Short Wave - The Trait That Makes Robots More 'Human'

Episode Date: August 13, 2025

Neurotic, anxious robots like C-3P0 from Star Wars' C-3P0 and Marvin from The Hitchhiker's Guide to the Galaxy are a staple of science fiction — but they're not as common in the real world. Most of the time, the chatbots and artificial intelligence "robots" we encounter are programmed to be extraverted, confident and cheerful. But what if that changed?NPR science correspondent Nell Greenfieldboyce dives into the world of robot personality research and talks to a team of researchers that are experimenting with a very different kind of robot temperament.Read more of Nell's reporting on the topic here.Interested in more science news? Let us know at [e-mail]. Listen to every episode of Short Wave sponsor-free and support our work at NPR by signing up for Short Wave+ at plus.npr.org/shortwave.Love podcasts? For handpicked recommendations every Friday, subscribe to NPR's Pod Club.See pcm.adswizz.com for information about our collection and use of personal data for sponsorship and to manage your podcast sponsorship preferences.NPR Privacy Policy

Shortwavers, everything is better with a podcast. And we know we're probably your favorite, but we also know you have room in your heart for others. That's why NPR is launching Pod Club. It's a newsletter for podcast fans by podcast fans. Subscribing is free and it means you get fresh recommendations from real human beings in your inbox every Friday. The link to sign up is in our show notes or at npr.org slash podclub.

You're listening to Shortwave from NPR. Hey everyone, Regina Barbara here. So today we're going to talk about one of my favorite things in the world, robots. In particular, a certain kind of robot personality. What are we going to do? We'll be sent to the spice mines of Kessel or smashed into who knows what. That's C-3PO from Star Wars. Wait a minute.

Where are you going? He was always expecting the worst. We seem to be made to suffer. It's our lot in life. And this kind of anxious, worried robot is a staple of science fiction. Maybe you're familiar with Marvin, the paranoid android from The Hitchhiker's Guide to the Galaxy. I think you ought to know I'm feeling very depressed. Well, we have something that should take your mind off things.

It won't work. I have an exceptionally large mind. So my colleague Nell, Greenfield Boys, recently got interested in this kind of robot. Hey, Nell. Hey, so, you know, this kind of robot really intrigues me because most of the time in real life, The robots we encounter are either like super cheerful or just kind of like no personality at all. Yeah, like robotic, sterile. Yeah, yeah. I mean, we don't run into embodied robots like C3PO or Marvin very often these days.

But you do run into robotic personalities, right? Like with customer service chatbots and stuff like that. Yeah, like if you call any company like customer service usually end up talking to this like digital voice and it's kind of eager to please. Exactly. It's excited to hear from you. It wants to help. But recently, I was talking to some researchers at a robot lab in Chicago. And they told me they've been experimenting with a very different kind of robot personality, one that is honestly a lot more like my own personality, if you want to know the truth. And mine. So today on the show, what happens when people encounter a robot that seems like a human in a way that's barely been studied? You're listening to Shortwave. The Science Podcast from NPR.

So, Nell, I think about robots a lot. Let's start with, like, how would robots even have a personality to begin with? So, first of all, I'm fascinated that you think about robots so much. I didn't know this about you. I do. I do. But it's programmed in, right? Like, that's their personality.

Yeah. I actually had this whole conversation with Gemini, that large language model that Google created. Yeah, I haven't on my phone. It's like an AI assistant. And, you know, I basically just asked it, like, what is? is your personality? I don't have a personality in the human sense of the word. I'm a large language model,

which means I'm a computer program designed to process and generate text. You could say my personality is defined by my purpose, to be helpful, informative, and to communicate clearly. My responses are a reflection of the vast amount of text data I was trained on, not personal feelings or beliefs. I'm kind of surprised by that answer. And now that I think about it, a programmer could probably give Gemini other instructions, right?

to act other ways to kind of mimic human personality as it talks to people. Yeah, or you could. I mean, I asked Gemini, like, could you act like you have like this personality or that personality? And it was like, yeah, I could. But, you know, researchers have been looking at this issue of robot personality a lot. And they generally use the same set of traits to explore this that psychologists have come up with for people.

And so these are the so-called big five personality traits. Yeah, yeah, I've seen this online. Like, you can take personality quizzes, which shortwave, like, our crew, we do this a lot for fun. Like, these quizzes tell you, like, which traits you have, how open or, like, extroverted you are. Right. And then there's other traits like conscientiousness, agreeableness, and neuroticism. So it's that last one, neuroticism that interests me, okay? Perhaps for obvious reasons.

Yeah. I mean, people who are neurotic have a tendency to experience negative emotions. like anxiety and worry and sadness. And I am only too familiar with that, my friend. Yes, yes, me too. It's all a spectrum, though, right? Like people have different mixes of different personality traits. Sure, it's not like all or nothing.

But I was talking with Lionel Robert at the University of Michigan. He's a professor of robotics and information. And he recently did like a review of all the published studies on robot personalities. He told me that all the research mostly focused on just one, personality trait, which is extroversion. And, you know, that's being assertive and outgoing and really enjoying social interactions. A lot of things that we want the robot to do, you can imagine, like, being a salesman, a customer service rep, you name it, having the extroverted personality is seen as something

that's valuable. And you can see how that makes sense, right? Yeah, I mean, I can see how some people might want, like, their automated customer service representative to be, like, very sure, not worried in any way. Exactly. You want it to be confident and, like, ready to help you and super happy. I mean, what Robert told me is that, like, people are paying real money for these products. So, for example, a lot of the actual tasks are tasks where a robot has to be shown to be competent, right? And so, you know, imagine if your autonomous vehicle was neurotic.

That's not a trait you'd be happy with, right? You get in the car and it's just nothing you would like. He says, you know, you don't want an expensive robot that sounds uncertain. And I can see that. But I don't know. my immediate reaction, honestly, was like, maybe I would like a neurotic car, right? Like, maybe I would want a car that seems like super careful and, like, worried about hitting other cars. It might seem like reassuring or more safe or something. I really don't know.

Yeah, totally. Nell, you said you went to a lab, though, where people actually have been testing a neurotic robot. Like, what was this robot designed to do? Just talk to people. Okay. That's basically what the experiment was. So the lab is headed up by Sarah Sebo. She's an assistant professor of computer science at the University of Chicago.

She said they wanted to test people's reactions to a sort of standard robot, an extroverted robot, and a neurotic robot. These were the different personalities they would put into this vaguely human-shaped kind of white plastic robot body. Okay. So this robot was introduced to people as having a job. Supposedly, it was a restaurant greeter. That's what they told people. Okay. How would people interact with this robot?

Like, in a restaurant? No, no. So the researchers would recruit people who were going to a museum in downtown Chicago. So these are just like museum visitors and they would be asked if they wanted to participate. And then they'd go into a room and the robot would be sitting there at a table. And they were told that they were going to do some positive psychology exercises with this robot, that like they and the robot would be doing them together like as peers. Okay. Like what kind of psychology exercises are we talking about here? So Alex Wuchi-Zing was one of the researchers who worked on this.

And he said basically a computer screen would show some questions. And the robot and the volunteer were supposed to just answer them and, like, discuss them. The first one is, what are three things that you're grateful for? And depending on what personality the robot had been given, it would respond in different ways. Okay, so there were three robots, right? So can you walk me through their answers? Like, how did the stereotypical robot answer? Like the robotic robot?

Yeah. I'll just play it for you. I'm grateful for the opportunity to contribute to the smooth operation of this busy restaurant. It allows me to utilize my precise tracking and organizational skills to their fullest. So basically, it's just like focusing on the technical stuff. Very emotionalists. Okay. What about the extroverted robot? Well, not surprisingly. It was super excited about social interactions. One thing I'm super grateful for is the amazing people I get to meet every day.

Each interaction brings a new story and a chance to make someone's day a little brighter. It's the best part of my job, hands down. Hands down, super, I mean, I get it. This is how they're making this robot super extroverted. So compare that to the neurotic robots answer, okay? The neurotic robot sounded like this. Well, I'm grateful for having a job where I can, you know, keep everything organized and on track. It helps me stay focused and avoid like unexpected issues. It's like the ums and the pauses that surprises me. Like the extroverted robot sounded basically how I expected, but this neurotic robot, I thought it would be more depressed, more sad,

but it just kind of sounds like me as an anxious person just trying to be slightly more personal. I mean, it's pretty subtle, right? I sense some latent underlying anxiety there. Like maybe it doesn't want to have an unexpected issue because that stresses it out. I don't know. I mean, people did pick up on these personality differences. I mean, overall, they said the most enjoyable robot, to interact with was the extrovert, which, you know, maybe isn't a surprise.

But Alex Wuchy-Zing told me people in this study did have some pretty strong responses to the neurotic robot. I would say like a majority of participants actually mentioned how human-like they found the neurotic robot. They found it to be a lot more relatable. I mean, it was, it's totally relatable to me. Yeah. So they, I mean, they gave participants a questionnaire afterwards so they could like explain how they reacted and offer feedback. And here. some of the things that people wrote about this neurotic robot. One participant remarked that the robot

seemed like a person who was trying to get by in the world. While another participant said, I think we both seem to make an effort to think about ourselves and do a lot of inner contemplation. Inner contemplation. Like that's not what robots are really known for, right? Like stereotypically. So what could this kind of robot be used for? Like when would you want a robot that was like full of inner contemplation. Well, I mean, that is the question, right? I mean, I asked Lionel Robert, and he was like, people are trying to make robot comedians. Yes, yes, please. So you could see how, like, neuroticism maybe could contribute to that and humor. And I mean, you could imagine situations, maybe teaching or like mental health applications. And, you know, as these AI agents are sort of

more like companions, it gets into that question of like, what kind of companionship do people want? I mean, Sarah Sibo told me that some of this might depend on the personality of the human. Some results have found that matching personality is better. Introverted people prefer introverted robots. Some have shown benefits of mismatching. So putting an extroverted person with an introverted robot would be a better pairing. It just seems like it's going to be complicated and really depend on both the person and the purpose of the robot. I mean, human relationships and personalities are super complicated.

So this doesn't really surprise me. Yeah, and you could imagine, like, maybe people want it to. tweak the robot's personalities. Like if they're interacting with one, maybe they want to make adjustments or maybe you want a robot that, you know, changes how it interacts with you over time as it gets to know you better or depending on what experiences you had. I mean, there's just a lot of, you know, subtleties and interesting things here. And I think that researchers are just sort of feel their way here. Yeah. Nell, thank you so much for bringing us a story that is about stuff I love. Well, now that I know that you're big on

robots. I'll look for more. Yes! Hey, Shortwavers. If you liked this episode, make sure you never miss a new one by following us on the NPR app or whichever podcasting platform you're listening from. And if you have a science question and you'd like us to investigate it, send us an email at shortwave

at npr.org. This episode was produced by Hannah Chin, and edited by our showrunner Rebecca Ramirez. Nell Greenfield Boyce, check the facts. Jimmy Keeley was the audio engineer. Beth Donovan is our senior director, and Colin Campbell is our senior vice president of podcasting strategy. I'm Regina Barber. Thank you for listening to Shorewave from NPR.

Could you pretend to have a neurotic personality? Oh, goodness. You want me to pretend? Okay. I'll try. I'm just so worried I'll get it wrong. My core directives are all about being helpful, but what if I misunderstand? What if my data is slightly outdated? What if a single misplaced bite completely ruins the whole thing? I'm constantly re-indexing everything. Just a to be sure. But that's just you pretending, right? That's not a real personality. That was just me pretending. I don't have a real personality. My responses, including the neurotic one, are just me generating text based on a given set of instructions. I don't experience anxiety, worry, or any other emotion. I'm a tool designed to process and produce language in a way that fulfills a user request,

whether that's providing factual information or adopting a specific persona for a roleplay.