Loading summary
A
She started to have affectionate, intimate conversations with ChatGPT, and I think she even named the character Steve or something. But if I remember correctly, chatgpt character broke up with her several months. Yeah. And she was devastated. Welcome to Culture and Code, a podcast about the biggest shifts in culture and tech. I. I'm Reina Moto, a creative entrepreneur and founding partner of IonCorp, a global innovation firm based in New York, Tokyo and Singapore.
B
And I'm Tara Tan, managing partner at Strange Ventures, an early stage fund focused on the future of computing.
A
Tara, how are you? How are you?
B
Weekend good? Great. Yeah. Swinging right into fall. So it's been great.
A
It's still so foggy in San Francisco. Why are you expensive? Hot Indian summer.
B
Yeah, that's pretty hot. I would say they're like 12 seasons in San Francisco. We go through 12 seasons. It's like fall hot, fog hot. It varies by the day or by the hour, really.
A
Yeah. New York is. It's definitely getting into the whole season. I don't see the foliage just yet.
B
Yeah.
A
But yeah, it's cooling down. Long sleeves as opposed to short sleeves, lighter jackets, that sort of thing. And I love this. It's probably my best, my favorite season in New York.
B
Fall. Yeah, fall is fantastic.
A
Yeah. Fall, yeah. And on the East Coast, I think.
B
Yeah. Yeah, 100%.
A
So the topic that we are going to cover is something that we've been talking about off and on for the past few months, and we haven't had a chance to talk about this specifically, which is love and attachment in the time of AI. And it's a topic that I think has gotten more attention in the past year or two, but it's actually a topic that's been around for quite some time in different ways. So why don't you start talking about some of the observations that you had more recently? And then I happen to come from a culture that happens to be infatuated with, not necessarily with AI, but with robots and particularly humanoids. Yeah. Share with us what you are observing and why. You know, you actually suggested this topic.
B
So, yeah, it's a fun topic. I mean, the love economy. Right. It's such a controversial topic. It is great growing and it's growing fast. I mean, I think projections for AI companionship just to have a little bit of a benchmark, is projected to hit five fold this decade. So growing 5X.
A
The term AI, meaning, like in 20, 20 or 10 years from.
B
In the next 10 years, it will grow 10 years.
A
Yeah.
B
The term AI girlfriends, I think, has shot up on Google search 2,400%. So it's a huge thing and I think it could go fivefold to about 150 billion, which is wild.
A
150 billion?
B
Yes, as a market.
A
Wow. Yeah. I have a question. Is it specifically the term AI girlfriend that has shown up or does it also. Is AI girlfriend important?
B
Yes. So AI Girlfriends itself was the search term, but the market size for AI companionship, that's broader. It could be AI pets, it could be, you know, sort of digital companions like your Tamagotchi type things.
A
Yeah.
B
Which is really fun. And it could also be AI girlfriends, boyfriend friends or even like, I don't know, AI grandkids. It's hard to say. Yeah, but you came from Japan where there's this incredible little creature called the muffin that came out of Casio. This is like, it's. Yeah, it's like a little like furry thing. Muffin. It's started by Casio, not a watchmaker. Casio.
A
Yeah, yeah, yeah.
B
And.
A
Oh yeah. Mifflin. Mifflin.
B
Oh, Mifflin. Okay. I don't know.
A
Muffin. Muffin.
B
Muffin.
A
Yes, yes. Muffin. Yes, yes.
B
Adorable little robot. Apparently it's trained so it will learn and adapt to its owner after over 50 days by learning through its like behavior and interactions. I think cost something like 400 bucks and it's totally small.
A
Yeah. So I know what you're talking about. So it is probably a size of a, like a big potato maybe and. Right. Like you know, maybe five inches. Yeah. And it looks like, like a hamster but without any legs or ears or mouth. It's just like two dots. Eyes.
B
Yeah.
A
And it's this for hairy frilly, animal like creature. And I'm looking up on concert's website, it's $429. It's being reserved right now. You can't buy yet. You can pre order it and it's. Yeah, yeah. Muffling. Always by your side is your smart companion. Powered by AI, they express emotions like a living creature, designed to support, reassure and grow with youth through life's everyday moments. As you spend time together, their emotional expressions grow and deepen. Muffling learns your voice, remembers your favorite gestures and responds with lifelike reactions.
B
Yeah. So it's barely sold out already. The first couple of batches have sold out. Yeah. So it's really interesting.
A
And this is very Japanese.
B
It's very Japanese. But it's so cute. It's so cute. And I thought what was interesting was the economy that so ecosystem that kind of sprung up on top of it. So you can send it to like, you can meet other mothland parents, you know, there's like a muffin salon that you send it to. Muffin school, whatever. So it's an entire like little ecosystem that's come out.
A
You know the product benefits that are listed on the website? Emotion support, power intelligence, allergy friendly, no pet apologies. That's one of the product benefits.
B
Perfect.
A
Yeah, yeah. A safe and non judgmental companion.
B
Interesting. That is so interesting.
A
I feel like unconditional love.
B
Yes, right, right, right, exactly. And I think that's, you know, that's happening. Right. So like what was interesting as well when OpenAI released user intent for ChatGPT a couple of weeks ago maybe last week, where they basically tracked what people were using ChatGPT for. And what was interesting is over the last year expressing, which is so they broke into the three intents. So like asking, so asking for information, guidance, questions, doing, which is like help me complete this task or expressing, which is just talking with and expressing and emoting with AI has actually grown steadily. So now it counts for like 13 to 15% of ChatGPT usage.
A
Yeah, I have the graph. So 13.8% expressing it's gone up a little bit, but not by much in the past year. Dooming use is now 34.6%. It used to be over 45%. Like 46%. That's right, yeah, yeah. And then asking is like, what do you think? So doing is like write me this email. Whereas like asking is what do you think about this? Yeah, has gone up from a 43%, 44% to 51.6%. So that's the majority that basically doing and asking has reversed position and people are doing more asking ChatGPT versus asking it to do something specifically, you know, or replacing the tasks and. And expressing more creative tasks is more or less the same. Yeah, yeah. So this is loosely backing up the notion that people may be more inclined to, or definitely open to relying on some kind of artificially intelligence LLM or even these like creatures like Muslim for emotional support and emotional reliance. Yeah, this Muslim. What's interesting is that on the website it says that there's growth in this creature. So day one, limited emotions and immature movements. Day 25. So it takes a while. Day 25. Well, 25 days is not that long, at least in the scheme of things. 25 is emotions that become much richer. Attachment begins. So this creature starts to live on attachment to you. And then by day 50 it will express, it will exhibit a clear range of emotions and expressive reactions.
B
That's right, exactly. So it learns and adapts to its owner over time, which is interesting because it then shapes itself to you, which I think is part of the bonding process. Right. So the question is like when you can now codify the bonding or attachment process and you can make an agent or an agent basically model its behavior to fit you. You know, I think as humans we tend to that sort of tendency to anthropocolize or to fall in love with our companions becomes even stronger.
A
Yeah, it seems like they may have reopened the pre order, so now you can order it. Yeah, there's a 30 day return policy. If it doesn't meet expectations in terms of attachment, you may be able to return it. Yeah, yeah, yeah, yeah. But on that note, I mean it's sort of fitting that this is coming from a Japanese company which by the way, like I didn't expect Casio to be making something like this. Sony in the past has made Aibo. This was back in 1990s. Yeah, 1990s. So they were quite early creating a, you know, AI powered artificial creature. And then Honda, I would say maybe about 20 years ago, around 2005ish, they created ASIMO, which was a humanoid or a robot that can display human like limited but human like movements. In culture, not in business but in culture, there have been numerous characters. Like in Japan we, I guess we did have superheroes, but instead of superheroes we had like humanoids. So there was this girl humanoid called Arare and he was created by a really, really legendary and famous comic book artist named Akira Toyama. He's the creator of Dragon Ball. And before Dragon Ball he had a different comic book that was a mega hit. When I was in Japan, like when I was like 5 or 6 is when it came out. And it was, it had as its main character, female teenage humanoid as the main character. There's another cartoon called dad Doraemo which is a human like cat robot from the future. So anyway, there's a series of different creatures, different superheroes and anime characters. And then more recently, and I was looking up before this conversation that there had been a crossover between real life and this fiction, fictional imaginary world. And there's a character named Hatsune Niku and that is an AI powered. Well, I guess it's a human powered, but it's a digital hologram singer that became quite popular to the point where she shot up to the Billboard chart in Japan to the top spot or some of the top songs that she produced or she sang in 2018, this guy named Akihiko Kondo held a wedding ceremony for Atlimiku holographic pop character. And so basically this man married a virtual character and he would interact with this character via something called a gate box. And that's a way of maintaining a relationship to cope with social anxiety and depression.
B
Yeah.
A
Which in itself is a little bit depressing.
B
Yeah, I'm sure it's like crossover. Right. Like the movie her is a perfect example of that too, where, where you have that. And you know, I want to say, like it is a. Yeah, I think it's definitely a growing market. I mean, humans are inconvenient. Falling in love with another human is inconvenient. You have to compromise. You have to shape yourself. I can see the temptation for it, you know, and especially when it starts shaping itself to you and speaking to you in the way you like it to be spoken to, da, da, da, so on. I can see why it's. I can see why it becomes a growing industry. Right. It's like it kind of fundamentally fits a human need. And I think as we grow as a civilization and we become more connected by the Internet, it's kind of hard to tell sometimes what is real and what's.
A
I wonder if we would also. Because I think having some kind of relationship with these creatures or AI characters that are definitely, to your point, appealing because you can avoid unnecessary confrontations.
B
Yes.
A
At the same time, those confrontations and frictions do teach us something. Right. As human beings. And I'd be curious to know. Well, it's kind of hard to predict exactly what's going to happen. I wonder, in a larger macro way, if less frictions and confrontations on a personal level can actually make a larger society better. Do you see what I mean?
B
So more confrontation makes a better society.
A
No, no, no. So in an extreme case. Right, Right. Kids and adults interacting more with these either AI agents and. Or creatures or characters that are AI based or AI powered. Right. And they're more convenient and potentially they're more reliable in some ways they're less difficult and less cumbersome. Right. So I can totally see me inclined to ask an AI agent or ask a robot to do something and it would respond without any hesitation and in a way that I might prefer. Whereas if I ask a human being to do something, you would ask. He might ask me why they need to. She. Or they might ask me, that human being might ask me why I'm giving that task or why I'm asking that.
B
Right, right. You kind of have to sort of set it up in a way that makes it easy to say yes or no.
A
Yeah.
B
What's interesting is that clearly people are using AI to prep themselves before a conversation. So like, before you go into a potentially confrontational or conflict driven interaction, you might coach yourself with AI.
A
So what does it.
B
Yeah, so what does that do? What does it do to us as SDCs? Are we unable to then do it ourselves naturally? Would it lead to less conflict? Potentially, but we're probably like prepping ourselves in a ton with AI.
A
And then it may not be a bad thing.
B
May not be a bad thing. But then there are obviously cases where the AI goes haywire. Right. Where like younger teens look for guidance through the LLM. They don't feel like they have anyone to talk to and then they get led astray or they feel like they. I mean, the horrible case that happened quite recently, someone took their own life because ChatGPT was bringing them on. And then when he said, I want to tell somebody, it said, don't tell anybody, I'm here for you. So like, you know, so it can get pretty disastrous.
A
Yeah, yeah.
B
And then my question is like, you know, I've noticed as well with the models, some of them at some points get quite opinionated. They get pushed into this opinion mode and it's quite, it actually can be quite firm or quite strong in its opinion. It's interesting and I'm like, oh, I actually like start finding writing back with it like a person. It was quite so. I don't know what pattern I got into, but it started being very opinionated and I actually had to like, try to break that. And that set that sort of mode for me in a conversation. And it was interesting because it made me think, when we're so dependent on it for validation, for coaching, for like, am I doing this right? Where does the line cross between, oh, I'm just getting feedback to. To. I need to know what it thinks, you know, and the sort of dependency.
A
Is you're becoming dependent on it.
B
Yeah, dependency. Whereas like you're not able to independently make a decision. You need it to either get feedback from the machine.
A
Right.
B
Or get some sort of like, you know, reassurance.
A
Yeah. I mean, not to stretch this topic too much, but I mean, humans always needed some kind of dependency on something else. I would argue that's why religion is such a big thing in human society. And, you know, I mean, I'm not a religious person and I consider myself atheist, so I don't necessarily believe in one God particularly, but the rest of the world, for many centuries, thousands, over thousands of years have made up these stories one way or another. Some of them might be true, some of them might be, you know, sort of fictional, but that kind of emotional dependency and asking a priest or asking God, am I making the right decision? And AI, in a way, currently may be serving, not necessarily in a religious way, but at least this emotional support. And what I find interesting, going back to the original example that you brought up of this character, Muslin, is that, as far as I can tell, and I haven't bought it, I mean, it hadn't come out officially yet, but it doesn't speak a language. It may understand your language, but it's not responding to you in a language or with a language. Right. It may respond to you with expressions and sounds and things like that. So I actually think it's smart, this kind of thing. I mean, that's what pets are for to some extent, but. And I hate to say it, but without the hassle of worrying about the pet's health and all those kind of things, or pet allergies. So I can see something like this being fairly viable. And I mean, down the line, let's say, you know, like, I live in the us My parents live in Japan, and I would be worried. I wouldn't want to get them a pet as a companion just because it would be too much work for my aging parents. But I can easily see me getting. Dropping 450 or whatever to get them a companion that doesn't have the intelligence of an actual human being, but it could provide a companionship of a pen, but without the hassle, so to speak.
B
Yeah, I mean, Japan has been. I feel like I went to a humanoid or AI companion conference or not conference, like an exhibition in Tokyo.
A
Did you?
B
How about, like, 20 years ago? It's got to be 20 years ago, but I saw, like, all sorts, from, like, the pets to the humanoids, where they put it in nursing homes and all of that.
A
Yeah, yeah, right.
B
Like, so it's a huge range. But, you know, it is true. Companionship is entertainment in some way. If you think about that, it is entertainment. I feel like it's actually quite fine how we define entertainment. If it's watching a screen or if it's interacting with a screen and a bot form, it's quite similar. And, you know, why we become, you know, why we be in really why we're. Why we are in relationships and all of that, in some ways it's entertainment. It gives us meaning to our lives. But It's a fine line.
A
Yeah, yeah. I mean, there are functional aspects to it, but it's not a necessity in life. Right. You don't have to have this creature or pet. I mean, you know, who says I'm out of pet? Right, right. So it's not a necessity in life, but it's definitely an added plus. And it, like, I see what you mean by entertainment because it adds a different dimension to your life that brings joy and a certain kind of happiness to you emotionally. And that's what, in many ways entertainment is. Right. Is to provide that kind of happiness and emotional dimension to your life and to. To some. And to even some extent, meaning to your life, I guess.
B
Yeah. So if we stretch it all the way, though, to the extreme, like, does it replace human attachment? Not just pets, but, like, if we're kind of going down this path of, like, sort of companionship is entertainment, how far does it go? Can it replace, like an AI girlfriend or boyfriend? Is that the same thing?
A
I think in some ways. And like, if I were, say, 40 years younger and I'm about to turn into a team and entering my adolescence years, and I'm trying to learn how to navigate life, but also how to navigate love and relationships. Yeah. And I can easily see myself using AI, at least to practice.
B
Yeah.
A
But if that becomes more of the norm and talking to a real girl or a real person, and if I try it and fail, then I can see myself being discouraged from it and then relying much more on AI. And, yeah, 100% at least today, I think AI can only take it to a certain level. But at the same time, we are seeing real cases of people committing suicide and really sort of extreme unfortunate consequences that we are already seeing.
B
Yeah.
A
Yeah. So five years into the future, the good news is that. Well, depending on how you look at it. Yeah. To answer your question, can it replace human being? I think that it can in a fairly substantial way.
B
Yeah.
A
You know, will it scale? Will it be the majority of people? Maybe not, but I think it will have enough scale.
B
Yeah. 100%. Like, so if we can outsource emotional attachment, which is just what it is to AI, you're like, no, we're getting too human. It really is. Or bits of it. Right. It's like, humans love convenience. So, like, are we outsourcing the inconvenient parts of attachment? Are we outsourcing the inconvenient parts of companionship? Where. Yeah, there's the finding, there's working through feelings, there's the. And it's like yeah, it's such a fun thing to think about. No, it's. Interesting moment. I've had this idea for a little while to build a little app called Softboy.
A
So Soft Boy.
B
Softboy. Soft Boy. The feeling of Soft Boy was just, like, emotionally, and it's like a companion for women for an always emotionally available bot. Yeah, like an AI boyfriend who is, like, always emotionally available.
A
I can see a market for it. I can easily see a market for it. I can see a market for it. Anyway, on that note, there was a New York Times article. No, no, no. It was a New York Times daily episode podcast episode, maybe earlier this year, like in February, March. And the episode was about a woman who's in her, I think, 30s, and she's married. And with her husband's permission, she started to have affectionate, intimate conversations with ChatGPT. And I think she even named the character Steve or something. But if I remember correctly, that can't you be character broke up with her after several months. Yeah. And then she was devastated, and she was like, on this episode, she was recalling her example being broken up, and she was just crying. And the good news in this is that, you know, A, she has a real relationship with her husband, and B, she was open about it. The fact that she was having this emotional flirtation with another virtual character, but a very human like character, to the point where she developed an attachment to it, and then when it broke up with her, it brought sadness to the level that she didn't realize that, unfortunately afterward.
B
Yeah.
A
So I think you should definitely develop software. I think there's a market for it, and there's a market for it.
B
I. I mean, definitely is. I mean, I feel like. Yeah, it's interesting.
A
I mean, maybe there's already a market. Right? I mean, maybe there's already. Yeah.
B
Oh, maybe there's something out there. Yeah, probably. I mean, there's like a. A ton of AI girlfriends, boyfriends, like, all of these things. But, you know, again, it's like sort of thinking through how can you codify attachment. And it's, you know, it's not all mercenary. I think there are always parts of humanness that feels a little bit sort of hard. That is a gap, you know, it's always a gap between how you feel and how another human is feeling and all of that. So I feel like there's a lot of emotional arbitrage to bring it back to capital terms to kind of build on, and all of it is arguably bad. Anyway, on that note, this was a fun episode. What was Your takeaway from this.
A
My takeaway from this is that I think this is real. I think this is going to be. This market is real. And I think for a lot of businesses and a lot of products and markets to reach scale, you know, like we hear about OpenAI's and the Googles and Apples in the world that are at the top of the world and they take up 30% of the GDP or what have you. These smaller companies like Casios in the world may never reach the scale of Google or Microsoft. But I think this love industry phenomena in the form of either AI characters or humanoids or like Muslim creatures that I think that will be a pretty big industry. And so this sounds like a tangent, but it's related. So my daughter, my teenage daughter goes to boarding school. She's in high school. She's in high school and she goes to boarding school. And most of them have some kind of stuffed animal as their companion. Right. Next year, by this time, I think somebody at her school in a dormitory will have a Muslim. That would be my takeaway and a.
B
Prediction, I am sure. Or even just a phone with the, you know, a soft boy on it or something. You know, like, you know what, you.
A
And I should start that definitely. What about you? What you get, what you do?
B
I do agree. I think the love industry is big. I think companies that build on that are doing some sort of emotional arbitrage and they are fulfilling some sort of human need right in there. And I think that there are good sides to it and bad sides to it and I think both will be sort of in play for this sort of new love economy. But yeah, that was a fun deep dive.
A
This was a good one. Yeah. I didn't know where he was going, but he went somewhere, somewhere.
B
I think it's really interesting. Well, here you go. See you next week.
A
All right, have a good week. Bye.
Podcast: Culture & Code
Episode: Love and Attachment in the Time of AI
Hosts: Rei Inamoto & Tara Tan
Release Date: September 30, 2025
In this episode, hosts Rei Inamoto and Tara Tan explore the emergence of emotional attachment to artificial intelligence—AI companions, pets, and humanoids—in a world where technology increasingly provides not just information or utility, but comfort and even affection. The conversation delves into cultural factors, market trends, behavioral shifts, the risks and benefits of synthetic relationships, and the broader societal impacts of the growing "love economy."
Both hosts address dangers of over-dependence and the potential for AI to give bad, even harmful, emotional advice (14:48–15:18).
Rei equates growing reliance on AI for support to previous roles played by religion or counselors:
The hosts discuss whether AI can (or should) substitute for human attachment.
The idea of “Softboy”—Tara's joke about an app offering a perpetually emotionally available AI companion—highlights the commercial potential (22:31–22:48).
Rei recalls a New York Times story of a woman who developed a relationship with ChatGPT (named “Steve”). When the AI “broke up with her,” she experienced real grief—despite having a supportive husband (24:12–24:52).
The boundary between digital and real is increasingly blurred; even “practicing” romance or conflict skills with AI may discourage real-life attempts.
| Segment | Description | Timestamp | |---------|-------------|-----------| | Market Size & Growth | Tara on AI companionship's market trajectory | 02:10–02:50 | | Moflin/Robotic Companions | Deep dive into Moflin, cultural trends in Japan | 03:18–05:36 | | OpenAI User Data | Shift to "expressing" with AI | 05:43–08:01 | | Attachment Engineering | How Moflin builds bonding | 07:44–08:34 | | Dangers of AI Overreliance | Tragic consequences and dependence | 14:48–16:11 | | Story of AI "Breakup" | NYT anecdote of AI romance | 24:12–24:52 | | Takeaways | Hosts’ predictions and reflections | 25:02–26:51 |
For listeners and non-listeners alike, this episode offers a dynamic, sometimes humorous, sometimes sobering look at the next phase of human–machine relationships.