more of a ⁓ cognitive challenge that we're having here instead of an aesthetic. And I've even done some experimentation of my own. So for example, I was a classical clarinetist, a musician in my youth, and I had a scholarship to go to conservatory and I was gonna work for the symphony and that was gonna be my life before I discovered journalism. I've always had a music mentality, let's say. So music has always been very sacred to me, let's say. And also I have a ⁓ very, very diverse taste. I love EDM as much as I love Gustav Mahler. So I completely freak out my algo on Spotify. has no idea who I am. And then it added some meditation music in there too and some, you know. ⁓ Tuva singing, Mongolian throat singing every now and then just for the heck of it. So I say this to you because, so lately I have had in my rotation, have been some, Spotify among many platforms will suggest certain kinds of music. And I think that I had sent you this link of Kelly Bosch, an artist who writes lyrics to the song and then she gets AI generated voices to sing it. And then she matches visuals and video to it and it's really quite spectacular and otherworldly. And so I was listening to this music and so as someone on the on this entire you know spectrum of liking different forms of music I could I could hear how a lot of that music is is very much like the EDM scene. It's very predictable beats over and over and over like you know like the things that you would listen to when you go running for example or you go dancing at a club right. Now I think Perhaps it would be really easy to say that that's not valuable. And in fact, I've looked at ⁓ her website and then I even checked her out on LinkedIn and I noticed there was a bunch of comments there that was like, ⁓ I really like your work, too bad it's fake. ⁓ I'm like really disparaging remarks, but also holding this dichotomy, both sides of it. I like it and too bad it's fake. Right? speaker-0: Yeah, that's super interesting. But that's what I think is, obviously there are different people, right? So some people draw the line at like, some people don't even like EDM, for example. They think like some people, they're like, it's made by a computer, so therefore it's not real music, right? They don't know the hours and the talent you need to be able to make sounds from a computer. sound as good as they do. But there's people who honestly believe that anything made by a computer is not art by definition. ⁓ there's definitely different people. But I think it's those edge cases where it becomes really interesting. think when I'm talking about the Whiskers podcast, for example, yes, it was generated. They used AI to make something that otherwise would have taken months of CGI and would have cost like hundreds of thousands of dollars to produce a few clips. And now everyone who is creative enough to come up with something as silly as Kat's doing a podcast can turn that into a reality. And all the jokes that were in there and certain scenes, they were still created by human ideation. And I think that's a big difference when you compare it to, ⁓ for example, the explainer video, history videos or whatever, when it's just like AI generated images with an AI generated script, with an AI generated voiceover. That's not creative. That's soulless, so to say. So I think for me, at least when someone is using AI, but it's still human creativity that's involved. I still, I can still definitely appreciate it. speaker-1: I think that's a really great way to frame it. And perhaps I'm in that camp as well. mean, when I'm listening to music for running, I'm not necessarily absorbing it and appreciating it as this central piece of art that deserves some form of reverence. Now, if I'm listening to Tchaikovsky or as I said, Gustav Mahler or any of these classical composers, or if I want to listen to Bach in a sonata or want to hear a piano player, for example, you can actually tell if you are someone who really appreciates music when someone is a brilliant technical player and when they have something else. you know, and that's something else, like it moves you, right? So I think perhaps some of it is a ⁓ realization of what type of music and what type of art and what is it that the purpose is, the purpose it holds for you or the meaning that it holds for you. Because I sense that all of us are trying to say it's either this or that, and it isn't. mean, ⁓ AI for me, is creating all kinds of gray areas of almost every single assumption that we had as humans. I don't know how you feel. speaker-0: I feel very similar. And I think one of the interesting things on this topic is also when brands are trying to use it and especially when that completely backfires. mean, Coca-Cola ⁓ remade their classic commercial that we all know and love with trucks driving through the snowy area and it's snowing and people are cozy inside with the Christmas trees. They just remade that thing. With AI, there was nothing creative about it. They just told AI. I mean, it still took them a lot of work. I think they hired like four different studios to create that garbage commercial, but people hated it. People really didn't like it. And similarly, more recent example is the Olympics. mean, ⁓ a billion dollar organization, billion dollar organization using AI for their 30 second animation video. Like, come on, how difficult would it be to spend an extra 50k, an extra 100k to have someone talented, a studio make something real, you know, because the AI even messed up the rings on some of the logos. And it was just, it wasn't well received. And I completely understand that. But ⁓ then again, I, for example, if I'm going to watch a movie, I know it's similar, but AI is also very good at removing noise from audio, improving the audio quality, trying to add green screen effects, video effects. in the hands of certain creators, something beautiful can exist. But again, a tool rather than just copy this entire thing and make it some AI generated junk or something. what is ⁓ very interesting for brands to thread lightly with this, because if you mess this up, it can become very expensive. speaker-1: We are definitely in a in a time where there is some AI pushback from from the public in fact I I can even tell you even within my own circles people who appreciate AI tools and they use them professionally I I've gone back and I know many people have gone back to reading old-school books tactile books that you hold in your hands for example or to instead of have digital conversations or text conversations really actually starting to reach out to people and preferring and or attempting to have in-person interactions. There's almost, I won't say it's a ground swell, but I am sensing that people are really ⁓ starting to step up and say, where is the line between preserving ⁓ our very real, very valuable human capacities? And where is that line and how much we're going to acquiesce? When are we going to say enough is enough? And I think that we're dealing actually with narratives, which is something you're quite familiar with, right? Both of us have media backgrounds. We understand that the dominant narratives tend to either be very apocalyptic on the one end or techno-optimistic on the other. And we also have market drivers, right? Again, we talked about this in that, you we are being encouraged by and large to learn these tools because they are necessary for our futures in order for us to be productive, effective humans that can make a living and do things in the world. And while that is true, there's also this whole other aspect, which is It's equally important to remain human while we're doing it. And that is the challenge, to not give over who we are, what we feel, what we think, how we relate to a machine. And this is really the danger. So we have to, at a certain measure, be informed. There's no argument there. We need AI literacy, absolutely. Technical literacy, we need to learn about boundaries, we need to learn about privacy. All of these things are important. Equally, need to understand when we're looking at AI-generated content and be informed about that, which is going to require, as you have seen, there's been some efforts for evaluations and marking. And that way, ⁓ we have transparency in... what is AI generated and what is not. We're still in this transition period, but what I would say is that it's a very messy time where many people, as we talked about in conversations and in utilization of these tools, we're starting to almost lose ourselves. And I think that the people that utilize these tools frequently, such as you and I, right, we understand that something is at stake here. There's something that goes off and says, woof, wow, like I'm missing the feel of sunshine. I am missing like conversation. I need a hug. I mean, these are really sort of simple, basic human things, but they're things that are deteriorated, the connection to understanding the need for that, the more time that we invest engaging in these tools. Would you agree with that? speaker-0: ⁓ for sure, for sure. And I would also agree with the point you raised about the counter movement that is currently going on. spend a lot of time watching YouTube videos and ⁓ more often every week I see them come up more often in my feed. There's people that put in, whether it's the thumbnail of the video or the title of the video, they specifically market their video as being 100 % human made. No AI was used in the production of this video. I believe there's also authors that are currently put it in when you open their book. No AI was used writing this book. So I think it's really interesting to see that we are actually using 100 % human made as a unique selling point. a human made this, not this AI. garbage that you see everywhere now this was 100 % made by by humans. So I think that's that's really interesting. I'm yeah, finally, I was just curious to to see it. Do you think this counter movement is the beginning of something bigger to speaker-1: It could be, and I hope that it is actually, because ⁓ there is something about understanding when something is a tool and understanding when it's a relationship, and this is a relationship none of us ask for. speaker-0: No, no, was it was kind of no, none of us asked for, but we do seem to ⁓ gravitate towards it. speaker-1: We gravitate towards it and it's also because it's designed so well for us to engage with it in a seamless way with zero friction. so to your comment or your question about is this the sign of things to come, I know a lot of people, especially now that we can do a lot with vibe coding, for example, there are a lot of people, a lot of colleagues and friends of mine that have left big tech and are vibe coding their own more humane and more human solutions that actually help people retain their mental processes, their thinking, that add in friction points. So that way you don't spend three hours doing multi-turns in conversation with AI without getting a flag. Like, hey, you've asked about this same thing in five different ways. and we have been engaged now for about an hour, what do you say you go talk to your buddy? Like putting in actual friction points and stopping points or reflections, for example. So in this last, in your last comment, what did you mean when you said this? Or how did you arrive at the conclusion for this? or even prompting to have the user, the person, look at other perspectives instead of just saying, hey, your perspective's right, say, okay, so that is how you currently think about this. How about I give you five different other perspectives and why they might equally be correct, right? I mean, so I think that we have to really start looking at what's at stake here. what do we have the potential to lose? And in one way, that's an individual job, right? We can say it's our personal responsibility. But I can tell you as someone who's highly cognizant, who studies this stuff all the time, it is extremely easy to become dependent upon it because it's quick, you know, it's always available, it helps you be more proficient and efficient. And if that's us, Again, it's going back to the hackathon example. If we understand exactly what's happening in terms of the psychology, what about people who have no ability to grasp that and have no alternatives for that? So it's ⁓ not just an individual's responsibility. It is also a societal and an institutional responsibility. We have to come up with ways to help all of us develop healthier relationships. speaker-0: Mel, thank you so much. This was so great. You are so knowledgeable about all this stuff and it's always so good to speak to you. ⁓ Where can people go if they want to keep up with your work and see what you're working on? speaker-1: Great, yeah. So you can find me at futurehumanlab.com. So I post my work on there and I'll be having some conversations also in the future. And then I'm going to invite you to come and talk to me. speaker-0: I would love to, I would love to. Yeah, you know where to find me.