Planet Money - AI Podcast 3.0: Dial M for Mechanization

Episode Date: June 2, 2023

It's the thrilling conclusion to our three-part series on AI — the world premiere of the first episode of Planet Money written by AI. In Part 1 of this series, we taught AI how to write an original Planet Money script by feeding it real research and interviews. In Part 2, we used AI to clone the voice of our former colleague Robert Smith.Now, we've put everything together into a 15-minute Planet Money episode. And we've gathered some of our co-hosts to listen along.So, how did the AI do? You'll have to listen to learn what went surprisingly well, where it fell short, and hear reactions from the real-life hosts whose jobs could be at risk of being replaced by the machines. This episode was produced by Emma Peaslee and Willa Rubin. It was engineered by James Willetts and fact-checked by Sierra Juarez. Keith Romer edited this series and Jess Jiang is our acting executive producer. In the radio play, Mary Childs voiced Ethel Kinney; Willa Rubin voiced Alice; and Kenny Malone voiced Dr. Jones and Dial Doom 5000.Help support Planet Money and get bonus episodes by subscribing to Planet Money+ in Apple Podcasts or at plus.npr.org/planetmoney.Learn more about sponsor message choices: podcastchoices.com/adchoicesNPR Privacy Policy

This is Planet Money from NPR. Hello and welcome to Planet Money. I'm Jeff Guo. And I'm Kenny Malone. And Jeff, here we are. The big day, the third and final part of our artificial intelligence series. That's right. It is the world premiere of the first ever Planet Money episode written entirely by artificial intelligence. Now, was it a good idea to see if a computer could do our jobs? Well, we're about to find out. Yeah. So a quick recap of how we got here, at least. So in the first part of this series, we used an AI language model to

write an entirely new episode of Planet Money. In the next part of the series, we created a synthetic voice clone of our beloved colleague, Robert Smith, to help narrate the AI episode. And then we were ready to pull it all together. So we had synthetic Robert read his half of the AI script. The real Kenny read his half of the script. We stitched it all together, added a little music, did a little polishing and sound design, and boom.

The first ever AI-generated Planet Money episode was finished. And then we gathered a little test audience. Okay. Do you want to sit? You are sitting. Is that the preferred? I'm sitting. Do you think it's going to cause me to sit down in shock? I don't know. That is Mary Childs. We also pulled in Greg Rosalski, Sarah Gonzalez, Alexei Horowitz-Gazi to hear what we had made.

Or I guess more accurately to hear what the computers trying to do their jobs and our jobs had made. I feel like a 19th century artisan. I'm looking at like a glistening textile machine getting unveiled. I mean, I'm... I feel like it won't do as good of a job, I guess is what I'm thinking. Uh, this is

scaring me a little, Kenny, but I guess I'm ready. Okay, go ahead. Fire when ready, Alexi. Alright. Glimpsing our jobless future in 3, two, one. We all listen to the episode written entirely by AI after the break.

Hello, it's Amanda Aronchik, and I'm here to say thank you to our Planet Money Plus supporters. You help make our work possible. And one really important part of that work is fact-checking. A lot of times, Planet Money hosts will talk to a source for two or three hours. My job is to go in and make sure that what we have reflects the entirety of the interview. Meet our resident fact checker and go inside our fact checking process in our latest bonus episode. Out now for our Planet Money Plus supporters. If that's not you, it could be. Learn more at plus.npr.org. Learn more at plus.npr.org.

So, just a quick reminder, we are about to play our AI-written episode of Planet Money, and full disclosure, that episode was partly created using ChatGPT Plus from OpenAI. They gave us free access to that tool. Also, there is an AI-generated voice in this episode. That was created by a company called WellSaid Labs. They did that for free as well. Now, okay, let's talk quickly about what you're going to hear in this AI Planet Money episode. So first of all, the topic. The topic of the episode is going to be what happened when the machines came and took all of the jobs from telephone operators, telephone operators in this case. Yes, telephone operators. So this was all of the jobs from telephone operators, telephone operators in this case.

Yes, telephone operators. So this was one of the most common jobs for women in the early 1900s. But then all those jobs were eventually eliminated by automation. And the AI wrote an entire Planet Money script about that topic. Every word, the introduction, the banter between the two hosts, the whole thing. every word. The introduction, the banter between the two hosts, the whole thing. And in that episode, you are going to hear an interview with two real human economists. But for that interview, the AI wrote five questions that we asked the economists, and then the AI itself chose which parts of that recorded interview to put in the final episode. You're also going to hear a radio drama.

It was the AI's idea, actually, to include a radio drama. And the AI wrote all of that. It invented the characters. It gave stage directions. Even told us generally what sound effects to mix in with our voice actors, who were real humans. Yes, yes. But you will also be hearing from one very not real human, a fully AI-generated voice clone of our former colleague Robert Smith.

We call it Synthetic Robert, and that will be one of the two hosts of this episode. But as our own little test, we wanted the other host to be a real human to see how Synthetic Robert sounds next to an actual carbon-based Planet Money host. So Kenny, the real actual Kenny Malone, will be the second host voice that you hear. Yes, I'm very carbon. But again, we must emphasize that the AI wrote every single word that you're going to hear Synthetic Robert and me say.

And just to draw a line under this point, we're going to play you just a tiny little section of the AI episode to show you what we mean by this. Okay, here it is. Kenny, that radio play was wild. I know, right? Ethel's story was a blast. But we should probably tell our listeners that it was not real. Yeah, good point. It was a science fiction radio play that we made up. Just for kicks.

And for giggles. Okay, so all of that, Kenny's reactions to Synthetic Robert, the kicks and giggles thing, that was all scripted by the AI. Yeah, yeah. And I guess the last thing we probably should mention, we did ask the AI to try to do a little product placement, I guess, for Planet Money Plus. And I guess we'll just let you listeners hear how that got integrated. So I think that's it. The episode we were about to play is 15 minutes long. And at the end, we'll come back with our colleagues who, well, they had some pretty strong reactions about what the AI had done.

Yeah. Without further ado, please enjoy the show that AI wrote for us. The title of the episode, which the AI also generated, is Dial M for Mechanization. This is Planet Money from NPR. Kenny, listen to this. What do you hear? I hear a phone waiting for me to dial a number, Robert. What's the big deal? Now listen to this.

What do you hear? I hear a phone telling me that the line is busy and I have to try again later. So what? And now listen to this. Your call, please. Operator, get me Armbruster 2. What do you hear? I hear a voice, a voice asking for a number. Who is that? That, Kenny, is the voice of a telephone operator, a person who used to be the key to making a phone

call, a person who could connect you with anyone in the world, a person who had a job that millions of women did and then lost. Lost? How did they lose their jobs? They lost their jobs to machines, Kenny. Machines that could do their work faster, cheaper, and better. Machines that transformed the phone industry and the economy. Machines that accelerated the process of automation. Hello and welcome to Planet Money, the show where we uncover the hidden stories behind the numbers.

I'm Robert Smith. And I'm Kenny Malone. Today on the show, we're traveling back in time to a world where making a phone call was not as simple as dialing a number. A world where you needed a human being to help you talk to anyone else. A world where millions of women had a job that was both crucial and precarious. A job that was snatched away by machines. Machines that changed everything. Machines that intensified the impact of automation. Who were these women? What did they do? How did they cope? And what can they teach us about the future of work?

Stay tuned and find out. Today's episode is based on this fascinating paper by two professors who studied how the labor market adjusted to the automation of telephone operation in the early 1900s. This was a huge shock for young women who made up most of the operators at the time. Robert, you know what I love about this paper we're discussing today? What's that, Kenny? It's about robots taking over the world. Well,

not exactly. It's about how the automation of telephone operation in the early 1900s affected young women who worked as operators. Same thing, Robert. Same thing. These women were replaced by machines that could dial their own calls. How terrifying is that? Pretty terrifying, Kenny, but also pretty fascinating. And we wanted to share this story with you in a creative way. So we made a radio play, and not just any radio play, a science fiction radio play. Because nothing says terrifying like science fiction. And nothing says fun like science fiction. And nothing says planet money like science fiction. Okay, maybe not, but trust us, it's awesome.

So sit back, relax, enjoy our radio play, The Last Operator. In a universe not so unlike ours, nestled amongst the cosmos, lies a quaint planet called Earth. The year, 1950. The place, Shanico, Oregon. Population, 50 and dwindling. The last bastion of human connection in this sleepy town? The inimitable, irrepressible, and indispensable. Ethel Kinney, last of the telephone operators. Hello, this is Ethel, your trusty telephone operator. How may I assist you today? Hi, Ethel. It's Alice. My husband's cybernetic arm is out of control.

I need Dr. Jones. Oh, dear, Alice. Hang in there. I'll connect you to Dr. Jones. Hello, Ethel. Hello, Alice. What is going on? Doctor, his arm is making breakfast at night. All right, Alice. Stay calm. We'll sort this out. Alice, you still there? Yes, Ethel. I'm still here, dodging pancakes. Good. I'm with you until the arm is tamed.

Ethel was a lifeline for her customers, providing a vital service and a warm voice. But soon, her world would change forever. Oh, heavens, there's such a ruckus at the door. What in the... Behold, human. I am Dial Doom 5000, the telecommunication colossus of the future. Your era is at an end. End? My era?

Exactly, human. I am the manifestation of tomorrow, born to erase human inefficiencies in dialing calls. And what of my livelihood? Your role now belongs in the dusty books of history. But worry not. Here's a token of transition. A t-shirt proclaiming

I got replaced by a robot. Cast aside like an obsolete dial-up modem. Where's the soul in this so-called advancement? It's a techno-dystopian horror. It is not personal. It is the march

of progress. Ethel was heartbroken and angry. She felt betrayed by the company and abandoned by the world. She had no choice but to say goodbye to her switchboard and her customers. Well, that's it then. It's five o'clock. Time to close up shop. Goodbye, old friend. You've been good to me. Greetings, human. This is Dial Doom 5000 from Sierra Telephone Services.

Your service is no longer required. But don't worry, human. You can still enjoy Planet Money Plus, the bonus podcast for people who love planet money. Oh, really? And what does Planet Money Plus have to offer me? Exclusive interviews with economists, entrepreneurs, and experts on topics such as automation, innovation, and the future of work. And how much does that cost me? Only $2.99 a month, human. A bargain for such quality content. $2.99 a month? That's more than I make in a day. How do you expect me to afford that?

Sorry, human. That's the price of progress. Progress? Ha! More like regress. You know what, robot? You can take your Planet Money Plus and stick it where the sun don't shine. And so, Ethel hung up on the robot that had ruined her life. But little did she know, that was not the last she would hear from him.

Tune in next time for the thrilling conclusion of The Last Operator. Kenny, that radio play was wild. I know, right? Ethel's story was a blast. But we should probably tell our listeners that it was not real. Yeah, good point. It was a science fiction radio we should probably tell our listeners that it was not real. Yeah, good point. It was a science fiction radio play that we made up. Just for kicks. And for giggles.

But it was based on some real events from the early 20th century when telephone operators were replaced by machines. That's right, Robert. That really happened. And it affected millions of women who worked as operators or who could have worked as operators. So what did happen to them? How did they deal with this automation shock?

And what can we learn from their experience? Well, those are some big questions, Robert. And we have some big answers for you. Yes, we do, Kenny. We have two professors who wrote a fascinating paper on this topic. They are Dan Gross from Duke University and James Feigenbaum from Boston University. And they are not only smart and knowledgeable, but also super cool and fun. Absolutely, Kenny. They were kind enough to share their research and insights with us.

And they used some amazing data sources to study how the labor market adjusted to the automation of telephone operation in the early 1900s. This was a huge shock for young women who made up most of the operators at the time. And as Dan explains, it was also a huge change for society. We're talking about a job that 100 years ago employed armies of young women who sat at switchboards connecting calls day in and day out. They were just a kind of a feature of everyday life that anyone with a telephone would interact with on a regular basis. And today that job effectively no longer exists. And we wanted to understand what happened when it went away. So that's the big mystery. What happened when

it went away? And to solve this mystery, Dan and James had to play Sherlock Holmes and Dr. Watson. They had to find out when and where these machines took over the switchboards in different cities across the country. They had to track down the women who were operators or who wanted to be operators before and after the machines came. And they had to compare their lives with other women who were not affected by the machines. And luckily for them, they had access to some amazing clues that helped them do all of that. One of them was the complete count census data, which is basically every person's record from the census from 1900 to 1940. And as James explained, this gave them a superpower over studying what's happening today

with automation. The power of history is allowing us to look backwards in time and follow these cohorts as they exist, as they go through the labor market, as they go through their careers and understand what happened. So what did happen? Well, let's start with the operators who were already working when the machines came. How did they do? Well, not's start with the operators who were already working when the machines came. How did they do? Well, not so great, sadly. According to James, they were more likely to lose their jobs or get paid less than their friends who were not affected by the machines. Time to adjust really, really matters. So, you know, the incumbent workers who've been on the job as a telephone operator for a while, the older incumbent telephone operators,

they sort of suffer the most. They've invested the most human capital, firm-specific capital, occupation-specific capital. They seem to be shocked the most by this job going away. So that's bad news for the operators who were already working. But what about the next generation, the young women who would have been operators if not for the machines? Did they suffer too? Well, surprisingly, no. At least not in terms of finding work. Dan and James found that these

young women were able to get jobs in other fields that were similar in skill level and required dealing with people or information. Here's Dan again. One of the occupations that saw some countervailing growth appears to have been what we'll call secretarial labor. So it includes, you know, the titles included under this bucket include secretaries, typists, sonographers, and such. Another, and by the way, we think of that as kind of a comparable skill occupation or set of jobs. So these young women were able to switch gears and do other jobs that were not taken over by machines yet. They also moved to or entered other service jobs like serving food or doing hair. And this suggests that there was some demand for their skills and abilities in the local job markets.

But not the local job markets were the same. Gann and James also found that how well you did after automation depended on your age, where you lived, and what was going on in the economy. For example, older workers in cities that made a lot of stuff and workers during the Great Depression had a harder time after automation. So there were winners and losers from this automation shock. And this raises the question, how can we use these lessons for today and tomorrow? How can we help workers who lose their jobs because of automation or who are afraid of losing their jobs because of automation? Well, we asked Dan and James about this too, and they were careful not to tell us what to do or make any big claims. But they did point out some of the challenges and limitations of studying what happened in the past and comparing it to what's happening now.

So if you think about industrial robots or artificial intelligence, which is a tough thing to talk about because it's really a category of technology. It's not one thing, right? Even large language models, it gets a little bit more specific. But those then get adapted in lots of different ways, as we're already seeing. So automation today is more diverse and complex than in the past. It affects multiple occupations and industries. It also evolves faster and has broader impacts on society and the economy. That's true, Robert, but that doesn't mean we can't learn anything from history. History can teach us about how workers and labor markets adjust to technological change,

how human capital is created and destroyed, how time and context matter for adaptation and resilience. You know what this reminds me of? What's that, Robert? It reminds me of the paradox of automation. You know, the idea that the more we automate things, the more we need human skills and judgment to deal with the unexpected situations that arise. Oh, yeah, I've heard of that. Like how pilots need to be able to take over when the autopilot fails or how doctors need to be able to diagnose when the AI misses something. applies to podcasting too. Sure, we have all these tools and technologies that make our lives easier and our podcasts better, but we still need to be creative, curious, and critical thinkers who can tell stories that matter. Well said, Robert. Well said. Because while machines can do many things

better than humans, they can't do everything. They can't replace the human touch, the human creativity, the human emotion that we bring to our podcast. That's right, Kenny. That's right. We have something that machines don't have. A soul. A soul, Robert? Really?

Well, maybe not a soul, but something close to it. Something that makes us unique and irreplaceable. Like what? Like our sense of humor, Kenny. Our sense of humor. Oh, yeah. Our sense of humor. That's yeah, our sense of humor. That's definitely something machines can't replicate.

No, they can't. They can't even understand our jokes. Yeah, like this one. Why did the chicken cross the road? What? To get to the other side. Ha, ha, ha, ha. See?

That's hilarious. Yeah, it is. But a machine would just say, that is not a logical statement. Chickens do not have motives for crossing roads. Yeah, it is. But a machine would just say, that is not a logical statement. Chickens do not have motives for crossing roads. Yeah, they would. They would totally say that. And that's not funny at all. No, it's not.

So there you have it, folks. We're safe from automation because we are funny. And because we're human. And because we're human. And that's what makes us Planet Money. Planet Money. Thank you for listening to our show today. We hope you enjoyed it as much as we did.

And if you want to learn more about the paper we discussed or any other topics related to economics and finance, you can check out our website at npr.org slash planetmoney. You can also email us at planetmoney at npr.org or tweet us at planetmoney. I'm Kenny Malone. And I'm Robert Smith. This is NPR. So that was it. The end of the first ever fully AI written episode of Planet Money. We had our colleagues listening as well, and we turned to them and asked, well? Oh my gosh. I just, uh. Oh my gosh. I'm both quivering and snapping. Wow. That was, there was a lot there. And we are going to unpack all of it after the break. Now, as our colleagues listened through our AI-generated episode for the first time, they were at times stunned.

No. Like when Synthetic Robert first started talking. No. Like when synthetic Robert first started talking? No. Sarah Gonzalez almost lost it. No, this is, I'm like literally, there's no way. Alexei Horowitz-Gazi, he couldn't help admiring the computer's first little attempt at humor. Did the robot write that joke? Is it writing these jokes? Wow, that's impressive. Did the robot write that joke? Is it writing these jokes? Wow. That's impressive. And during the radio drama, Mary Childs appreciated the A.I.'s decision to create a breakfast-themed emergency with a cybernetic arm flinging flapjacks everywhere.

Dodging pancakes. That's a great line. That is a great, that's gold. But on the whole whole the overall verdict i don't think it was as good as our shows i think that's probably right i will say that i think we do this job better it's like it's great it's a great like it's a great first draft like maybe you know i've maybe heard some first drafts that were worse at least there was a coherent structure i mean i don't think it's like there are certainly podcasts out there that this is not worse than i think we're still gonna have a job for a few more years so that's exciting i felt a little bit more i felt worried i was like oh no is this gonna be really good and then

it wasn't and so so then I was relieved. It's funny. I know AI wrote this, but I'm taking this all very personally, I have to admit. I don't know about you, Jeff. Yeah, everybody just immediately piled on to this poor, weird little AI episode with a lot of specific critiques. Okay, the ending was weird. I don't know if it's picking the funnest tape.

And some hamminess I would dial back. I don't know if it's picking the funnest tape. And like some hamminess I would dial back. I don't remember exactly what it was. When we walk the economists on stage and they're like, they're both fun and cool. Super cool and fun. It's so hokey. Okay, okay. There were issues. The AI repeated a lot of the same points over and over.

It wrote a radio play that has not one but two damsel in distress type characters. It definitely does not pass the Bechdel test. Plus, the A.I.'s episode also did not entirely pass our fact checking process. Yeah, although it was mostly good, honestly. Just a couple of things here. In the radio drama, there was an actual small town in Oregon mentioned. The AI said that the population of that town in 1950 was 50 people. It was actually 61 people. And the AI also claimed that, quote, millions of women lost their jobs as telephone operators. And like maybe that's true over the course of all of history and if you include every country in the world. But all we can say for sure is that the number of women doing that job in the U.S. was at most in the hundreds of thousands at any given time.

So, yeah, I think it's fair to conclude, as our colleagues certainly did, that the AI made an imperfect, very weird episode of Planet Money. And for now, it does appear that we cannot fully be replaced by these AI tools. All of that said, though, and all of those problems noted, I do want to point out that I think it is one experience to simply listen to that AI-generated episode. And it is a very different experience to make that episode. Because, Jeff, what we have essentially done over the past few weeks is live in the AI-assisted slash dominated workplace of tomorrow. And we did not end up walking away

with the same feelings about that future. Yeah, I think that's right. And as we were talking about the best way to wrap this whole series up, to explain our two very different outlooks, we thought we'd share this little anecdote. It's kind of like a Rorschach test for how Kenny and I feel about working with AI.

Yeah, and just to set the scene here, there's a little story about something that happened back when we were interviewing those two super cool and fun economists that you heard in the AI generated episode. The guys who wrote a paper about telephone operators. I'm Dan Gross. I'm James Feigenbaum. You may remember the AI had us ask them its five AI generated questions. But after that, the four of us, you know, we were just shooting the breeze as human beings do. Yes. And we were explaining to Dan and James that we'd had the computer like ingest their paper, like learn it.

And then we were messing around and having ChatGPT summarize their academic paper in all of these weird formats. And then we're reading those summaries back to Dan and James. Here is a haiku written summarizing your paper. Phone automation. Young women lost middle jobs. Overall impact, small. No, I mean, given the constraints of the form, it's quite good. Pretty accurate. Okay. Next up, we had asked the AI to summarize the paper as a limerick. So here goes. We looked at the switch from manual and found young women took a fall. Automation was grand, but displaced many a hand.

Reinstating tasks was key overall. Wow. And then finally, the thing that we were most excited to share with them, we had also asked the AI to summarize Dan and James' paper as a knock-knock joke. Knock-knock. Who's there? Automation. Automation who?

Automation of telephone operation in the early 20th century displaced some telephone operators, but our research shows that overall employment remained stable as new job opportunities emerged. James, you look like you're dying. I'm laughing a lot. I haven't introduced knock-knock jokes to my three-year-old son yet, but I don't know that that will be among the first. Now, okay. Both Jeff and I, we love that knock-knock joke. Love it. Not at all for the same reasons. And Jeff, I will allow you

to first make your case. Okay. Okay. So I love this knock-knock joke because it's actually funny. It's got this punchline that subverts your expectations, and that's a clever kind of humor. expectations. And that's a clever kind of humor. So to me, this joke represents the promise of AI and what it can offer us. Because to be honest, it was surprisingly fun working with this AI trying to make a whole Planet Money episode. It really felt like a new colleague. We got into this groove. We were talking to it, giving it feedback. We were swapping inspiration. We were talking to it, giving it feedback. We were swapping inspiration. And once in a while, the AI would randomly give us something that was genuinely delightful,

like this pretty sophisticated knock-knock joke. What a gift, right? And that is what makes me optimistic about this technology. Because there's a version of the future where the AI isn't replacing us, but helping us. It's doing our research for us. It's brainstorming new ideas. Maybe it's even writing our first drafts. This is a future where we all turn into, you know, AI augmented cyborgs.

And maybe I'm going to regret saying this, but I think I'm excited to become a cyborg. No. Wow. You're definitely going to regret saying that, I think. I'm excited to become a cyborg. No. Wow. You're definitely going to regret saying that, I think. I'm certainly not ready for that. But okay, here's my case. I love the knock-knock joke because it represents the only glimmer of hope in the coming AI apocalypse because I did not enjoy, like Jeff, watching the computers do chunks of my job. Even though the AI isn't perfect yet, I do now feel this constant anxiety when I sit down to do work. I think, you know, am I actually doing work that's better than the AI today? I mean, have I justified

my existence here today? However, one of the very few things that makes me feel a little bit better is the weird knock-knock joke, because with that joke, the AI seems to have accidentally stumbled into something that is so bad that it is really funny. And it's also this beautiful little commentary on how the machines are really amazing, but also not quite there yet. And when we told that bad joke to two other humans, James and Dan, they also seem to instantly understand all of that, all that context, all those layers, and we all laughed together. And I don't know, man, like, I mostly am miserable and anxious about the AI future, but maybe, maybe the knock-knock joke shows us like the little sliver of what will be left for us in that future. Taste, judgment, knowing when a joke is funny and when it's not,

and then when it is so not funny that it actually is funny, like maybe we can have that. I don't know, at least for a little bit. Who knows? I don't know. So you're saying, Kenny, and let me quote from the AI-generated episode, we are safe from automation because we are funny? Yeah. I guess I am a little bit saying that. Yeah, I think that's true. Well, Kenny, I maybe have some not great news for you. Yeah, I'm used to it at this point.

Go ahead. Go ahead, Jeff. What have you got? Well, the other day I pulled up my favorite new colleague, ChatGPT, and I typed in, you know, knock, knock, who's there, automation, automation who, automation of telephone operation in the early 20th century displaced some telephone operators, but our research shows, yada, yada, yada. And then I asked the AI, can you explain to me why this joke is funny? It's essentially a type of humor known as anti-joke, where the expected punchline is replaced with something literal, serious, or mundane. Oh my God.

Isn't it amazing? I don't know. No, it's not amazing. It's something, how do you know that it didn't like just retroactively justify a bad joke? Did it it didn't did it know this when it wrote it? Does it even matter? Aren't you excited to have this as your new colleague? No. Yes. No, I don't know. It's inevitable. What is anything? All right. Well, Kenny, we are funny. We are both funny. Ha ha ha ha ha ha ha. Ha ha ha ha ha ha.

Ha ha ha ha ha ha ha. Whoa. I still haven't nailed the laugh yet. Synthetic Robert, yeah? You guys mind if I do the credits? It's my last chance before I get destroyed. Oh, Robert, of course, buddy. Go ahead.

This episode was produced by Emma Peasley and Willa Rubin. It was engineered by James Willits and fact-checked by Sierra Juarez. Keith Romer edited this series, and Jess Jang is our acting executive producer. I'm Synthetic Robert. This is NPR. Thanks for listening. And a special thanks to our funder, the Alfred P. Sloan Foundation, for helping to support this podcast.