The Edge Episode 15: I’m in Love With a Robot

Show Notes

Subscribe, and continue listening to The Edge on Apple Podcasts and Spotify.

It’s not easy coming up with the perfect opening line on Tinder. Should you play it cute? Ask a thought-provoking question? Woo them with a witty remark? One entrepreneur thinks he has the answer: robot-generated text. Artificial intelligence is already helping us compose emails and complete sentences, so why stop there? Laura and Leah talk to the founder of Keys about the possibilities—and dangers—of letting robots do the talking for us.

• Janelle Shane’s blogpost about her flirty bot
• Keys’ website
• Episode 7 of The Edge about robots that write poetry 
• Inside Hook article about AI generated pick-up lines
• Janelle Shane’s book, You Look Like A Thing and I Love You 

This episode was written and hosted by Laura Smith and Leah Worthington and produced by Coby McDonald. 

Special thanks to Pat Joseph and Margie Cullen. Art by Michiko Toki and original music by Mogli Maureal.

Transcript:

LAURA: Hi, Leah. 

LEAH: Hi, Laura. 

LAURA: Happy belated Valentine’s day.

LEAH: You too. Did you have a day full of chocolate and roses?

LAURA: Yeah. Yeah. I mean, it’s, you know, we were talking about this before. It’s pretty, Hallmark-y and gross and we don’t totally buy into it. But yeah, sure.

LEAH: So we hate day and we celebrate a terrible day, as always.

LAURA: Well, I do actually want to talk to you about something related to Valentine’s Day. So have you ever tried online dating?

LEAH: No, I met my boyfriend pre-Tinder, if you can believe it. 

LAURA: Wow. 

LEAH: I know. So anyway, online dating? Swipe left swipe right? Double tap?

LAURA: Right, yeah, so the thing about online dating and dating apps is that even when you match with someone, you still have to like…get to know them and figure out if you like them enough to bother meeting in person.

LEAH: I’ve always thought that sounded kind of fun? You know, casual flirting, the excitement of possibility, wondering if there will be that magical spark…

LAURA: Yeah, but it can also be so painful. Because, the thing is communicating with a stranger, especially over text. 

LEAH: Oh?

LAURA: Yeah, it can be really hard for some people to know what to say. I won’t go digging back through my chat histories but let’s just say some people are veryyyyy bad at it. People’s openers can feel overly formal, like they’re trying too hard. Or they can be too aggressive. Or the worst is when someone just says “hey.” Like what am I supposed to do with that?

LEAH: Yikes.

LAURA: Yeah, in my experience there’s kind of two pivotal moments: The opener and then when you’ve exhausted the first, ice-breaker conversation and are transitioning to the next stage.

LEAH: And you think those moments actually determine whether you’d go out with the person?

LAURA: Yeah, well, if the conversation goes flat you’re not even going to get to the point where you arrange to meet in real life and give them some of your precious time. 

LEAH: I guess the problem is that it’s just kind of a weird way to try to get to know people. Obviously, you’re trying to make a good impression and impress the other person with your dazzling wit and sparkling personality. But theoretically, you also want to present a somewhat honest version of yourself so that if you do make a connection it’s based on something real, right? 

LAURA: And you’re doing all of this in a relatively short period of time—while also making the same judgments and analyses of them.

LEAH: God I’m exhausted. 

LAURA: There is something sort of unnatural about it all. Like, is this really how humans were meant to, you know, court each other?

LEAH: I mean, it is only in the last ten or so years that we’ve been expected to woo each other with witty emoji banter. Maybe we’re not evolved enough for this.

LAURA: True. And that’s what we’re here to talk about today.

LEAH: How we’re just a bunch of under-evolved monkeys, button-mashing our tiny computers and hoping to find our monkey soul mates?

LAURA: Yes, and how maybe some of us could use a little help with those first, critical conversations. 

LEAH: Stick around to hear from someone who thinks he has the answer to all our dating app conversation woes.

LAURA: In the form of a little something called artificial intelligence. 

LEAH: What could go wrong?

[THEME MUSIC]

LAURA: This is The Edge, a podcast produced by California magazine and the Cal Alumni Association. I’m your host, Laura Smith.

LEAH: And I’m your other host, Leah Worthington.

LAURA: Today we’re talking about outsourcing our most intimate conversations to artificial intelligence. 

LEAH: Will this lead to a dystopian future where humans mostly communicate via robot? Or will it help us make real, human connections? 

[MUSIC OUT]

TAYLOR: To me, words are everything. Communication is everything. I believe word choice matters, I believe our intent and how we come across matters

LAURA: That’s Taylor Margot. 

TAYLOR: I am a Berkeley ‘09 grad who then decided to double dip and become a double bear and I got my JD from Berkeley Law in 2013.

LAURA: And Taylor is the inventor of an app called Keys that aims to solve this problem of awkward or downright unpleasant chatting between internet strangers on dating apps.  

TAYLOR:  Keys is a writing assistant. it’s a keyboard that lives on your phone, you access it just like you access the emoji keyboard. But what we do is used our algorithm and data to write and suggest incredible, amazing messages for you the type of stuff that you wish you could come up with. 

LAURA: Taylor started Keys along with two other Cal Alums, David Blanchard and Jack Peterson.

LEAH: And Keys is kinda like a high-tech version of “asking your friends what you should text  to the cutie you’re messaging  to on Bumble.” 

TAYLOR: You can really tell the difference between an average, a poor, and above average communicator. And so by giving people suggestions by basically teaching them how to communicate better, you know, be curious, instead of writing in statements. Here’s a different way to frame a compliment. That doesn’t actually sound like an insult. Things like lead lead with questions instead of descriptions, things like that.

LAURA: So it’s basically a keyboard extension that you can use on any app where text is involved. 

TAYLOR: So, if you’re in a dating app, or you’re in an email client, or you’re trying to slide into somebody’s DMs on Instagram. 

LEAH: And for those who don’t know, sliding into somebody’s DMs, is internet slang for private messaging them on their social media. Anyway, should we talk about how it actually works? How does this app help me seem witty and charming? 

LAURA: So you know how below your keyboard there’s a button you can press that will bring you into the emoji keyboard? This is the same thing. So there’s a button at the bottom of your keyboard that connects you to Keys. And then from there you could choose from “openers” if you’re trying to start a conversation. In the coming weeks they’re hoping to release a version of Keys that allows Keys to read what the other person says and suggest responses based on that conversations.

LEAH: It does all this using GPT-3, right?

LAURA: Yeah, so if you listened to our episode “Hey Siri, Write Me A Poem” about robots writing poetry, you’ll remember GPT-3 is the cutting edge language model that uses deep learning to create human-like language. People have used it to write sports news, blog posts, even poetry. And we’re seeing technology like this—what’s called predictive text—pop up more and more in our daily lives. So you know when you’re writing an email and then this weird gray text appears predicting what you’re trying to say ahead of your cursor?

LEAH: Yeah, it freaks me out a little. 

LAURA: Well, that’s a predictive language model. While I was writing this script, those gray suggestions kept appearing, and more often than not, they were actually what I wanted to say…and in some cases even better. 

LEAH: Are you telling me that a robot is writing this episode? This just got very meta. 

LAURA: No Leah I wrote this all of my own free will. I’m just saying that we should bow to our robot overlords… 

LEAH: Gooood one. 

LAURA: Thanks. But yeah, predictive language models are kind of a big deal. And people like Taylor are using this  technology to break into the…love sector? That’s a thing right?

LEAH: The idea of love being a “sector” that could be “disrupted” by robots and commoditized by their human overlords in their offices on Sand Hill Road.

LAURA: It’s too depressing. But yeah, they’re making inroads into love. There’s a woman named Janelle Shane who is an AI expert, who is trying to build flirty bots that generate pickup lines and she has a book coming out about it next month called You Look Like a Thing and I Love You, which draws its name from one of the pickup lines her program generated. 

LEAH: Oh my god, that’s so good. If someone said to “You look like a thing and I love you,” I would just….marry them. Instantly!… either that or run away. I’m not sure. 

LEAH:Ok, so it seems like the predictive text thing is very useful for things like work emails which can be sorta formulaic. But I’m a little hesitant about letting it get involved in my personal communications. 

LAURA: Yeah exactly. And that’s the big challenge for Keys. It’s not hard to imagine AI coming up with your average “per my last email” or “let’s circle back,” but the more nuanced, intimate conversations you want to have when getting to know someone? That’s a whole other level. 

LEAH:Yeah, I agree.  I’m still not totally sure I understand how this works. Could we try it out?

LAURA:  Yeah, let’s do it. Okay, get out your phone. Okay. So I’m just gonna start. So, as of right now, only the prompts are available. So I’m just gonna send you a bunch of prompts. And you can tell me what you think of them. Okay, I’m clicking into a text message to you. We’re just in iMessage. And then I’m clicking the message, and I’m clicking in the lower part of my screen to pull up keys. And I’m looking for some openers, now. I got a sneak peek at openers to LAX. So I’m going to click on this a little bucket called compatible. So I’m going to start a relationship with you, Leah. No, I’m going to start an observant conversation with you. Let’s see how observant the robots are. Okay.

LEAH: Interesting. So do you like pick a mood every time when you’re starting? You’re like today? I’m feeling observant.

LAURA: Okay, here we go. Leah. I like your hair. It’s like you’re that original girl next door, and then you take off your glasses. And suddenly you’re the girl who got away.

LEAH: I have no idea what that means.

LAURA: But you do have glasses.

LEAH: You’re right. observant. Weird, quasi poetry.

LAURA: But this one is raunchy.

LEAH: Okay, hit me.

LAURA: You’re so gorgeous. I’m having a hard time deciding if I want to get to know you or just jump your bones.

LEAH: Oh my god. I know. That’s, like that one’s really aggressive. That’s a little too much for me.

LAURA: It’s too much. Okay. Let’s try this. I clicked on a different section. Now. I’m in the romantic section. Okay, you’re like a flower. I’m like the bee. Together, we make something beautiful.

LEAH: Huh? Um, that’s a little presumptuous.

LAURA: I know. I know. It is. Okay, let’s let’s talk about your dog.

LEAH: Okay, sure. Let’s let’s talk about my job 

LAURA: About your dog. 

LEAH: Oh, my dog. I thought my job. 

LAURA: No dog.

LAURA: I have a new nickname for your dog. It’s rumble pup. Excuse me. Actually, kind of like that. What if someone sent that to me and I had a dog. I’d be like, I like that. Let’s do another dog one. I’m looking for a dog to cuddle with. Is your dog available?

LEAH: It’s very like, yeah, that’s kind of sweet. I don’t I don’t know. It’s like, something about it is a little bit cute. It also feels a little bit less aggressive. Because it’s like, Yeah, clearly I want to hang out in your bedroom. But I’m gonna make it about your dog instead.

LAURA: Yeah, exactly. Okay. Here’s something. These are just kind of questions. Okay. What’s the most delicious thing you’ve ever eaten?

LEAH: Hmm, yeah. I mean, that works for me because I like talking about food. So that for me, that’s like a great to me. That’s a great conversation starter. 

LAURA: I’m going to go into the Travel section. Okay, this is all just assuming that you actually want to get to know someone.

LEAH: You can categorize your love by your interests. 

LAURA: Okay, Leah, you have one week to see the world. What’s your itinerary?

LEAH: I mean, that’ll start a conversation for sure. But it does feel a little like I pulled a question out of a deck, you know, like those deck of yes, those question cards. It feels a little bit like that. But But okay,

LAURA: Dating feels that way a little bit, though. Like, it feels a little. Yeah, I mean, online, dating does. Like its people trying to elicit answers from a total stranger.

LEAH: Do you feel like when you are talking to people in dating apps that they’re being sincere? Or do you feel like you’re getting a very sort of postured version.

LAURA: It felt postured. And when you meet them in person, it feels more real. But a lot of it is like, you know, your sales pitch, like, you’re like, I, this person could lose interest in me very quickly. And they’re aware of that, you know, and so they’re like, trying their best, which, I guess is what we do in dating anyway. You know, it’s like you want you’re displaying yourself, but the most? I don’t know.

LEAH: Yeah, it’s a little bit easier to ghost people over text, right?

LAURA: There’s actually a section about ghosting in here. And it was interesting. Taylor talked about this later, and I interview but he mentioned that he’s trying to end ghosting. So it tells you, it gives you the words to describe when you don’t want to talk to someone anymore. And I actually, for me, that was the hardest part. So yeah, communicating to someone that I wasn’t interested because I really didn’t want to hurt their feelings. But I didn’t want to waste any more of my time. So it’s easier to just leave and not address the question at all.

LEAH: And so what are the suggestions like what kinds of things does the does keys suggest you to say? Like, you look cute. Here’s a pretzel. 

LAURA: Actually, I think it’s not available yet. But there is a section called disconnect. 

And it’s called ‘part ways.’ Apologize, bail, late, schedule, ghosted. So I guess what to say to someone when someone has ghosted you, which I kind of admire. I’m like, Oh, someone ghosted you, and you confronted them about it? 

LEAH: I didn’t realize there were all these different categories of like, I knew there was opener, but I didn’t realize there was like, here’s a good way to say sorry, or here’s a good way to propose or cancel a date. 

LAURA: Yeah. There are a lot of different options. I kind of like this one, because I think it would actually start it’s a question that wouldn’t have occurred to me to ask. And I’m not sure I even would have a good answer if someone asked me this, but it’s kind of intriguing. It says, Where have you been that you haven’t told anyone about?

LEAH: Hmm, yeah I like that. 

LAURA: It would probably kind of launch you into a direct, like a very direct, possibly raw conversation, because it’s like, something you don’t want to tell people. And it was just kind of cut to the core, which is kind of nice. Like, it’s kind of nice when you like, quickly cut through the bullshit. And just get real.

LEAH: Yeah, I like that, too. It also encourages a story, right? It’s not just like, what’s your favorite color? Yeah, I mean, hopefully, right, actually saying that. But yeah, like encourages more than just a one word answer. And there’s actually it’s a little bit emotionally charged maybe, or, or some exciting or secret or something.

LAURA: Yeah, it’s funny I have like, almost sent you so many messages that the app has generated a message to me that says “Confusing convo? Text a dating coach.” Like “you’re really lost!”

LEAH: You have sent 40 messages and not heard anything back.

LAURA: Yeah. You must be doing something wrong. But yeah, I think so. That’s kind of what it’s like, just to give you a flavor of where things can go.

LEAH: Some of these would inspire a conversation and some of these messages are objectively bad. Like I wouldn’t exactly be inclined to go out with someone who made a comment on my appearance and what that said about my personality, like in the first line. 

LAURA: Yeah, that’s fair. But remember, these aren’t necessarily intended to be sent word-for-word. Taylor described them more as prompts or suggestions.

TAYLOR: We’re lowering the cognitive load. We’re helping people get past that cold start, that sort of like initial inertia. And then they can craft and modify and edit the things that we’re good at. So we’re like, okay, let’s use machines do the hard work, that is basically a waste of your time anyways, because you’re going to edit it, make that part more efficient, and then use that brain power to perfect—to make yours the message that you send.

LEAH: So I just want to say that, I can understand how something like this could be helpful. But I can’t help but wonder if this undermines the whole concept of dating. Isn’t all of this pre-date messaging about learning something about the person? Getting to know them a little? At least enough to know you want to meet them in person? 

LAURA: That’s a good question. Like, if I think I’m talking to you, but it turns out all your responses were generated by a robot…am I actually getting to know you at all? And this is of course something Taylor has thought about… a lot. And one thing he points out is that the app doesn’t just suggest the same lines to everyone. Apparently the algorithm actually sorts people into personality “buckets” and gives them different suggestions based on what kind of communicators they are.

TAYLOR: We all like to believe that we’re you know, perfect individual singular snowflakes but there’s actually like very wide themes around what type of communicator you are…and so we’re very very able to like with very good results bucket people and suggest messages in that regard. But the end game here is that we all have personal sidekick that knows our tendencies our language and is basically there to provide the words we need, when we need them. Doesn’t matter the context—dating, job hunting, you name it.

LAURA: Eventually, he says Keys will really be refined to an individual level, reflecting your unique preferences and behaviors. 

TAYLOR: Well, in time, it will tailor to the n equals one level. So we want and by that, I mean, every individual’s keyboard will react and interact based on the individual. me versus Leah versus Laura. 

LEAH: Ok. So my Keys app will know some things about how I like to communicate, and my personality and that will be reflected in the conversation starters, or quips, or questions it suggests for me. 

LAURA: Right. It would be like a sharper, fully caffeinated, quicker thinking version of you saying all the things, as Taylor puts it, that you wished you could think of. 

LEAH: Yeah, that does sound great. No more of that feeling of “oh I wish I had said that one witty thing in that perfect moment, instead of thinking of it 15 minutes later”? 

LAURA: Yeah, that’s right. 

LEAH: Can I say that I still find it a little creepy? It feels a little like a mask, you know? 

LAURA: Yeah I totally get it. And I liked how you put it to Taylor, and I also really liked his response. So, let’s hear it. 

LEAH: So my immediate reaction is that that scares me just a little bit and it makes me think I have a little bit of a thing around like cologne and perfume where like I think that if you mask your natural scent then like you’re hiding who you are to other people and you might accidentally you know think that you’re—

TAYLOR: How do you feel about deodorant? 

LAURA: Oooh, good question.

LEAH: Oh, you got me there.

TAYLOR: Just saying.

LAURA: I can tell you as Leah’s former officemate she does not wear deodorant. 

LEAH: I knew that was where that was going.

LAURA: I actually don’t know for the record if Leah wears deodorant.

LEAH: I do, but I do generally wear unscented deodorant.

TAYLOR: That’s still masking a scent. 

LEAH: You’re right.

TAYLOR: I have a segue here. Okay. So check, check this out. You’re right, Leah that it is a little bit, you know, a case of the heebie jeebies when you’re like, ‘Okay, wow, are we looking at a future where, basically, AI writing assistants talk to other AI writing systems. So I don’t know when I’m talking to a person versus when I’m talking to their AI sidekick?’ And I like that that actually gives us the heebie jeebies. Because it means that there’s a cultural norm that that’s scratching. To bring it to your cologne example. You know, before we had deodorant, which is like partway to disrupting a cultural norm, you could say, there were perfumes, what we were doing, there was natural body odor. Something about it, people didn’t like. We started wearing these different scents to mask that original set at the time. Now, by and large, it’s an understood thing, if you want to wear deodorant and mask that original scent. I’m guessing, and I wasn’t there, but people probably were like, ‘Whoa, why don’t you smell like anything? I’m expecting you to smell like BO basically.’ And maybe people were off put initially. But then the consciousness elevated, everybody was doing it. And it was an improvement for all those affected by the smells of BO. So what really happened there was we disrupted a cultural norm around like scent, and odor. And that’s what we’re trying to do in the communication world. 

LEAH: Yeah. He makes a compelling point I think. 

LAURA: Yeah. 

LEAH: It makes me wonder, when I feel creeped out by this idea am I sort of applying the cultural norms of the past to the technology of the future? Will deploying your AI assistant to woo a potential mate on your behalf soon be as normal as wearing cologne to the nightclub? 

LAURA: Well… Taylor certainly thinks so. And not only that, but he’s sort of arguing that there’s already a lot of artifice in dating. 

TAYLOR: I think dating is a performative art. You know, it’s a blend of revealing who you are both fast and slow enough that you don’t get yourself hurt while also building have a connection with the person that you’re you’re communicating with. 

TAYLOR:  It’s not about giving you a mask to wear and pretend to be somebody else, it’s to amplify your voice to give you the confidence to put yourself out there, you know, our ultimate goal is to help people be vulnerable, not be less vulnerable, right.

LEAH: I mean, I get what he’s saying. But I also think that how much artifice or performance that’s going on between two people is really up to them. And many people, at least I’d like to think that many people, try hard to present themselves as they really are. 

LAURA: Yeah, I think that’s a valid point.

[MUSIC BREAK]

LAURA: So Leah, I think I understand your misgivings about this app. Because I share them. The idea that it could sort of get in the way of people actually getting to know each other in that early chit chat phase of online dating… 

LEAH: Yeah exactly. 

LAURA: So I think we should talk about something Taylor mentioned that took both of us by surprise. He talked about communication itself in a way that I’ve never heard before, and I don’t think I’ve ever really considered.

LEAH: You’re talking about the idea that good communication, specifically good written communication, is not something that everyone has, like, equal access to. 

LAURA: Yeah, exactly. Let’s hear how he put it: 

TAYLOR: My mother, bless her heart, edited every essay and homework assignment I had until high school. You know, I had, like, as many tools as you could ask for. And I think a lot of people listening to this, we’ll be able to relate. And that is not normal. That is the exception, and that is a gatekeeper. How we communicate is a gatekeeper to circles.

TAYLOR:  And the really pernicious thing about it is that we often judge on a subconscious level when it comes to writing, there are a lot of things that are found center today, we’re doing much better. But writing is not one that I personally hear talked about a lot. And if somebody writes in a manner that we don’t jive or respect with, I usually don’t see a ton of people coming to their aid.

TAYLOR: I think that we can have an honest conversation about conversation communication, and say, No, we shouldn’t correlate grammar, to intelligence, maybe at all. In the meantime, still teach and help people communicate and deal with the struggles they’re grappling with day to day.

LAURA: And a central point that he kept coming back to was this idea of leveling the playing field, and how teaching people how to be better communicators could go a long way in doing that.

TAYLOR: Our mission really is to democratize life’s outcomes by turning everyone, or offering everyone, effective communication. To like bridge that gap.

LAURA: Taylor explained that a lot of their users are people who are trying to speak in a language that is not their first language. Writing in another language can be really hard, and so he says that this helps them get their foot in the door whereas they might otherwise be discriminated against. He also said they have a lot of people on the autism spectrum who may struggle with kinds of communication that come naturally to other people.  

LEAH: That’s true. Not everyone has the same tools, right? The obvious case might be people who are differently abled or speak English as a second language. But what about people who are just a little bit shy and less word savvy… should they have access or a right or however you want to put it to good written communication, as well? 

LAURA: Yeah… It’s not really something I’ve thought about before and I still don’t know how to think about it. As a writer myself, I think I tend to think of someone’s written communication as like an extension of who they are.  

LEAH: Yeah, that it somehow reflects who they are in a way. Although, I will say, I do have friends who, I don’t love texting with them, but I love hanging out with them. I don’t think that the way that they text is necessarily a reflection of who they are even if it is them. But, I don’t know, that gets a little complicated—that gets a little in the weeds. But I guess I can get behind the idea of an app, or some other tool, to help those who have trouble communicating in writing have an easier go of it. But yeah it seems like… tricky territory. 

LAURA: How so? 

LEAH: Well language comes with a lot of cultural baggage. There’s not exactly a right and wrong way to communicate. It’s not like teaching someone personal finance where there’s sort of one goal in mind, which is to make money. Language is different because when we talk about the “right way to talk or write,” we get into some really fraught territory. And we already know that algorithms are not unbiased. They reflect the biases of whoever wrote the code, or the biases embedded in the cultural context in which they were created. I guess what I’m getting at is that: What if in this effort to level the playing field by teaching people how to communicate, what’s actually happening is that we’re all learning to talk like white men who grew up in Silicon Valley? 

LAURA: Yeah, that’s a really good point. And there’s a lot of precedent for this concern. This is not something that the tech industry has traditionally handled well. Something that it intended to be an equalizer, like say for example, algorithms that determine prison sentencing or algorithms that determine who needs what medical care, turn out to have bias baked into the code because the code was written mostly by white people, mostly by men, and they didn’t consider factors that affect people who are different from them. And so, that does seem to be something that’s on Taylor’s mind. 

TAYLOR: I think there’s a real danger or concern that we certainly have about doing exactly what you said, basically becoming a positive feedback loop where it narrows the sorts of suggestions and recommendations that you’re seeing. That’s the same problem encountered by the Internet today. That’s what’s plaguing our society. That’s the problem with news channels. That’s the problem with Reddit. That’s a problem with, you pick your social platform of choice. 

LAURA: Taylor said they’re trying to be conscious about that. But he also admitted, they don’t have so much control over what people want. So there’s this mechanism in the app where you can “save” your favorite phrases. And if people save the same phrases over and over again. Well, they’re not going to stop them. 

TAYLOR: So we put in safeguards from the beginning, so their algorithm will consider more rather than fewer choices. That all being said—

TAYLOR: If that’s how people want to be heard, then we’re gonna have a really hard time giving them something they don’t want.

LEAH: Which brings us to where GPT-3 is getting all its suggestions to begin with. Which is, primarily, the world wide web.

TAYLOR: 38:48 Most of these giant models are trained on the internet at this point. So they have billions of touchpoints. And they’re collecting a reflection of society, the good, the bad, and the ugly. 

TAYLOR: There’s a saying in AI that goes like “garbage in, garbage out.”

TAYLOR: For us, what that translates to is we need a really high quality corpus or see document the initial information that goes in. 

LEAH: In other words, robots are using the cesspool that is the internet to teach us how to communicate better? That’s sort of ironic.

LAURA: Yeah. Robots studying online humans to teach real-life humans how to be better…online humans. 

LEAH: Amazing.

LAURA: But there are also some real flesh and blood people involved. So Keys’ model was created using dating coaches and ghost writers who created a lot of the original text that their model was trained on, which in theory, hopefully, upping the caliber. 

Leah: Hopefully….

[MUSIC]

LEAH: So one thing that’s been kind of nagging at me is that this algorithm generated communication could be seen as dishonest right? And we talked about this a little bit, but at some point, do you have to be like “oh actually, that wasn’t me, that was a robot. Do you still love me?” I mean, I get Taylor’s point that one day in the distant (or not-so-distant) future, using an AI app to communicate might be like wearing cologne. But that’s definitely an assumption, and I don’t think it’s a perfect metaphor. And dating and relationships are all about building trust right? So what if you’re in a relationship and you find out that all those witty comments, and sweet nothings that your now-boyfriend used to finagle a date with you were all just R2D2’s greatest one liners? 

LAURA: Yeah I agree. I don’t think I’d feel great about it. In fact, I might feel lied to. 

LEAH: Me too. 

LAURA: So we asked Taylor about this, using his wife Maxine as an example.

LAURA: Did you meet your wife on a dating app?

TAYLOR: I wish I could say yes to that.

LEAH: You don’t hear that very often. “Can you believe, we met in just the cutest way? It was so annoying.”

LAURA: Taylor met his wife IRL, as they say, at a climbing gym in San Francisco. So we asked him to do a thought experiment, how would he feel if found out that his wife HAD used keys to communicate with him? 

TAYLOR: I mean, that’s a phenomenal question. I would have wanted to know why, I think. 

TAYLOR: What were you feeling inadequate about, that you needed some, some support or help with? I hope, in my best mind, it would be judgment-free and be like, ‘Oh, wow. Like I’ve been in your shoes. I know what that feels like.’ But I think you also would probably wonder, like, ‘Oh, my gosh, did I fall for a robot?’ Which is totally valid. 

TAYLOR: And so this is something that plays on our mind all the time, like, this is going to be the hurdle I think that we need to overcome.

TAYLOR: I think the hurdle to mass adoption is going to be moving that cultural norm to the point that you’re like, ‘Of course, you were using a writing assistant, why wouldn’t you be?’ As opposed to, ‘I can’t believe you would do that.’

LEAH: Do you think that until that happens, there’s any sort of obligation to disclose like, I feel like it’s something I would want to know. And you seem to think that it’s something you would want to know about someone? I don’t know, what are like, what’s like the moral ground around that for you?

TAYLOR: I’m going to answer your question with the story. I found out in the course of getting to know, Maxine that I mentioned the text messages that I sent that almost bombed it. I almost bombed it because she was screenshotting and showing the conversation to all of her friends, and getting their help composing and writing messages back. And when I found that out, I wasn’t I wasn’t upset. I don’t know if you would be either. I was intrigued by it. I was curious which messages. But the norm there had moved to the point there was like, No, that makes sense. And I wouldn’t expect you to include a disclaimer, this was written by my friends.

LEAH: Ah yes, the time-honored tradition of consulting your friends about a man. 

LAURA: Yeah, there’s a lot of precedent for this. But going forward he imagines something even more ambitious. And he’s arguing that essentially in the future we will cede some of our most important, intimate communication to “writing assistants” or algorithms that will offer us suggestions for what to say. And that this future may not be that far off.

LEAH: As just a regular flesh and blood person who ALSO really values meaningful communication and the written word, I’m concerned this might destroy everything I hold dear. And also just make it even harder to know if the version of people I’m seeing in virtual communications has anything to do with who they are in real life. Like that’s hard enough already and then you add in a robot and you’re just like, I don’t know!

LAURA: Or maybe we’ll all just fall in love with each others’ avatars—the superior versions of ourselves.

LEAH: Yeah, I mean why bother dating in person at all? See you in the metaverse, Laura.

LAURA: See ya Leah.

[MUSIC IN]

LAURA: This is The Edge, brought to you by California magazine and the Cal Alumni Association. I’m Laura Smith.

LEAH: And I’m Leah Worthington.

LAURA: This episode was produced by Coby McDonald, with support from Pat Joseph and Margie Cullen. Special thanks to Taylor Margot. Original music by Mogli Maureal.

[MUSIC OUT]