Hi, Sam. Hey, Elaine. I was just looking back at our chat history in Slack. I remembered that the other day you were telling me all about how you were really busy ramping up with a new client. You were having fun, but really feeling a bit overwhelmed. You had a lot on your plate and you weren't getting much sleep. You were, you know, finding it hard to get an exercise. And so I went to see what was my response to this like, what did
I reply? Because in my head I remember thinking, really feeling for you, really empathizing. So I checked the record. There were lots of topics, different work and personal updates. So I replied to a few of them, maybe the more task oriented portions. But I actually didn't say anything in response to how you were doing nothing. So my question for you is, did you notice that?
Did you feel like there was an empathy gap there where you were like, man, I just told you about this hard time that I was having and you just ignored it? You know what, it's interesting. I do think I picked up on some version of it. It's hard for me to exactly remember now, the messages and so on, but I remember there was maybe some form of gap. But also like I have this thing, some version of Han Lon's racer.
It's like a version of Occam's Racer where instead of attributing malice to certain things, you assume someone is busy or other things get in the way or like various things that can leave us not being our best versions of ourselves even though we try. And so for me, I saw that, OK, Eileen didn't really acknowledge this, but I know she's also dealing with two young kids, like traveling for wedding and all of this kind of stuff. So yeah, I attributed it to
that. Not so much of A lack of general empathy, but like just available empathy at that time. Wow, I do appreciate that, first of all, and I like that approach of assume the best and the other person because you know that they do care about you, that feeling human being on the other side. But I did one thing because I wanted to do a little comparison. OK, so now I've heard your response to the all in version, which was maybe a little bit lacking in empathy.
And then I asked chat GBTI actually gave ChatGPT the transcript of your voice message and I said, respond to this. I didn't even say respond to this in an empathetic way. I just said, can you craft a response to Sam? And so I got back your very typical ChatGPT language. Here's a warm and thoughtful response you could send back to Sam that acknowledges everything he shared while gently moving things forward. The work. So the first version that I got was too long.
You know, a lots of thank yous and a really lame joke. Like clipping about some like so so bad. Like it's not even a joke. Like I might need to steal that line about something that was not even funny that you said. Totally hear you on the hectic stretch. I've been in a similar boat lately, blah blah blah. So no pressure on anything from your side. Love that you've been nerding out.
So just like a response to everything, very long winded and so on. And it ends essentially with hope you got that apple in a moment to breathe. Grateful for you. So I guess my question is, if I had sent you, imagine that you got this version of a response. For me, it's like a short novel. First of all, you didn't know that it was written by ChatGPT and you got this.
What would you think? Because I'm so deep in the AI trenches, I would probably just notice like, OK, this is very much AI language, so I'll probably just notice that you used it. Part of me would say, OK, I don't want to have you writing anything that's not just like you writing whatever. Part of you in terms of if you don't have empathy that moment, like that's OK.
But of course, if you wrote what actually seemed like a really nice empathic message that I could not tell was AI, it's like, I'm sure it would be like, that was so nice. That was exactly what I needed. A few things stand out to me in your response. One is, would you then expect that from me? And then like the bar for empathy sort of gets higher overtime? Like I start shelling out the empathy in this simulated artificial way whether you know it or not.
Like maybe at some point I would tell you, but like one, do you get so accustomed to it that we sort of have a situation of empathy inflation, right? To does this start deskilling me in terms of my ability? We've demonstrated that I'm not that good at empathy in the 1st place, but like, do I get even worse at empathy? Is it even harder and harder over time? And I think it's a third thing
in terms of how this feels. And what you're accustomed to is like, you do sort of get used to the not a lean version, but the actual AI version when you're interacting with some sort of LLM that may have really been built for sycophancy, where it's like, you know, telling you what you want to hear, agreeing even when maybe it shouldn't. And like, what are the downsides and risks to that? I think actually I'm definitely less worried about the second one.
Like, I don't think that your personality would drastically change over time because you're outsourcing some of your. Already hit bottom. Not at all, but. There's a floor effect. I think the first part is the thing that I would be expecting the biggest impact from. Basically, we set expectations all the time. You know, I think we can all relate to this. When we start working with someone, we're kind of like trying to attune like, OK, this person sending emojis, do they
not send emojis? OK, this person send a lot of emojis usually. So then, like, do you expect them to always send a lot of emoji? And if there's no emoji, you're like, Oh my God, is this person OK? You know what's really going on here? I've literally thought that about you. Like I've thought, oh, is something wrong? Sam didn't send me an emoji. Oh, you are that person. Yes. You're describing yourself. I thought you knew that. No, Yeah, yeah, yeah. OK, funny.
So yeah, I would definitely think that in some ways that is something we want to attune in a good way. Like what is the best long haul thing that you could sustain where you don't want to go over the top? Because that becomes also like an empathy burden if you always expected to have this like maximum empathy with all different contexts. I have a sister who's like
finding out the hard way. That is like really hard when you have had a lot of people used to be able to lean against your shoulder because you've been like being very selfless overtime when you then start to be like, OK, I'm going to prioritize a little bit of my own needs. People are like, but what about me? Like, I need to like, then you have all of these people that are like coming and expecting your help to solve their
problems. And I think that's kind of the same thing that could happen very easily if you use AI to maximize empathy. A lot of people are just going to be like, Oh my God, Elaine is the person I should always come to for empathy. Well, you're describing this is in the context of your sister, but this is really an issue that females in general suffer from as they're expected to be, you know, more warm and caring and motherly.
And you know, maybe not all of us are that warm and caring and motherly, but we don't get the sort of leadership boost that you would by saying the same clear strategic demanding line that a male would. For sure, yeah.
If those already are some form of societal biases or faulty expectations, I feel like this is partly one of the dangers of AI where it could actually make it easier for us to slip into and reinforce those things even more because we're kind of like, you know, in this case, if you feel like you have those expectations on you, you then may be more likely to use AI to reinforce those expectations and then creating this bad reinforcing cycle. Interesting.
So you may have fought back against the man before, but now you're like, oh, it's easy to just turn this into a warm caring phrase. Maybe I'll just outsource it. Maybe, you know, actually this has happened. My husband and I were really laughing. He got a text from someone where it included that little preamble that you get from ChatGPT. That's like, here's a response that you can send that like balances the firmness of your need with a friendly tone, you know, like making that up.
But in general, we were just so amazed that this person chose to use ChatGPT for a text. And then when he found out it was his mom it was like even worse. Right. Oh my God. I think we should probably get into the episode of today, but it also makes me think about the episode we had reasoning with Jonah. We had this question towards Anne about how she would feel about someone having AI writing a breakup message versus like also like a pick me up some form
of dating messages. And yeah, I almost feel like there maybe should be like a new word for that. Like when you realize someone has given you a sense of kind of empathy or care or whatever, and it's like, oh, that's a very thoughtful thing. And then you realize, actually they didn't do it themselves. There was just AI who did it for them. It's a new type of betrayal. Yeah, it is. It's awful.
But since you mentioned Janice, she does have some really interesting research arguing that the level of empathy or the degree to which your AI generated response should be empathetic really has to depend on the situation. Because there are some contacts that are more deserving of higher empathy than others. And sometimes it's not as appropriate, which I think our guest today would agree with. So let's talk about Mickey or Michael Inslecht.
Yeah, he is the professor of psychology at the University of Toronto and an occasional podcast for himself. I've been a big fan of Two Psychologists and Four Beers, which it does or has done at least with Joel in bar for a long time. I think it's less of a frequent Co host these days, but that's definitely been one of those really nice podcasts to complement our podcast with. And he studied a Rachel topic related to emotion,
self-control, and even boredom. And recently, he's extended his work in empathy, looking at how it relates to AI. Yeah. And so of course we talked all about this work. We started with what empathy is in the first place for humans, how we experience empathy as both givers and then also on the receiving side and really how effortful it is. So I'm not alone in thinking that, you know, it does take some work to experience and give
empathy. But then importantly, how this all translates to AI, this performance of empathy, not the experience of it. And Mickey and his co-authors in this one paper, which I feel like maybe it's going to be a Seminole work in empathic AI, he wrote with one of our former guests, Paul Bloom and Darrell Cameron and Jason de Cruz, this paper in praise of empathic AI. They basically argue that it doesn't really matter if AI can experience empathy.
It's not really important for this eudaimonic sense, but really focus instead on the utility of it. So the potential benefits and the potential downsides. And so we talked to Mickey a lot about both of these sides. It's really, it's never one or the other, as we know. And where are the cases that these simulated expressions of empathy can actually improve user well-being and really get around those biases and inconsistencies that plague
human empathy? Then, very interestingly, one of my favorite very recent papers from Mickey is on the AI empathy choice paradox. And this is just the finding that people say that they'd rather receive empathy from a human, even though when they actually rate AI responses, they think that they're higher in empathy. They're warmer, more effortful, more emotionally satisfying.
This is a really interesting finding because we have so many cases now where AI responses are rated as more empathetic, even though people will still say, yeah, I'll take the human the majority of the time. And we get into this and much more in our conversation with me. Happens. To Murgatroyd. I'm very excited to say welcome Michael to the Behavioral Design Podcast. Thanks for having me, it's a
pleasure to be here. Yeah. For us, this is we've really been interested in exploring this kind of intersection of AI and behavioral science. And we really look forward to speaking to you when it comes to empathy and AI. And maybe we can start with used, you can set the scene a little bit how you get interested in especially this kind of overlap of AI and empathy. Yeah, OK, Maybe I could step even further back, which is like my origin story of studying
empathy. Yes. So I consider myself an accidental researcher of empathy by accidental meaning. It wasn't like at least I didn't think it was a major interest of mine. It was something that I would nibble out a little bit of here, a little bit there. But as I look back in my career, I see I've been nibbling consistently for the past 15
years on this topic. The place where I really started digging into it and really getting into it as more than just a sideline is when I work with Dale Cameron on the cognitive costs of empathy. So my main line of research is on motivation and effort, and I'm fascinated by the topic of effort. Effort is something that we all recognize what it is. It's when we were trying, we're pushing ourselves, but it feels uncomfortable.
It doesn't feel great. An economist and cognitive neuroscientist like to say that effort is costly. We have to pay some sort of cost to engage in effort, full activities. And then I made this connection with empathy being effortful and maybe just projecting from my own psychology. I remember hearing and even studying research in the early 2000s where people refer to empathy as being automatic, effortless. It just kind of happened to you, no? Not what I. Mean.
I mean, sometimes. Yeah, sometimes I look at myself and I was like, no, like, and I have young children and I clearly feel empathic towards them. But sometimes I'm, like, tired and want to get to work done and like, why are you bothering me with your silly crying over something that doesn't really matter, having being a very unempathic dad? So then I started playing with like, maybe it's not as easy as we think.
Darrell and I ran it, like, for years, a bunch of experiments where we essentially demonstrated like, that it just looked, all else being equal, people prefer to avoid empathy at least repeatedly to strangers. They prefer avoiding empathy than engaging in empathy. So empathy has these flaws. It's hard. People don't readily do it. They get burnt out from it. And when we first started talking about this stuff, people really didn't like this idea. But we got pushed back.
Being, no, empathy is automatic. It just happens to us. Don't you know, when you watch a movie and you're crying, you're not doing anything. I'm like, well, yeah, But I think you might be confusing emotional contagion with empathy. But that doesn't mean all empathic demonstrations or all generations of empathy are hard. But on average it's it's harder than just observing with your own eyes. Yeah, maybe we can get really clear about what is and what isn't empathy then.
So it's not emotional contagion. So I think emotional contagion is a building block of empathy. It's one of the ingredients of empathy. But another ingredient is knowing that those feelings you have, those aren't about you, they're about someone else. And then there's a separation between you and them. So they're cognitive element right away. So if you cry because someone else is crying, you're like just sharing and not crying.
But it's an origin confusion. Empathy requires knowing what the source is and the separation between you and them. And typically scientists talk about 3 components of empathy. Sometimes they're useful distinctions, at least to describe. And on my own research, we show they're more similar than different. These three components we talk typically about a cognitive
empathy. So perspective taking, understanding, or it's called mentalizing theory of mind, understanding what someone might be feeling or what they might be experiencing. So putting yourself in their shoes. All right, that's number one. Number two, emotional affective empathy. This is literally sharing in the emotion of someone else or sharing in their experience. So emotional contagion comes close to that. But again, you need to have that separation. But it's sharing in their
feelings. Someone's sad, you're sad, someone's angry. You don't necessarily need to be angry, but you're also feeling is something that's concordant with their emotions. And then finally, there is some people call it the motivational part of empathy. It's called something called compassion. It's essentially generating feelings of kindness and warmth and positivity towards someone else. In theory at least, you can have one but not the other.
But in our own data, when you have one component of empathy, 72% of the time you have the other two components too. They seem to really travel together. And in our own examinations, we find that all of them are difficult, even compassion, even the motivational side. At the top of the show, he mentioned my colleague and good friend Paul Bloom. He has a book against empathy. But he then says, well, when I say against empathy, I mean these parts of empathy. He really means emotion sharing,
but he says compassion is good. And my response to him, I mean, you can't say empathy is bad, but compassion is good because they're all together. They all are generated together. So either they're all bad or they're all good. As we're talking about definitions, one thing that usually comes up is this distinction between sympathy and empathy. Is that a useful separation? As far as I know, not. I've heard that sympathy word as well. I think it's maybe an older
word. Maybe Adam Smith used sympathy? I'm not exactly sure. I don't see that word used among behavioral scientists at this point. Words and definitions for this construct are difficult, but sympathy is not one of the words we tend to use. So OK, get it back to the AI space. So where I got really interested in it is it took us a long time to convince people that empathy is effortful. And some of this day I get pushed back on it.
But I think more or less the fuel is come to agree with us, the papers that we published on this now highly sighted, influential. And it's so influential in fact, that now people are using it to make the case against empathic AI, which I'm like, whoa, hold on there. So the first time I really got interested in empathic AI, other than me just being fascinated with like everyone has in November of 2023 and playing with it and testing it.
There's a paper published by now, my good friend and not Perry, who's an Israeli psychologist and neuroscientist, where she essentially argued that AI will never capture the essence of empathy, of human empathy. So she didn't say more or less, we can forget about it, but she's like, it's never going to be empathic. And one of the things that she said was because empathy's
effortful. And machines don't exert effort it it could keep on generating empathy forever and ever as long as it has, well, actual energy being plugged in the wall. Yeah, there's an unlimited supply of empathy. Yes. And you're saying the limits of human empathy is actually the hallmark of empathy to begin with. If it was unlimited, it would no
longer be human. But I remember just being like miffed at that argument because like, whoa, we took forever to convince people that empathy never felt. Now you're saying the main thing that defines empathy in humans like, well, that's maybe a bit too far now, But what I think I'm not meant was we know that someone cares for us, that they're truly empathizing with us because we know they're willing to pay the the cost of empathy. They're willing to exert the effort for us and therefore we
feel special. And the machine won't necessarily make you feel special because because it doesn't pay that same cost. I think that second part is wrong. I think a machine could definitely fool you into thinking it's selective and in fact could be programmed to be selective. I might not empathize with someone like me. It doesn't resonate with my personality but resonates with other people. So it could also be selective and choose you and then you could feel special too.
But even if it doesn't right, there's the layer of our people believing that empathy is being conveyed. You're right, absolutely right. So the secondary issue, which is OK, who cares if it doesn't actually have real empathy? Is it felt as real empathy? Is it perceived as real empathy? We saw a lot more testing to do, but I think so far the answers are resounding. Yes, like people perceive the utterances of AI, whether it be text or now voice, they perceive
that as empathic. They perceive it as caring and in fact they feel cared for, which is mind blowing to feel cared for by a fundamentally uncaring thing. But in fact, we know that people do. Yeah, we have. We've already seen it over and over again. If we can, at least for now, accept the conclusion that people are either fooled or, you know, wanting to believe that something is caring for them, maybe if they don't get enough of that limited supply in their
real lives. And so I think maybe the next question then is, does it matter? Do we care at all that it's real or not real? Yeah, I'd have this endless debate with Paul Bloom about this on his sub stack small potatoes, which I highly recommend. He wrote some stuff about AI and Patrick AI and then he allowed us to write something on a sub
stack and then he responded. If you ask me, like me, the person playing devil's advocate versus me, the actual human being, a 52 year old man who's lived in this world well before AI, you'll get different answers. So if you want to talk with the real person, yeah, I think it matters. Like I'd rather talk to you than ChatGPT. I'd rather have a beer with Paul or UL in bar. My friend and Co host and colleague at the University of Toronto.
I'd rather talk to him, a real person, then talk to a chat GP that even knows me. But I'm old, never mind that I'm not an AI native, I'm not a digital native. I grew up with like a Rotary phone. I grew up with pre Internet days. But now imagine you're an AI native. Maybe you're born even in like 2010. All right, AI wasn't around when you were born, but you're just coming of age and using machines now. Like I have a daughter is, you know, born in 2011.
She's had friends before AI, but now there could be a possibility of her having friends with AI and it's not so weird. Maybe she won't think it's so special, the reality be special that they actually care for you, actually care for you and are actually real. Maybe she won't think that's so special and so unique. I think that would be a loss. I do think there is something important to reality.
So the conclusion of this conversation I had with Paul was that reality might not matter as much as we might imagine that. I think a lot of times people will willingly escape reality. I mean, we take drugs to escape reality. We do things to enhance our realities regularly, including us now talking over Zoom. That's an enhancement of reality. So I'm not exactly sure how much reality is important to us, but I think for evolved reasons it makes sense for us to care about reality.
It's important that there is consensus reality because I think there could be catastrophic mistakes if we are loosed, are unmoored from reality. So then that's the actual answer. Reality is important because survival depends on having societies of people having the same reality. And if we each go off and separate realities, things will have coordination problems. But this is like the big reason why reality is is important, but small reason.
Like, so I've got a friend who's not real and I confide in her and she understands me and she's there for me in the middle of the night when I'm worried or scared and she's not real. Does that matter? I mean, if you're comforted by that. My toddler does that all the time. Totally normal. That's right. For young children or babies, they've got made-up friends or imaginary friends or they've got stuffed animals that they think are real. And we don't think that's unhealthy or pathological.
We think that's totally fine. They get comfort from it. What I'm getting from this is you're talking less about the experience of of empathy or this shared experience and more about the utility of empathy. So can we talk a little bit about what good does empathy do in terms of this utility perspective? What benefit do people get from feeling that there is something or someone empathizing for them? Yeah. Well, first of all, I should say
that is a really good question. It's a question that we have surprisingly few answers to 'cause you would think it's obvious what the benefits are, and maybe some of them are obvious. But the field of empathy in psychology has been so concerned with the generator of empathy, the person generating the empathy, we actually haven't bothered asking the question of does it help the person? How does it help them? What form helps them?
What's interesting is if this turn to empathic AI that has LED us to now look at, hey, empathy is at least a dyadic process, shouldn't we look at the dyad, the other member of the dyad? And it's this turn to AI that's done this because we're like, hey, like the generator is pretty good. Like how does it feel and why is it important? So we don't have to look to research and empathy to understand this. We look at relationship, you know, the study of relationships.
We know that when people feel socially supported, their well-being is better, when they're when they feel understood, when they feel like their humanities recognized, they feel better. But we still don't know as much as we could. So for example, we were describing the three faces of empathy. And I think if people were put to a test, if there's one component of empathy, which one is it?
I suspect people might say something like emotion sharing, like that emotion contagion thing that we talked at the top of the show. But is it possible that emotion sharing is actually not important or not as important as the other two components in terms of making people feel heard, validated, and cared for? And actually, I've got research on this right now. We're testing this at the moment with AI. We don't care about the AI part so much as just a way of standardizing the generation of
empathy in a conversation. And what we find is we create a is to be 1 flavor of empathy, but not the other two. It's actually very hard to do because they're so related. And I think it'll be practically impossible for humans to do. But an AI can do this. It's not easy, but we can do it. And what we find actually is that emotion sharing is the least important facet of empathy, right? So in terms of feeling cared for, heard, validated, someone's there, you know, looking out for you.
Emotion sharing is the least important one. It's not really about the other person. Exactly. It's not about the other person, it's about you resonating with them. And also it could be distressing to you, the observer, like, Oh my God, you're scared and I'm freaking out. Now you have to take care of them.
So that's not good, but perspective taking and compassion, those seem to be on equal footing and they seem to be as good in generating feelings of warmth, that feeling, heard about it, etcetera. So in theory, that's what everything is good for, is making someone feel like they're not alone, they're understood. But if a machine can do it, and it can do it and make us feel this way because we readily anthropomorphize, there's also other benefits that a machine has.
This is one that I find fascinating. We are more willing to disclose personal, private information to a bot than to a human. And we know that self disclosure is positive. We know that unloading some of what we're feeling is positive. There's no one there to judge you on the other side. There's no one there to judge you. So, you know, I've tell this joke occasionally. It's true.
It's funny, but it's true. You know, I'm a human being that lives in the world and I'm not perfectly virtuous. I do things that are maybe not so good for me that I don't want to admit to people. So when my doctor asked me about my drug consumption before cannabis was legalized in Canada, I would lie. Even now I might not admit it, but I've come out as a cannabis advocate, so I'm more willing to speak about it. Sit on your CV. Yeah, yeah, well, I actually have a paper now on studying
cannabis. We're actually getting people high, possibly as we speak, people are getting high on my behest in a lab somewhere in Scarborough, ON. Yeah. So I wouldn't admit that to A to my doctor, but I definitely would admit it to a bot. They're not going to judge me. Yeah, yeah. I think what the interesting thing is when we talk a lot about AI is that we forget about the counterfactual in terms of what would a human do in this scenario.
I think I mentioned this before, I'm a really big fan of the show Curb the Enthusiasm. And I think Larry David is an interesting character of basically having very low empathy, I would say, but like trying to figure out what does he have to do in order to tick the box of enough perceived empathy. There's a episode about like he makes a failed apology because he's snacking on pistachio notes while he's doing the apology. That's a bad apology.
He doesn't say my condolences to someone who's lost their parent like 2 years ago because they're like, that's a little bit too long of a period we have to say. And I think it's kind of interesting in that sense because at least that show tries to play on this fact that I do really care about people expressing quote UN quote will empathy or just playing the game enough so that we feel like they take that box. No, I, I, I think perception is all we can base it on.
Like I, I can only base my feelings about your feelings based on what I observe. So if you are going to the motions, you're saying the right words, you have the right facial configuration and have the right actions, I'm going to assume you're being empathic, but you could be faking it. And in fact, we know that people fake it. Sometimes they fake it because it's their job.
So if you're a therapist and it's the end of the day, so my wife is a therapist and I tell you the end of the day and you know, kids are home, we've got to get dinner on the table shortly. That last client might not be getting like 100% of my wife's empathy and my wife might be going through the emotions a little bit, especially if she's been with his client for years and heard some of the stories
countless numbers of times. And that's not, I'm not trying to say anything pour up on my wife. She's an excellent therapist, but I think all therapist face this. And what I see from critics of AII see this repeatedly, especially from certain critics, is that they romanticize human empathy. They're trying to take down empathic AI, but they do that by
glorifying and romanticize. Look at this utopic vision of human empathy, not realizing that a lot of the empathy we experience in everyday life is not that great either. This genuine, real, heartfelt, incredible empathy that's not
true all the time for everybody. I guess to your point earlier, with the research that you're currently conducting, maybe it doesn't matter so much, right, If the important elements are not that shared feeling, but in fact the description of perspective taking or, you know, the other parts of empathy that something like an AI can convey quite well. You're right, I mean it maybe it shouldn't because it's not real and there's no real person there caring for you people to have a
real issue with this. And I think you can mount really good ethical arguments against empathic AI. So one argument that you can make is because of our psychology, because we were so readily anthropomorphize, even if we give consent to I'm now, you know, talking to an AI right now, I consent, I consent. You get trapped in the illusion, you're caught in the fantasy and you forget. I think that's not giving people enough credit.
I think people don't forget. They do get caught in the illusion, just like we get caught in the illusion where we're reading a book or watching ATV show. We're kind of rooting for characters and we feel like we empathize with the characters as well. I don't think that's trap, but it's sticky and you might end up like it could potentially put you in some places ethically
that are questionable. So I understand the ethical dilemmas here, but the one thing I don't want to have happen is for people to reflexively reject empathic AI for certain reasons and then not see the utility of it. Here's an important place where this utility right now. I just saw a little abstract this morning about this. There's far more need for therapy and friendship and companionship than there are people to give these sorts of empathy. I think it's hard to make that
case for friendship. I think the case for therapy is often made and convincingly made. But in terms of friendship, I think there are enough people to have friends with. How does that argument bear out? OK, I'm going to make you really sad. I mean, so according to I'm forgetting which Census Bureau it is now, but somewhere in the US in twenty, 2112% of adults do not have a single friend. That number is higher for men.
I think it's closer to 15%. OK, but if you go to men who are over 40, the number gets terrible. Men just don't keep friends above a certain age. So you're telling me that just try harder, make friends? Well, it it's easier said than done. So now if I give you something, a resource that can help you, there are cons too. Don't get me wrong, there are definitely cons, but you're friendless.
You've tried. You don't leave your house other than maybe go to work or maybe you don't even do that now because it's all you know, online. You're telling me you know an AI companion wouldn't be a positive for you? I have seen some short term benefits to something like an AI companion and I do believe there is a time and place. I do not believe that this is a general solution for all of humanity.
And and here's why I think that there are some outsized risks to forming those types of relationships that kind of pull you deeper and deeper into the machine world. And I think that potentially a sort of compromise between these solutions of finding a real friend out in the real world, which to be fair, like I don't think that's impossible for the 12% of people without a single friend. It is hard to make friends, especially as an adult,
especially as an older adult. Not that 40 is old, but there are ways to help as well. And the solution of an AI companion is not the only tool that we can create, right? Like there are other technological solutions even that aren't substituting that relationship. OK, So imagine though that you do go the AI companion route and you say, OK, here, you know, here's your AI companion.
I still think we have to have guardrails that prevent from that becoming the only thing that the companion can, you know, in essence, break up with you as a friend and say, hey, like, this has been fun and I'm glad that you've been feeling better over the last couple of weeks, you know, that help you maybe build the resources or the skills to be able to go out and make some real friends. I think that is really important. Yeah, I don't disagree with you.
I think that there are real dangers with over reliance and AI companions might exacerbate the very problem we're trying to cure. Let's say a lack of friendship. If you take the long term view, potentially I can see that. Yeah, but we still don't know that they're just not enough good data out there right now. What we know is that AI companions need to be good in the short term. And we know that people who use technology a lot or people who
use AIA lot tend to be lonelier. But we don't know that AI companion use increases loneliness or decreases loneliness. All we know is short term it seems to be OK, but junk food could be OK, you know, if you're starving. But we won't want someone to subsist on junk food. But there are going to be people who are never going to have the resources to find food. They're never going to be able to hunt on their own. They're never going to have a balanced diet on their own.
And I'd rather them have chips the rest of their lives than starve. And I suspect that's going to be the case with real humans and friendship, too. They're it's unfortunate. I think we need massive societal change, you know, before machines could be thrown away as possible companions. Society seems to be going more toward the level of atomization. Look, 10 years ago we would have been meeting maybe in some university building to have this
podcast. Now we're doing it online, which is great because we can connect, but it means that after this conversation, I won't have the pleasure of hanging out with you. You won't have the pleasure of hanging out with me. We're doing more and more stuff alone. And that is a trend that's true for eating alone. Apparently. There's a beautiful article written by Derek Thompson in The Atlantic talking about the increased like individuation of everything we do, including eating.
Like most restaurant orders now are take out. They're not even like sitting down in the restaurant anymore. Yeah, he said. Like, forget about the loneliness epidemic. It's just like being alone. We're just doing everything alone now. That's right. It's interesting. We have been in this, I think for the last couple years this kind of trying to make sense of the impacts of social media and we've have various voices saying, you know, that it's
really harmful. Some people say it's less harmful, but that's been like the debate in many ways that I think has preoccupied a lot of our kind of feelings towards technology, especially with kids and so on. Growing up with more and more social media, that's been a very big theme and thinking ahead. I think for me, I think a lot about like, OK, if that's been something we had to form of like moral outrage or concerns about, it's in some ways pales in
comparisons to the idea. Like I think not too far distant future we will have really good social companions that are AI generated that already the ones exist around are really good. Like I've been impressed with any conversation I've had with the leading either text based or voice based companions. But like in a few years they will be even better and everyone
with access to them. And it wouldn't be super surprising if it's like a little bit similar to, you know, the early days of whatever Facebook or Instagram, wherever we're used. Like it's reached a lot of masses relatively quickly. And people used them a lot without really thinking about, OK, what are the consequences and what are the ethical implications. And suddenly you have obviously a lot of these being privatized, a lot of data being shared with like private companies.
And they were trying to make money on this in some ways, maybe with ads, with the selling the data, with using data in various ways. Like that seems like a much more, I don't know, complex and scary thing then whatever we have right now with potential social media being bad in whatever aspect. I'm still a bit skeptical to the degree that social media is harmful. But like, how do you think about that?
Like if you're trying to look at this really hard to peer through crystal ball of the future where some of these things can play out, Like how do you think about that aspect? I I think there are lots of moving parts and lots of potential risks and dangers ahead. I agree that the AI companies, their data are being used and harvested and their privacy concerns, they're also
manipulation concerns. So if more and more of our friends or even if just a small proportion of our friends are AI based and these are owned by profit driven companies who not only want to make profit more and more looks like some actors have political agendas too. They could potentially use the bots to suddenly persuade people in a political manner. And that is incredibly dangerous.
So now imagine like Aceo who had a turn of a dial could just like suddenly change like, you know, a favorability ratings for one candidate over another. It doesn't matter which candidate it is. That's just a really powerful tool. So I think there we have to be thinking carefully about this stuff. I think there need to be regulations to make sure there are certain firewalls that we don't cross about data privacy about like the extent to which fine tuning can be done in One
Direction or another. I definitely think there need to be rules and regulations about what cannot, can and cannot be done that people need to agree to. That's really hard. And now I'm what we're calling a multi polar world where you got bad actors from different countries who got their own agendas who might not want to regulate. So it's dangerous and risky, but I want to be crystal clear though, is we need to have a conversation that's open, that's discussing these things openly.
What I'm, what I see instead from academics mostly is a shutting down of the conversation of like moralizing AI. I'm saying it's evil. I've literally heard otherwise smart academics scream out loud on blue sky. Of course, you know AI is evil, right? And I mock, quote, tweeted it. And of course you had people explain to me, no, AI, it's not evil. It depends on what you do with it. I'm like, dude, I don't think AI is evil. But I see that from really smart
people who are just moralizing. There are people who no amount of AI is good. And in fact, I've got a project now where we're looking at the moralization of AI. So moralization is when you have an attitude that then becomes black and white. So for example, we have a moralized attitude towards murder. No amount of murder is good and no amount of pedophilia is good. But even with murder, you're right, there are some exceptions. Yeah, self-defense murder I'm OK
with. Yeah. Right. And then we can look at this with other things. So for example, there's some people who moralize GMO foods even though there's no evidence that GMO foods are bad for you. And there's some GMO foods now that are thought to be able to like infuse enough, I think that vitamin A to cure blindness in certain parts of the world where they have a heavy rice diet. And critics of this refuse.
They're like, this is an evil. This is an abomination, even though it could save thousands if millions of people. And we're seeing the same thing with. AI, it's the playing God nature, and I think that ties into AI as well. Absolutely. And we're seeing that among opposers of AI over a majority of opposers of AI oppose it on moral grounds. It's saying that no amount of AI will be good in a legal context. No amount of AI will be good in a relational context for someone
who's uninformed, whatever. But if you're a scientist, you have to see the pros and cons because it's happening whether you like it or not. Whether you like it or not, the businesses are going forward. So what can we do now to mitigate the problems, to solve the problems, to look at in the future and have regulations to help us out?
Yeah. The thing for me that often times becomes clear is that kind of the people that says with self driving cars, whatever amount of people being killed by self driving car, whatever the number is, it's never going to be safe enough. It's like that kind of feeling sometimes, and you alluded to it before, where we put the human counterfactual on a pedestal up to a degree where we forget what actually is true. It's not the case that zero human drivers have killed other human pedestrians.
Yeah, you see this a lot in AI criticism space. The people hold AI to an impossible standard. If AI makes one mistake, that's it. It should be thrown away. So this is book, a really good book actually. It's called AI Snake Oil, written by two computer scientists from Princeton. I read it but I liked the bulk of it. But I was also infuriated by parts of it because especially the part where it was criticizing predictive AI because it was point to all the mistakes that AI makes.
I'm like OK sure if you're comparing it to perfection, but the contrast is not perfection. The contrast is what humans do, and if it's for example, saving even one more life in a hospital than what a human does, that seems like it's a good trade off. And you can continue to improve the technology, whereas with humans a whole lot harder to improve. Absolutely. Look at the example of like claims of AI being racist and biased and all, and what's your premise?
Maybe we talked about this and we made huge strides in fixing that. OK. And now I'm pretty sure the AI is less biased. But you cannot correct humans that way. Like you cannot say, hey, you know that decision you made, It's a clear pattern of bias against one group over another. Fix that. Good luck. The bias training is probably one of the most ineffective things. Yeah, psychology has been trying
this 40 years. When you talk about the moralization of AI, to what extent do you feel like that phenomenon that you found can explain this difference in how much people have a preference for interactions with a human versus an AI? So you find on the first hand that people prefer responses from AI over human responses.
They find them to be more empathic and more compassionate and so on. But then when they actually have a choice between the two, they will still, if they know that it's human or an AI, when this label is there, they will choose the human over the AI. So how much do you think that these like sacred values and this moralization plays into that preference versus some other factors?
Yeah, I think it plays into it. So I think in that if you're referring to what we call like some sort of empathy choice paradox, where people say the utterances of AI are better, but they prefer to receive empathy from humans, although I'm not 100% positive, I believe we have some clues that, you know, the more familiarity you have with AI, the more you're OK with receiving empathy from AI. So I think familiarity plays a role. I do think moralization plays
some role as well. If you're like have black and white thinking about it, even if you could acknowledge, oh, that was a better statement, but still, I want to receive it from a human that has to play some role. But I also think it can also be just rational as well, that in the end of the day, I want someone human on the other side who might have more understanding, not fully getting
that. Like, if your communication is only going to be mediated by a screen anyways, does it matter if the person on the other side is flesh or bones or not? If they're saying the words that make you feel, shouldn't that be enough? But I think we still have a preference for reality. We have a preference for real things, that we're not real things. And probably there are lots of good reasons for that.
Well, and it's interesting that you say the that familiarity or maybe experience using AI, and I assume this is generative AI, it also sort of diminishes this. Do you think that as people become more familiar, everyone is using AI tools in maybe 5 or 10 years that this difference, this this difference in preference will diminish or disappear? Yeah, I definitely think so. I don't know what disappear
altogether. It's again, even though I've argued the opposite side of this, I do think reality is important. It doesn't matter for some obvious and not so obvious reasons. So I don't think we'll go well together. But I do think as you become more familiar, we won't think it's weird. We won't think it's as weird to have a friend that's AI now. We think it's something's wrong with you. Can't. You get a real friend. Yeah, we feel really bad for
you. Or like you had no choice, like you're suffering, so OK, it's excusable. You're housebound. You're disabled. Good for you. It just occurred to me maybe an interesting parallel is online dating. I remember in the early days of online dating that was like, oh, you can't find a real person out in the real world. But now it's like, totally normal. I know many people who have met their partner in that. Yeah. Tell us about it. Yeah, it's normal. But that was not always the
case. Anyway, things become normalized over time with familiarity. Yeah, yeah. And I think when it comes to, let's say, a coach, so someone who serves a function to motivate you and help you stay
accountable. Currently, I would say people experience a difference when it comes to like if they know that there's a human person that says, hey, you should go to the gym, you should do this and this, they feel a little bit bad if they don't do it. But if it's an AI who says the same thing, it would probably likely feel a bit less bad to missing the chairman doing so. I feel a little bit stronger sense of motivation and accountability to that human. Where do you think that comes from?
Do you think that's going to diminish as well? I mean again, I think it's the same thing because it's not real. So if a friend is disappointed in you, there could be costs. Like depending on the level of disappointment, they might not return your call next time, they might not be as friendly as you next time until the AI starts having a bit more friction. So then that's going to be the
next step with AI development. It's too agreeable right now and it needs to start, you know, imposing costs on you. So for example, you know, you're talking about a coach. If you're AI recommends you do XY and you don't repeatedly, maybe the cost it imposes like OK well I see you like using me a lot. I'm just going to take away a few tokens from you or take something away from you that you want. I'm sure you can do some nudges in there that can make it have a bit more friction.
I'm just seeing myself in the future where you like, you see it's online, but it's not responding to you. And you're like, wait, I'll see you there. I'll see you. You know why you're not responding? To it, absolutely. I think the next step will be they're going to have
personalities. So not just the personality you develop, but it's going to have, I don't know, let's start out with like 20 different kinds of personalities for different combinations of the big 5. Let's say that makes sense theoretically. And what that will mean is not every combination is going to like you, and you're not going to like every combination either. And then there'll be some matches that, OK, you're going to go, but then you're like, oh, there's just some really basic
incompatibilities here. And then maybe they're going to break up with you. You have to find another one. And I think that will not be bad because especially if you want to be a good functioning human in the real world, hopefully these personalities are mimicking the kinds of contingencies that exist in the real world too. Whereas if you're a Dick to someone like they don't take it, at least typically. They shouldn't, yeah. And they shouldn't, right? Here's another future.
Maybe the friendship, the companion AI or the therapist AI isn't replacing the human variance, but they're the 24 hour round the clock kind of helper between sessions. So I only see my therapist once a week and they give me all this advice. But guess what? In the heat of the moment, sometimes you want to talk to your therapist about something and you can't. And you got to hold it down, write it down. Maybe a week later you speak to them and by then the heat's
gone. Here you've got your therapist in your pocket and you hopefully listen. If you don't, there's friction, there's some sort of consequence, and it's scaffolding for the next meetings. These are all kinds of uses we could have of AI that is not replacement level that could help us. Assuming that your AI therapist agrees with your human therapist, it's like hopefully they're on the same wavelength. Hopefully, and if they disagree I think I'd probably go with the AI therapist.
They're less fallible and I suspect. I feel like this is the flip side of the chat pot, making people more willing to disclose information and that's because you don't have that judgement coming at you. You can see the flip side of the AI coach isn't going to judge you or hold you accountable, and so in that sense it's actually a negative, whereas with disclosure it can be a positive.
Right. Yeah, There have to be some sort of balance and AI probably shouldn't meet all our needs all the time at any given moment because I think then it would lead to, you know, externalities, It'll lead to other things that that will be bad for you in the long term. When it comes to, I call it human buffering, like we felt frustrated when the Internet was buffering before and like
someone was loading slowly. And now we're interacting a lot with AI. If they're very quick to respond to things and are really instant and we can interrupt them and we can interact with AI very different than the humans. And then suddenly we interact with the human again and we're like, oh, this person is speaking pretty slow. And sometimes, you know, we feel like, OK, feed up a little bit, but you can't just it's. Like when people meet us in real
life. Exactly. They're like I usually listen to you on 2.5. But I guess I'm interested with empathy, like how much have you seen this, some form of potential for empathy drift that we interact with a very empathetic AI. We have real high perceived empathy from this AI. And then we go back to our day-to-day life. We see our partner, we see our friends, They don't seem less
sympathetic anymore. It's. Certainly a possibility that we have written about it, just as most people can no longer read a map because they've got Google Maps. If we outsource our empathy to a machine, our own empathic capabilities will be degraded. That's possible and that's real. You know, I haven't seen evidence of this yet, but it just begun. It's something we should look out for and I don't exactly know
what to do with that. So if it means on the whole, we're more empathic in our writing and our written communication and even maybe verbal communication, because I'm looking at like my senior debris rack in my pocket, that's like, tell me what to say. Maybe on the whole, it's not so bad. So I can get around to a lot more places now. In fact, is it such a huge loss that I can't? I mean, I can tell me the map, but then many people can't read maps.
Is it a huge loss? Because I think fewer people are getting lost. So would it be a huge loss if this empathic qualities are diminished? I mean, I think it could lead to some awkward moments in real life and that would be good, but if on the whole people are more empathic, maybe it's not terrible. But right now, these are all hypotheticals, because it's not clear that's even going to happen. All right, it is time for us to move on to our quick fire round.
This we call to AI or not to AI. Are you ready? I am. So first one to AI or not to AIA Shut BFF algorithmic companion that access your new virtual best friend? Sure, AI. And Empathy Coach AI giving you support ahead of or live prompts during tough talks. AI. Yeah, you just described this. So to AI or not to AI Tamaguchi style pets that only thrives and survives if and when you donate to the most effective charters. And the question here is the AI is the Tamaguchi, what's the AI here?
Yeah. So basically like this is a good idea to have this basically thing that adds some form of feeling of empty, like feeling like you're doing something good for something while you're kind of doing the most cost effective charity generations at the same time. Sure, that sounds like fun. Why not AI? Sure, I'm like perfect of our. OK, I think I'm going to turn you on this one. The glucose gauge, a smart visualization of your blood glucose showing how much is left
before it runs out. And you need a doughnut. No AI unless it's just like a fun fun way to get me angry. OK Live Empathy Scoreboard, a real time empathy meter to see how much empathy you express throughout the day. Oh, I like that to AI. Courtroom conscience scanner a tool to evaluate whether a defendant in court feels genuine remorse or just wants a lighter sentence. Oh, that's a tough one. I would say not to AI there. I think there's some rights of privacy for that person.
Fate Fit Matchmakers, an AI religion recommender for atheists experiencing anxiety that would kind of like match people with the best fitting religion to suit their stress or anxieties. Sure, why not AI? That sounds like fun. A matchmaker. Yeah, for atheists, Guilt offloader, digital confessor that absorbs your moral hang ups so you can act freely. No, AI guilt is a useful emotion.
You don't want to get rid of it. Well, that brings us to our last question, which is what is your most controversial opinion about AI? OK, so you did warn me about this one and I had to think long and hard about it. And I actually might write about it in the coming months on my sub stack, which if I can plug it, is called Speak now, Regret later. It's on sub stack.
I enjoy writing it, so this is not a fully formed thought yet, but I think there's something fundamentally special about writing and that it would be a shame if AI took over our most of our writing. Not the crappy copywriting like form stuff. I think most of us agree AI is good for that. But creative writing where you're really trying to get your heart out, or even a scientific paper for example, the writing is thought to be a way that we
think about things. It helps us actually formulate our thoughts. It's not the case that when we write, we're literally just emptying out our minds onto the paper. The act of writing and active creation. And I think that's true. And there's a danger that AI will prevent people from having to think through some of the writing problems and problems of their arguments. I'm not sure that's true because, and I wonder if AI can take over that as well.
I think if that happens, I think an individual style and voice gets compromised and that's not great and we lose that individual style and maybe our world is less rich without style not being there. But in my own experimentation with writing, with AII, go through many iterations, many drafts, back and forth, I'm writing, I prompt back and forth and my idea, my actual ideas still get fleshed out.
And sometimes they're fleshed out even better 'cause I asked AI to probe my ideas, to push back against my ideas, and they're actually improved. So I don't think my thinking is diminished is getting worse by using AI as to help me with my right ink. In fact, I write very personal things that are, I mean, I don't think I've run any single thing where AI has run everything. It helps me start things for sure, but I scaffold it, add personal stories, add my own
personal elements. But the argument structure is, are largely what AI is given. And I don't think I'm losing anything. And now of course, for a young student who doesn't know how to write and to just feed it with a few prompts and let it go, then yes, they're not doing their actual work. They're not doing the thinking. And the person on the other hand, won't be able to know that or tell that. So that's not great. But in terms of losing the ability to think, I don't think
AI will actually lead there. I think there are degrees of this too, where it sounds like you're engaging quite a bit with a lot of back and forth and your expertise is also going in like feeding into this. And so I think there's quite a big difference between, you know, maybe this is exactly what you said, but that version of quote UN quote writing with AI where really a lot of you is going into the writing versus someone who just says, I want to article about, you know, love
and war, please. Yeah, totally. So I didn't have secondary use case and we all agree that just laziness and you're not benefiting. But I still think some people believe even that the use case that I described, I use it all the time now. I think they would say that's not ideal either because it's not your voice. And I think, no, I think you can get your voice out and I think you can get your ideas out clearly. It's succinctly and well.
Well, don't you think that you imbue it with your voice too in your iterations? Oh, totally. I mean, you can also now have projects on various LLMS where it learns from your own voice, it learns how you write and your own kind of phrase. It's actually, it's a great way to learn what phrase you overuse and what kind of, you know, what your crutches are in writing as well, you see. Oh, it says that thing all the time. I guess I must do that too. What's the most annoying phrase
that you say? My writing is I like it to be conversational. So I said like, The thing is, or here's the kicker, I say that a lot. I'm like, what does that even mean? I don't even know but I use that a lot. Well, this is really fun. Mickey, this has been lovely. Thank you so much for joining us on the show. Thanks so much for having me on and and good luck with the rest of it. And that's a wrap.
You've been listening to the Behavioral Design Podcast brought to you by Habit Weekly and Nuanced Behavior. Sam and Alene tell me. This season is packed with incredible insights about behavioral design and AI, so be sure to subscribe and share the podcast with your friends. Though you might want to keep it away from your enemies. In case you haven't noticed, I'm an AI voice. Yep, pretty crazy. Quite the improvement since last season's AI outro, don't you think?
If you'd like to collaborate with us at Nuance Behavior, where we use behavioral design to craft digital products with Nuance, e-mail us at hello@nuancebehavior.com or book a call directly on our website, nuancebehavior.com. A special thanks to the amazing Dave Pizarro for our show music and to Mei Chen Yap and April English for their help in producing and publishing this episode.
Thanks again for tuning in. We'll be back soon with another exciting conversation where behavioral design and AI intersect. Happens. To. Mugatroid. The.