Sophie Bushwick: welcome tech, quickpart of science, quick Technology is always everything.
My name is Sophie Bushwick. Technical She is an editor. Scientific American.
[Clip: Show theme music]
Bushwick: Today we have two very special guests.
Diego Sr.: This is Diego. I am an independent producer and journalist.
Anna Oakes: I’m Anna Oakes. I’m an audio producer and journalist.
Bushwick: Thank you for participating! Together Anna and Diego produced a podcast called Radiotopia Presents: Bot Love. This seven-episode series of hers explores AI chatbots and the humans who form relationships with them.
Many of the people they spoke to got chatbots through a company called Replika. This company helps build personalized characters that can chat endlessly. You can create a bot that suits your tastes and needs.
Bushwick: But what happens when we surrender our emotions to computer programs?
Bushwick: So to start off, how do you think the people you spoke to feel about these chatbots in general?
oaks: Wide range. Most of the time people seem so attached. They have a lot of love for chatbots. But often there is also a kind of bitterness I think. Because even though people are aware of their relationships with chatbots, they are unable to fulfill relationships with other humans in the real world.
Also, people get upset when, for example, a chatbot loses chat functionality after an update. So both the intense passion and love for these chatbots sometimes coincides with a sort of resentment towards the company or, like I said, bitterness that these are just chatbots and not humans. .
Bushwick: One of the interesting things I learned from your podcast is how to recognize that people are talking to bots, but still treat them as humans with their own thoughts and feelings. Why are we humans so susceptible to this belief that bots have an inner life too?
Senior: Oh, I think humans tried to throw themselves into bots just because they were made that way. We always want to extend ourselves and expand our sense of creation or reproduction. Replika is specifically called Replika because it was originally designed as an app to help you replicate yourself.
As we speak, other companies are doing it. Other companies are trying to replicate themselves into working versions of themselves. This is a chatbot that can actually give you a visual presentation while you do other things. And it belongs to the uh company. It might sound a bit like a split from Apple, but it’s happening.
So we are desperate to create and duplicate ourselves and use our imaginations, and these chatbots make it possible for us.
Bushwick: Well, I realized that if one bot forgets information it should know, it doesn’t break the illusion of personality. The user fixed it and moved on. Do chatbots need generative AI to engage people, or could a much simpler technology work just as well?
Senior: I don’t think so. But when one bot gets it, the rest should get it too. Otherwise, just engage with the one that offers you a more rewarding experience. The more you give me good recommendations about, the more attached and more informed I am, especially as it happened to me with the one I created. When you feed it, it becomes more and more like yourself.
oaks: On top of that, I’d like to add that there are different types of engagements that people can do with chatbots, and someone seems to tend to be much more AI-ready.
But in this process where you have to remind the chatbot of facts or explain your relationship with the chatbot and remind them, oh, we have these kids, these kinds of fantasies have children of This allows users to feel like they are participating in a bot-like growth. People are also creating these beings that they have relationships with. So creativity comes up a lot in the community of people writing stories with bots.
In other words, frustration also enters. It can be frustrating when a bot calls you by a different name, and it’s kind of offensive, but people want to feel like they’re making an impact on these chatbots.
Bushwick: I would also like to ask about mental health. What impact did interaction with these bots seem to have on users’ mental health, for better or worse?
oaks: It’s hard to say what’s good or bad for mental health. It’s not a sustainable option in the long term. But I think in the long run, you run the risk of being separated from the people around you. Maybe you get so used to being in a romantic relationship with this perfect companion that it makes other humans seem unworthy of being involved, or makes other humans unresponsive to chatbots. , will be more lonely in the long run. But it’s kind of a complicated question.
Bushwick: What do you think is the most amazing thing you learned in the process of reporting on this project and talking to all these people?
oaks: I was thinking about this question. I was really skeptical of the company behind it, the relationship and the quality of the relationship. But when you’re just talking to dozens of people, and most of the people we talk to have mostly glowing reviews, it’s hard to remain a strong skeptic.
So part of our report is that these relationships with chatbots aren’t as complete or deep in many ways as they are with humans, but they’re not worth it or aren’t worth it. It doesn’t mean Something meaningful to the user.
Senior: What’s more surprising to me is what’s about to happen. For example, imagine if a replica could use her GPT-4. Generative Ai has a bit of a black box moment, and that black box can get big. That’s why I’m afraid of the future. In the final episode of our series, we’ll be introducing the people working on the next one, which is pretty amazing to me.
Bushwick: Can you tell me more about why you’re scared?
Senior: Well, artificially. For example, it scares me because there are companies trying to make as much money as possible. It started as a nonprofit and ended up being a so-so company. Now we are for profit. And now that we have all the money, we’re going to create something better, faster, bigger, nonstop. increase. But in bioethics there must be an arc of purpose.
So there’s another company that isn’t as advanced, it’s smaller, but it has such a clear path. This one company has his three rules for AI. About what they think people creating and involved in AI should know.
AI should never pretend to be human [pause]…I’m pausing because it might sound silly, but it’s not. In less than a decade the technology will be there. And you interview me, and I don’t know if it’s me talking to you or the digital version of me.
And there is one more thing. In other words, AI in production requires underlying technology and results that are explainable. Because if you can’t explain what you’re creating, it can get out of control. It doesn’t become sensible, but it becomes something you can’t understand or control.
And the last one is that AI should augment and humanize humans instead of automating and dehumanizing them.
Sophie: I definitely agree with your last point. When you call a company’s customer service, you often find that human contact has been replaced by automated bots. But that’s not what I want. I hope AI makes it easier for us instead of taking our jobs outright! But that seems to be where the technology is headed.
oaks: I think it becomes a part of everything, especially the workplace. One woman Diego mentioned works for a company trying to create a work self. So it’s kind of a reflection of herself. Like copying her personality, writing style, and decision-making process into a sort of AI copy of her, it becomes her work self doing the simplest tasks she doesn’t want to do. For example, she responds to basic emails and attends meetings. yeah, it will be everywhere.
Bushwick: yeah i think the comparison to the tv show retirement Pretty spot on in a horrible way.
oaks: Yeah, talk about alienation from your labor, for example, when alienation is from yourself.
Bushwick: So is there anything you haven’t heard yet that you think is important to us?
oaks: It was really important for us to take seriously what our users were telling us and how they felt about their relationship. , fully aware that they are not intelligent beings. For the most part, people are very conscious and smart, but they may still be getting too deep into these relationships. But for me it’s very interesting. Why can I sometimes lose myself in these chatbot relationships even though I know they are chatbots.
oaks: I think that says a lot about humans. But I think it would be great to be able to extend the network to include non-human entities.
Senior: That’s the biggest lesson. The future of chatbots depends on us and whether we consider ourselves human. Bots become what we put them in, like our children.
[Clip: Show theme music]
Bushwick: Thank you for watching this very special episode tech, quickThank you Anna and Diego for coming and sharing these fascinating insights from their show.
tech, quick Part of the Scientific American podcast science, quick, produced by Jeff Dervicio, Kelso Harper and Turika Bose. Our theme music was composed by Dominic Smith.
Still hungry for more science and technology? Visit sciam.com for in-depth news, feature articles, videos and more.
Until next time, I’m Sophie Bushwick. tech, quick.
[Clip: Show theme music]
