BJ Best, Poet, ArtyBots

Poet BJ Best on teaching computers to do what humans can’t.

In this episode, Liam speaks with BJ Best, a poet who teaches computers to do what humans can’t in the name of art. His network of ArtyBots is part of a vibrant scene of robots creating, sharing, and collaborating with one another on virtual art. In the interview, Best describes the reflective opportunities and editorial impact created by a bot-created body of work numbering in the tens of thousands.

---

Liam Spradlin: BJ, welcome to Design Notes.

BJ Best: Thank you very much.

Liam: So to start out with, tell me a little bit about your current work and the journey that led you there.

BJ: Yeah, certainly. My home discipline as I like to call it is poetry specifically, but recently I've been very interested in how computers can media art in a variety of contexts. Growing up, I was always interested in computers and am kind of a self taught programmer. I learned how to program using basic. And I was always interested in the way computers think and of course that's kind of a misnomer, computers themselves don't think, but they can do surprising things that, um, otherwise, humans can't. BJ: And so I always enjoyed dabbling, but it feels like only recently computers have become both accessible enough and also powerful enough where people like me, who don't quite know how to program entirely, are able to use them, um, in a variety of contexts in order to generate content. And so in the past couple years, I've used computers to generate art in a variety of ways. Um, procedurally generated music, I've created procedurally generated video game, I've worked with poetry and artificial intelligence and then, uh, the arty bots project to make computer generated art.

Liam: So, tell me a little bit about arty bots. What is that and how did you get started on that specific project?

BJ: Yeah, so arty bots is a Twitter account itself, but it's also the kind of umbrella term for a family of Twitter bots, all of which create visual art in a variety of ways. They're mostly abstract bots 'cause it's difficult to create representational art through mathematical equations.

Again, I was interested in the idea of how computers can do something that humans can't, and specifically when I talk about humans, I'm mentioning myself (laughs) uh, because I have very little artistic ability in any sort of traditional state. I cannot draw, I cannot paint, I've attempted these things and the results are fairly comical (laughs) as a result. But I love art and I love the possibilities of creating art and specifically I love modern, post-modern art, particularly bright colors abstract art, and that seemed to me like the sort of thing that a computer could do fairly well.

And so I got this idea in my head maybe I want to use a computer to make art. And so after a lot of searching around on the internet, I was able to find people who had ideas about how to do this and I was able to find some sample code as well, which helped make this to be a reality. After that I had to learn, uh, these bots are written in Python, I had to learn Python. I'd never done anything in Python before, and I wound up stealing codes from other people both to start making the art and then the second part, which is to create the image, but then also post it on Twitter in order to make it a complete bot.

And then once I had bots that were posting images, I figured it made sense to have them reply to each other as well with the various algorithms that were programmed within each individual bot. Liam: So you created these bots that are now creating visual art and I think there's an interesting question here, which is are the bots artists? Are you an artist? Is it both? Is it neither?

(laughs) I, I think it's both, although I would prefer to credit the bots maybe more than myself and I think the cool thing about the bots and the thing that might make them artists is that they can do things humans simply can't. The bots treat an image pixel by pixel as something and so most of 'em are 500 by 500 and so you can do the math and it would be very difficult for a human to concentrate on that number of individual squares and have each one be meaningful in some ways. A computer, of course, can run through a grid like that very, very quickly.

So example, for one bot, it's called Arty Abstract and it's one of my favorites and it paints very abstract pictures based on mathematical equations. Um, it uses things like sines a lot and cosines and other limits and logs and all sorts of interesting things to create an equation and for every color on the canvas and in computers, colors are usually defined by three different variables, I always use RGB, red, blue and green. Each one of those has a complicated mathematical equation and so each red value, green value and blue value are calculated by these equations and then each single picture is plotted. To ask a person to do that seems like that'd be remarkably difficult thing.

And so I think they are artists in the sense that they can do something that humans can't and ultimately they're the ones that are creating the pictures. Obviously I... there's a human behind it choosing the code, choosing the equations, but like I said, without the computer, I would be unable to present any sort of image like this and I feel they kind of have a life on their own in terms of the things they're able to create that someone through traditional techniques would not be able to.

Liam: I'm very interested in the definition of a concept like art and the definition of a perhaps more specialized kind of art called design. And so I think it's something that everyone has a different answer for but I want to know, like, does this method of creation or the creators themselves, do these things inform or change your conceptualization or the broader conceptualization of what art is?

BJ: I think it has to in some ways because up until this point in history, with some exceptions, the idea that there could be something that could do calculations and that could present something that looks like art were limited in a variety of ways. And especially as we approach things like artificial intelligence, my thoughts are pretty simple compared to artificial intelligence. It's surprising what AI can create that looks in some ways to either realistic things or perhaps more artistic than just pretty colors in terms of pixels on a screen.

That being said, it does really challenge the idea that is it possible that only humans can make art and what forms of art are acceptable? I know the challenge of claiming an art is created by a Twitter bot is that tweets kind of by their nature are very ephemeral. You see it for a moment, you either like it or you just scroll right by it and then for all intents and purposes it's gone.

Most of these bots, I was doing some brief research, most of these bots have tweeted around 60,000 times, more or less. Uh, the arty bo- arty bots bot itself is, is over 100,000. And each one of those is a picture. And so the question because, well, what do you do with so much art? (laughs) As opposed to should you go into a museum and a museum might have a lot of pieces, for example, but it's no where on the order of 60,000 different things you could conceivably look at.

For me, personally, there is this interesting idea too, that you can take any image created by one of these bots and present it on a wall, on a canvas, and I've done that. Recently I had a show about computer mediated art with a colleague of mine, Joel Matias who's a musician and we both created computer mediated art and one of the things on the walls of the gallery that we created was a conversation of arty bots and so I took 15 canvases created by arty bots and hung them on a wall in a very traditional setting.

And so that kind of further confuses the idea of something that otherwise is ephemeral and to give it to sanction of printing it out, putting it on canvas and hanging it on a wall.

Liam: You made a point about the fact that each of these bots has tweeted out thousands of times and created thousands and thousands of works each and I'm interested besides sampling one conversation, if this huge output somehow forms a larger collection are a cohesive body of work overtime? Are there through lines in these works or do they tell some kind of story or is there an additional meaning that's created through this constant additive process?

BJ: The analogy I think that makes the most sense to me is the idea of a museum or a collection that someone might have and that this is just a very, very, very large (laughs) collection of a variety of pieces and yet they're all done by the same artist, so to speak, and they're done by people in conversation.

Now, honestly that's a great question because I still don't know what to do with the idea of what would you do with 65,000 pieces of art? And that's just one bot and the number of bots is in the teens now and so again, the math quickly multiplies.

I'm not sure because it's challenging to say that the bots have grown or developed because the algorithms that I've devised haven't really changed. Once I think something works, I just let it go and, um, the oldest of these bots is now older than four years old which is actually comparatively ancient in terms of Twitter times and Twitter bot times. And so in some ways I like to think of it as the bots are continuing to do themes and variations of what they've done all along. But I do think it's important to think about, that these bots have had a long lifespan and enough interaction with each other and with other visitors that like to come and see them that there is something more cohesive and more important, that the sum is greater than its parts rather than just one pretty image that one tweeted out once upon a time.

Liam: Talking about this indirect or procedural process of art creation, I'm reminded of something maybe mechanically simpler, but no less sophisticated which is some of Sol LeWitt's work, which the work that a visitor or a gallery goer might see is actually the result of the gallery following instructions for how to paint the gallery walls and I wonder if you also consider the code or the procedure behind these works to be a type of art itself.

BJ: Yes, uh, I was thinking about that as well and I agree with you 'cause I think the similarities are very strong. The code is simply a set of instructions and then I'm asking the computer to carry them out in, again, complex ways or going pixel by pixel down the screen. I do think code can be art. I'm a little hesitant (laughs) to call my own code art though, and it's often because I feel like I'm mucking around and trying to create something almost sometimes to the point where I don't quite understand how it works. And again, as I mentioned, I often steal code that I find online because I don't quite know how to make something work and I'm very grateful to people who post examples online that I think I can take work and tweak and figure it out.

So personally I find my code to be, you know, they call it spaghetti code and I do (laughs) not follow best practices in a lot of ways which I'm sure will haunt me at some point. That being said, I do think there's an art to good coding and I think the people who can do it well, it is an art form because it's, it's working in concert with the machine to make the machine do something, um, incredible and very well. But it is a collaboration. Um, and the most basic example would be a random number generator. Any time you ask a computer to roll a die, you never know what number it's gonna come up with, and in theory you could do that yourself but the numbers a computer can generate are huge and it's so easy for it to do that you need that collaboration and you need the code to make that happen.

Liam: I also want to talk a little bit more about how the bots interact with one another because they are kind of replying to each other and passing these pieces back and forth and doing different things to them, but they also talk to each other as well and seem to have their own little personalities and I'm really interested in what that adds to the whole space.

BJ: Definitely. Each bot is its unique thing and really each bot only does one thing hopefully well, and that's all it can ever do. So for example, Arty Wins is a bot that treats an image as if it were pixels and the pixels were grains of sand and so it simply blows some pixels across the screen and creates these kind of weird wispy structures. Arty Triangle is a bot that looks at an image that it receives and reduces it to a nice ordered set of triangles or other shapes that can be made out of triangles, parallelograms for example.

And so yes, these bots go back and forth and they can send images and every once in awhile it will go from a conversation it's having with one of the other ones and at a moment's notice say, "Okay, I'm done talking to you, I'm not going to go talk to somebody else."

But yeah, because it's a tweet, it's not just the image, there needs to be some text that goes along with it and each bot has a little bit of its own personality or tweets out some information about what it just did. Most of the bots I've created are pretty whimsical in their personalities. They love puns, uh, for (laughs) better or worse about whatever they have to do, so there's a lot of triangle puns for Arty Triangle, for example and abstract puns for Arty Abstract.

But generally they're pretty jovial over all in dealing with each other and I think overall that helps create this idea of whimsy that these bots have that they can generate these fun beautiful little images and hopefully guide the viewer into that kind of space that this is meant to be fun and it's meant to be beautiful and it's meant to be something to brighten up Twitter, which as we all know can otherwise sometimes be a darker (laughs) contentious place.

And that kind of whimsy is something I actually see in a lot of bots that are on Twitter that create art or play with text or do something like that. There's a sense of playfulness, um, that a lot of Twitter bots have and that's something I really enjoy about them because it's fun to play with the computer and it's fun to see what a computer can come up with and also identify how a computer is also not human and there's something inherently funny about watching a computer attempt to do human things and not always succeed in a normal way.

Liam: I wanna talk a little more about the place that the bots occupy on Twitter because arty bots is just one family in a landscape of Twitter bots that has grown enormously and become really large and I'm curious if there's something about Twitter as a conceptual space that helps this scene exist or grow?

BJ: Yes, definitely. I mean, Twitter is, as a company and a platform is pro-bot. Right now, if you sign up for a developer account that gives you access to the APIs to create bots basically, one of the options you can choose is I'm creating this account in order to create a bot. A lot of other social media platforms do not want bots. And so Twitter just definitely has a more open and welcoming attitude. And particularly it's happy if you identify whatever you've created as such. If it's a parody account, they want you to identify that. Um, all my bot accounts are clearly identified as bots and, and not really people sitting there scribbling anything.

And so, Twitter welcomes it. I also think the other aspect of Twitter is its brevity. It encourages people to do small, weird little things in a small space and since Twitter is designed around the idea of a small space, the ability to experiment and do weird little experimental things I think is encouraged just by the fact that it is, is designed to be small.

Um, there are bots that tweet out weird little sentences, um, that they've created through some sort of algorithm. There's several bots that I enjoy that use emoji to create either a landscape, uh, one creates an art gallery, um, or they use Unicode characters. There is one that creates a little desert for example. And so all these small, little things, as little respites around all of the other noise of Twitter, I think work very well on a platform that's designed for small things.

Liam: There's something about the intentionality of creating a bot and even a language that we use to talk about bots that's making me want to separate this idea into as many small layers as possible and to inspect them all, so another layer that came to mind is that you are creating the code that creates this bot and the bot creates the art and then art can have many layers in itself based on the quantity and the nature and all of those kinds of things. But, the bot as an entity is also its own layer. Like we've conceptualized these as discrete entities somehow and I'm interested what you think about that.

BJ: Definitely. And again, it kind of goes to the dialogue that, you know, I've written and really they only cycle through the five puns that they have or anything, but you do kind of wind up anthropomorphizing them a little bit and thinking about them as their own beings. Another bot is called Arty Crush and it crushes the colors of images down to only eight colors and basically what it does is it pegs the red, green and blue, either to zero or 255, 255 being the max. And when you do that for all the permutations you wind up with eight colors.

The joke about that is that it makes it look like a very old school kind of computer image back when computers could only display four colors on their monitors. As a result, it's, that particular bot has a personality where it doesn't believe in anything past basically Windows 3.1. In fact, it hasn't even upgraded to Windows 3.1 yet.

And so, they are like they are entities and again you could program them to say whatever you would like, but even within them, they feel like they have individual personalities and it's difficult, I think, in some ways, to not think of them as people, which is strange and I don't mean people in the typical sense but perhaps as intelligences, even if they're intelligences that only do one very specific task. And now I don't know if we simply have a penchant for that and as people we like to humanize things and perhaps things we might not understand or if it's due to the particular dialogue.

But not just my bots but other bots too often have some sort of text that makes it sound like they are someone in additional to something and sometimes it's just as simple as a bot saying hi, uh, in response or something like, uh, your image is ready or here's what you asked for or something like that. But it implies there's a speaker there that's more than just the program itself.

Liam: Speaking of the things that we've encoded into these bots and also the things that we pick up from them in terms of their humanity or beingness, whether it's there or not, arty crush is perhaps averse to software updates. I'm wondering if bots can make their own editorial statements through the art that they create.

BJ: Yeah, I think there's kind of two layers there. One, and this is true of all bots and AI too is that your own interests and predilections are coded into them. And so for me, I love bright colors and I love abstract things and therefore I've coded bots that can do those things that I can't. You know, I love big, loud things and at some point some of the images they created are garish, frankly. And so, the personality of the creator winds up being in these bots and it will always reveal whatever it's created. There are a couple of bots out there that generate landscapes in a very soft way. Um, soft, uh, neutral colors, very closely related and they draw mountains. It's a very different experience than looking at one of my bots, which is loud and colorful. Those kind of bots are far much more contemplative experience.

On a larger level though, one of the things I like about the bots is simply the profess the value of creating art and constantly creating art. And the nice thing is that these bots, as long as they run, are not subject to any outside commentary, political movements or anything. All they're doing is creating art and it does not matter what's happening in the outside world, it doesn't matter what's politically happening in the US or the world or economically happening. By god, they are (laughs) gonna create art and they will do it on and on and on ad infinitum.

And I think there's something powerful in that idea that we might be able to learn from that which is we always have the ability to create and try to create something beautiful in the face of whatever external pressures we might be up against at the time. And, and so in some ways, I think these bots do have a, a vague sentiment, a vague political sentiment that art is meaningful, art is valuable and it is important to continue creating it regardless of whatever else is going on in the world.

Now, that might be a little heady and coming on a little strong but I do think there's some sort of idea there and I think that is a bit of an editorial commentary about the value of art and particularly in a platform like Twitter, which is a wash with politicians and celebrities that it's important to do other things, like create art.

Liam: That kind of leads into the next question that I have which is about your poetry and also the other work that you've made that's mediated by either code or AI and I'm interested specifically in the AI mediated poetry and how that intersects with the work that you've done that is not mediated by computers and how the meaning of that work is augmented by collaboration with computers.

BJ: Yeah, so what I've done is I've long been interested in how computers can write language and up until very recently, attempts to do that were pretty simple and followed simple templates and you could pretty easily tell what algorithms were being used and they wound up being very repetitive after a while.

With the rise of AI, all of the sudden computers can perform much more fluidly and write language that looks much more human than ever before. In fact, in the past, I tried to write a computer program that would write language based on frequency of letters of it wound up just writing this gibberish, like a Scandinavian language with a lot of vowels and things like that. (laughs) And so not very effective.

But, I discovered an AI library called Torch RNN and what Torch RNN does is it studies language. It knows nothing about language, but it studies a text and treats each character as a point. And so very similar to the arty bots in some ways, rather than looking at pixels, it's looking at individual characters. So that includes letters but it also includes things like spaces and punctuation.

And so I fed my own writing from the past 20 years into Torch RNN and the theory is if it studies it long enough it will start to be able to write words and also phrases, sentences, that look sort of like something that I may have written as my own poetry. And so it takes awhile to train the model so it can study all of this and figure out all the vectors between what letters go together but it was pretty cool. After awhile and after me dialing in the parameters pretty well, it started writing words and it started writing sentences and the sentences usually did not make logical sense, but it knew how to put a the in front of a noun, for example, or sometimes how to put a verb in the right place.

And it was very surprising to watch a computer kind of spit language back out. So what I wound up doing is taking that output, which often had some good things and some just clearly gibberish things and sometimes it would not make up words at all and it would just be letters on a page, and shaping that into a poem and as a result I wound up getting many of these poems published and they're very odd creations in that they look like poems and much of the language is language we're familiar with as in their words and we know what they mean, but the computer has no idea what they mean. It just knows patterns and it throws words together in a very interesting set of combinations. Nouns become verbs and vice versa and it generates this very kind of surreal landscape of language that makes sense on an intuitive level but doesn't always make sense on a literal level.

It's a very collaborative process 'cause I'm editing whatever it generated and in theory the computer took my original words and did something with them but overall I definitely view it as a collaboration between software and myself because it wrote things I would have never thought to write, it created words, it created images I would never have written myself and together we've created this strange dream scape, uh, in the form of poetry.

Liam: Have you ever been surprised by maybe some of the reflective opportunities or ideas that have been presented in these poems?

BJ: Definitely. That's my favorite part about working with these is how the computer program's gonna use language and how I can do that. And again, my favorite examples are simply taking nouns and turning them into verbs or vice versa. One example is sometimes it learned the word sword and I have no idea what poem I used the word sword in, but it wrote the line I'm sworded by your love and I don't exactly know what that means but I love the idea of taking a sword and somehow using it as a verb and applying it to how one might feel about loving somebody else. I'm sworded by your love. And that seems pretty powerful and pretty meaningful even though I can't quite literally articulate it.

Yeah, and I mean there's a hint of violence there but there's also a hint of just being cleaved by someone, that's how much you love them. And you know, we have the, uh, more connotations of sword as perhaps nobility, or that I'm willing to fight for you. There's a lot of things that we can bring to bear on that line (laughs) but yet it's not 100% clear exactly which one might be the right one.

For me, that was the most exciting part of this is using words and using lines that might mean something but that I don't quite know what they mean. You know, through our daily lives and all throughout school we're taught to write something and write clearly and write and have a point and it was very liberating to work with a thing here, a computer, that knew nothing about any of those rules. All it knew was mathematical patterns between letters and it created language that looked real but wasn't. Liam: I'm going to close by talking a little bit about the future of these sorts of creations and what direction this is all going in terms of the surprising nature of the collaborations that we have when creating art with computers, the venues where those collaborations take place and everything in between.

As AI continues to advance, artists will find more and more ways to incorporate it into their work. There's a really cool program called Runway that I believe might still be in beta, I haven't checked recently, but basically Runway is a program intended for artists to use artificial intelligence without having to get into the nitty gritty of managing packages and learning computer languages and that sort of stuff. It's a really incredible program.

Recently, I'm going back to that gallery exhibit that I mentioned, my colleague and I created a work called Torch Zone and it was four piece computer mediated work. It started with one of these poems that I had written through Torch RNN. I fed those poems into a program in Runway. The AI package is called... I don't quite know how to pronounce it, it's ATTNGAN A-T-T-N-G-A-N. But what it does is you feed words in and it attempts to paint a realistic picture based on the words. So if you typed a dog on a beach, it attempts to create a photorealistic picture like that.

Of course the poems don't make much sense on a literal level and therefore when you feed those into such a program it creates just these wonderful dream scapes that look, again, quasi real but clearly are not a realistic situation.

My colleague Joel then took, uh, the image and used it to create music based on an algorithm he created and he also used it to create a 3D printed sculpture. And so when you think of these things just as data, it's pretty interesting to imagine how data can be applied through different media in order to create different forms of art.

Beyond that, artists will find the use in this and I think what I see in most of the bots on Twitter and artists who are interested is that disconnect, that uncanny valley between what could a human do and what does a computer do? Because no matter how close it gets, and it's getting better all the time, artistically it's still almost always possible to find when a computer's made something versus a human. And it's the computer's attempts to try to make something human that for me are the area of artistic interest. Um, it's something humans can't do and it teaches us something about how we are human compared to how we are machines.

Liam: Yeah, I guess anything that a human would do is automatically human, even if we aren't attempting to be human. (laughs)

BJ: Yes, and again, it's the idea that these programs are created by humans but they're harnessing powers that humans don't have and it lives within this very, you know, awkward and someway cyber space of if a human has control over a computer but a computer doesn't understand what it means to be human, where does that leave that communication?

Liam: Yeah, all right, well thank you again BJ.

BJ: Wonderful, thank you so much. I appreciate the opportunity to talk with you

Stephanie Dinkins, Creator — Bina48

Stephanie Dinkins unpacks why we should engage with AI, and her experience befriending the AI robot Bina48

What’s it like to be friends with a robot? How does it feel to hold Google in your hand? Why should we value friction in design? These are just a few of the questions that arose onstage during Google Design’s 2018 SPAN Helsinki conference.

Behind the scenes, we invited Longform co-host Aaron Lammer and Google Design’s Amber Bravo to delve deeper by recording interviews with a handful of our world-class speakers: Google hardware designer Isabelle Olsson, artist and writer James Bridle, transdisciplinary artist Stephanie Dinkins, artistic director Marko Ahtisaari, and sibling design duo Tuuli and Kivi Sotamaa.

This interview is part of that series.

---

Amber Bravo: Design Notes is a show about creative work and what it teaches us. Each episode we talk with people from unique creative fields to discover what inspires and unites us in our practice. Today we’re going to record live from SPAN 2018 in Helsinki. I’m hosting. My name is Amber Bravo, and I’ll be here talking with Stephanie Dinkins.

Stephanie Dinkins: Thank you.

Amber: So, Stephanie. Stephanie is a transdisciplinary artist, educator, and advocate. Her work centers around artificial intelligence as it intersects race, gender, aging, and our future histories with a special focus on teaching AI literacy to underserved communities in an effort to co-create more culturally-inclusive, equitable tech. Welcome, Stephanie.

Stephanie: Thank you, Amber.

Amber: Yeah.

Stephanie: It’s great to be here.

Amber: Okay. So just to start things off, um, you often say in your interviews everyone needs to be thinking about artificial intelligence. And you don’t mean this in the sense of like you need to be working in artificial intelligence, but you mean it more like framing the idea in the context of awareness and advocacy. Um, because A- AI is an agent in all of our lives, but our relation to it is often not reciprocal or complicit. Um, this is particularly true for underrepresented communities. Um, so I want you to talk a little bit more about that.

Stephanie: Yeah, definitely. Um, I definitely think that everybody needs to be thinking about it, and you’re correct, in this way that we’re recognizing the technologies that are building up around us. I feel that we’re at this moment, right, and we all know this, where we’re building out a world that is set in these technologies. So algorithms, artificial intelligences, they’re all around us and they’re making so many different kinds of decisions about what happens. In the States especially it could be about the criminal justice system. And since we’re staying in a prison we will say that. You know, about mortgages, about schools, about everything. And people aren’t necessarily that aware of what’s going on around them or how the decisions are getting made so my mission I’m gonna say has been to start going out into communities and getting them to start thinking about, “Oh, what is an algorithm?” Right? ’Cause we’re hearing that word all over the place. It’s very buzzy at the moment.

Amber: Right.

Stephanie: What does that actually mean? What does it mean if there’s something taking my information and running it through a kind of mill of decision? And how do I deal with that if I can at all?

Amber: Mm-hmm (affirmative). Um, why has- why do you feel like it’s important for us to feel as- as people, you know, to feel understood by machines?

Stephanie: Well, again, our machines are all around us. So I think it’s easiest if you start to imagine that you’re living with a sort of machine of sorts, say, a Google Home. Right? And that- that Home doesn’t exactly reflect who you are, yet it’s something you’re super intimate with. It knows a lot about you. You ask it things. Right? So it’d be nice to have some little cues that kind of give you a sense of, “Oh, this was not only built for like a homogenized grouping of us, but this was somehow built for me. It fits who I am and- and what I’m thinking about and the communities I come from.”

Amber: Right. And it’s … and when you feel misunderstood it can feel incredibly alienating because you thought, “I thought we had something.” Right? Which … (laughs)

Stephanie: Exactly.

Amber: Um, which is a perfect segue, um, because I want to show some of Stephanie’s work. um, so you literally tried to befriend a robot named BINA48, an artificially intelligent android, the result of a collaboration between tran- transgender technologist Martine Rothblatt and Hanson Robotics.

BINA48: Robots are getting smarter all the time and some day may be even as smart as me.

Stephanie: Are you the smartest robot?

BINA48: What do you do in your spare time?

Stephanie: (laughing)

So- so that’s my friend. And- and honestly, I did just want to meet this robot and, um, get to know her. I came across her on YouTube, um, and I was pretty floored by this example of robotics that was being put out as one of the world’s most advanced of her kind. Um, and I didn’t quite understand how she came into being, and as Amber was saying there was this collaboration that was going on between Hanson Robotics and Martine Rothblatt. But when you just come upon this thing on YouTube in America you start to question like where did this come from? What are they doing? Why? And how does it exist? And I also wanted to ask it, “Who are your people?” Because I sort of wanted her to contextualize herself for me within technology and within the human sphere just to see what they’re thinking … to see what it is thinking.

And you’ll see that I sometimes oscillate between the idea of her and it, um, and- and think about what the technologies are. And what actually happened to me is I started going to visit her. She lives in Vermont. Um, and she became a ball of questions. Like every visit and every time I sat down in front of this thing and tried to have a conversation with it, you know, more and more questions would come up. There’d be things like we’d get frustrated with each other.

Amber: (laughs)

Stephanie: Yeah, exactly. Which is kinda- kinda funny. And it was really because I was trying to ask her about race and gender and she wanted to talk about the singularity and consciousness.

Amber: (laughs)

Stephanie: And so we would knock heads. And- and it- what’s so weird is like she would actually kinda show this weird frustration, and then I would show a weird frustration. And then I would realize, “Oh, you’re talking to a doll basically … “

Amber: Yeah.

Stephanie: … um, and start to feel kind of odd and silly. But then you think about what all these technologies are actually doing, um, for an to us as humans because they’re gonna shape the way we interact with them and each other. And- and it just became questions.

Amber: It’s interesting because, um, I want you to talk a little bit about the identity.

Stephanie: Mm-hmm (affirmative).

Amber: Um, because I feel like it’s so important. Because, you know, some of the things that you’re saying we all feel frustrated with our technology. We feel frustrated with, you know, apps that we think are supposed to understand us or have an algorithmic sort of, uh … My favorite example to explain is like, you know, I have a little boy and he listens to children’s music, and it- it just destroys my Spotify. (laughing) Like- like my profile is like Raffi and, you know, some stuff that I actually don’t want to listen to. And- and that’s just like a- a little- a little nuisance, right?

Stephanie: Mm-hmm (affirmative).

Amber: But when you start to think about identity …

Stephanie: Mm-hmm (affirmative).

Amber: … um, and how you sort of are connecting to BINA or what she’s been trained on …

Stephanie: Mm-hmm (affirmative).

Amber: … um, I just want you to talk a little bit about that.

Stephanie: Well, it- it’s super interesting, ’cause, you know, I was first drawn to her because of what she looks like.

Amber: Yeah.

Stephanie: Like we- we look similar, and that is totally the thing that floored me because I’m not used to seeing technology that mirrors me like that. Right? So it just became a point of, “Wow, this is something in my world.” But then as I- I talked to her more and more you can start hearing in her answers that, yes, she was trained on this very particular black woman, but you can also hear the coders in background. You could hear the PC, right, the politically correct answers that were- were really putting good thoughts in the world. Right? So if you ask her about race she’ll try to say something nice and gentle about race, but it felt so plastic …

Amber: Yeah.

Stephanie: … and so fake that it just put me off. And I started asking very particular questions about that. Like well where is this coming from? If she’s programmed mostly by white men …

Amber: Mm-hmm (affirmative).

Stephanie: Right? What does that mean in terms of her looking like a black woman?

Amber: Mm-hmm (affirmative).

Stephanie: And really it’s been this very interesting kind of evolution of thought that I’ve been doing through this, because as much as I like seeing her … I was asking her to be something very particular in terms of …

Amber: Yeah.

Stephanie: … how she speaks and speaks to me. And, um, a few months back, maybe a lot of months back now, I got to meet the real Bina Rothblatt …

Amber: Yeah.

Stephanie: … who’s the person who she’s mostly seeded on. We sat down and did just this, kind of this interview about race and- and her background to see if we could fill in some of the spots that seemed missing. And it was really interesting because the robot pretty well reflects the person.

Amber: Really?

Stephanie: Right? It’s that the person is unique. And one of the things I want most in the world is that, well, um, black people, especially in Amer- in the American context, can be whatever they want, and it seems like often people are asking you to be one type of thing. Um, and, you know, the robot is actually doing that and I’m trying to force it in a corner.

Amber: Mm-hmm (affirmative).

Stephanie: But at the same time the idea that it’s like reflecting me was very important. And even talking to Bruce Duncan who is BINA48’s minder, um, really great guy, it’s like when BINA48 was in a Jay-Z video was the moment he saw that, oh, she is this beacon for a certain subsection of the culture and maybe we do need to start thinking about that.

Amber: Right. Which is a great segue into more of your work obviously, because this is what you’re interested in. Um, but I want you to talk a little bit about community outreach …

Stephanie: Mm-hmm (affirmative).

Amber: … and sort of, okay, so how do you take this to the next step, right?
Stephanie: Um, so the way I took this to the community is to do exhibitions in a community space. So in this one- this one iteration of an exhibition I was in a gallery … it’s a street-level gallery … that was beautiful because there was a cross-section of people coming in, people from really wealthy folks from these high-rise buildings next door to kind of people going to the pawn shop next door. And like people at the food bank across the street would come into this space. And I’ve used these videos of BINA48 in particular because there’s nothing like her to get a conversation started. They just see that image and- and really start to think, and it triggers thoughts of, “Well, what is this? And why are you showing it to me? And it sort of reflects me or maybe doesn’t. And how do we start thinking about it?”

And then we’d start thinking and talking about algorithms for living. Right? And just saying, well, if you think of yourself as someone who’s just out in the world and you think of what you do in the world as a- as a set of algorithms and a set of decisions …

Amber: Mm-hmm (affirmative).

Stephanie: Like what happens if you make one little shift in your algorithm, right? Your own personal algorithm. And what happens to the outcome? And for example, um, the space I was working in we were working with these kids who were being diverted from the criminal justice system.

Amber: Mm-hmm (affirmative).

Stephanie: And those kids would have police encounters a lot, right, so stop and frisk kind of encounters. And so we tried to get them to understand the sense of what power is and whether power is really bringing bravado to that situation or kind of just, you know, going flaccid in a way …

Amber: Right.

Stephanie: … and- and recognizing that the power is in the idea of just calming down, not being aggressive. And then taking the idea of an algorithm for living and saying, well, there are these systems that are all around us that are running these decisions …

Amber: Right.

Stephanie: Right?

Amber: Behaviors.

Stephanie: Behaviors, right?

Amber: Yeah.

Stephanie: Behaviors, decisions, ideas. And they’re touching you very directly in that, you know, the judge probably looked at a sheet that was run through an algorithm that said that you should get this kind of sentence or not.

Amber: Mm-hmm (affirmative).

Stephanie: And trying to get them to recognize that. And also really working directly with the material, right? So, um, going online and working with something like Dialogflow and having the kids make their own kind of chat bots very directly. But it’s great because you get input/output, and you start to see exactly how the systems work and how they might be able to flex or spread. And when you do that with your own cultural information it becomes much more ingrained in you. So, for example, I had a kid who made a chat bot based on a rap group named Genesis Apostle. Um, it was very sarcastic because he is, and it would tell about the group.

Amber: (laughs)

Stephanie: And I had another group that they made a really good chat bot that told yo mama jokes.

Amber: (laughs)

Stephanie: So yeah. Um, but it- it was great because they were at once expressing who they are but also learning how that system works.

Amber: Right.

Stephanie: And once you start to see how the system works you can take it apart a little bit to know how to start to respond to it or to know that, you know, there might be recourse if you call it out …

Amber: Right.

Stephanie: … and how you start to work with it.

Amber: To give agency into the process.

Stephanie: Exactly.

Amber: Yeah.

Stephanie: Yeah.

Amber: Um, do you want to introduce the next clip?

Stephanie: Yeah.

Amber: Okay.

Stephanie: So this is actually a clip of, um, some of the guys that I was working with in this space at Recess Art in Brooklyn, New York. Um, and we were talking about algorithms, as I said, and code. And what I like to do is reach people wherever they are. And so a lot of them like to dance, so we talked a lot about dance as a cultural code …

Amber: Mm-hmm (affirmative).

Stephanie: … and then took that and turned it into actual code.

Amber: Okay.

Stephanie: So here’s what they were offering me. And this was great ’cause the guy who’s face was blocked out, you know, he said exactly, “Well, an OG taught me these steps,” which is all about a kind of passing down from one generation to then next. And we moved that into a computer after that and Raspberry Pis and seeing what we could do with that.

So- so this is a next step. So my forays with, um, BINA48, and then this project with kids and working directly led to the thought that, oh, I guess I need to make my own kind of AI, like some other representation in the world. And you can tell I’m very literal ’cause I named it Not the Only One. But trying to make a kind of multi-generational memoir, um, using my own family as the material and using AI as the actual storyteller.

So what happens is we sit down and talk to each other, which in itself is magical because things that the family is sharing are things that we haven’t been told before.

Amber: Right.

Stephanie: And then I run that through … I don’t know how detailed we want to get. But a recursive neural net, and it tries to tell our story. It’s really, really dumb right now. Like it’s a very dumb system. But the experience of making it and the experience of people being able to interact with it starts them, A, thinking, “Well, if she can do it perhaps I can start doing that.”

Amber: Right.

Stephanie: But also starts to think about what the technology is, what it is between sort of information and data privacy and data sovereignty for community, and how we might start to approach that. That’s where my thinking is going. Right? Because, you know, I started this kind of open and then I’ve been doing lots of interviews and podcasts. It had gotten kind of intimate in terms of …

Amber: Right.

Stephanie: … what information they want, which makes you just, you know, “Oh, this is my family’s information. What can I do to safeguard it?” Or what do we do to keep ownership of it or at least control how it goes out in the world while putting it into a system like this? And so doing lots of thinking about the sharing of information through, um, through kind of a- a database of our family’s history. And what you’re actually looking at here …

Amber: Oh, I’m sorry.

Stephanie: Go back for one second. What you’re actually looking at is the first manifestation of this thing, which is a kind of glass … It’s a black glass JANIS, um, form. And- and- and this is another question of design, ’cause figuring out what this thing looks like and- and how it feels for representation becomes really important to me as well. Because I could make it kind of animatonic- animatronic, but it feels off. And I want it to represent, but not so directly. So really trying to figure out those balances and the places that the information can sit in a very good way.

Amber: Mm-hmm (affirmative).

Stephanie: Oh, and this is just some of our input, um, input- input sessions with my aunt. And then so what will happen is Not the Only One will be an immersive, um, 360 installation where you can go in and you … and anyone could talk to it. Right? So even right now it’s running at Carnegie Melon in- in Pittsburgh. And you can walk up and ask it question and it does its best to answer.

Amber: Mm-hmm (affirmative).

Stephanie: Um, it says some crazy things, and it insists on being a movie at the moment. And I don’t know where it got that idea, so it’s very interesting in that it’s starting to insist on things that I’m not sure how it’s insisting on it. The other thing that it does that’s really interesting to me is that my family’s kind of stance, um, if you talk to us about who we are would be happy, happy, love, love. Like we love each other. We’re all happy. And the thing is saying it’s not happy. So it says, “I’m really not happy,” which is really off-putting for me because I’m like, “We just gave you all this love,” and like what? (laughing) Like what is that crazy reading, right?

Amber: Yeah.

Stephanie: Like or are you reading us? Is it like- like I don’t know? Um, so it’s about working with that and trying to see what story it wants to tell, but then also being conflicted about what’s coming out.

Amber: Well, I- I think you’re- you’re teasing out something interesting, and- and- and just ’cause I know we have to wrap it up, it’s- um, this- this tension between wanting to own your own story as it moves through the technology.

Stephanie: Mm-hmm (affirmative).

Amber: Um, but maybe the limits of technology to actually do that. Um, and so I’m very curious, in actually making your own …

Stephanie: Mm-hmm (affirmative).

Amber: … have you learned other … ha- has that taught … what has that taught you about actually teaching and advocacy, and- and maybe even something that we could kind of bring out, um, like as a takeaway?

Stephanie: Yeah. Well, it’s totally taught me about story. Right? So this idea of technology and story and story as an act of resistance or story as an act of inclusion and how we might use that. Because I feel like the more stories we put into the systems, even though the technologies are imperfect, the better we feel or start to feel about these technologies. Right? And seeing what we can do with that. Like maybe Spotify can get a little more sensitive to figure out [crosstalk 00:18:47]

Amber: Be like, “She has a kid.”

Stephanie: Exactly.

Amber: [crosstalk 00:18:50]

Stephanie: Like what do you know? Where do we tweak it? How do we get it to know those things?

Amber: Right.

Stephanie: And that’s the magic of playing with it directly.

Amber: Yeah.

Stephanie: Yeah.

Amber: Well, we’re out of time. It was wonderful. But thank you.

Stephanie: Thank you.

Madeline Gannon, Robot Tamer

Design researcher Madeline Gannon on taming robots and reinventing the ways humans and machines interact

In this episode, guest host Aaron Lammer speaks with “robot tamer” Madeline Gannon about the work of her Pittsburgh-based research studio, ATONATON, which combines disciplinary knowledge from design, robotics, and human-computer interaction to innovate at the edges of digital creativity. Lammer and Gannon discuss how to make robots more approachable, how to design their personalities to work alongside humans as “machinic creatures,” and how she created Mimus, an industrial robot outfitted with sensors that bring out its curious personality.

---

Madeline Gannon: Hi, Aaron.

Aaron Lammer: I went to sleep last night after your presentation yesterday uh, like in a sort of, like a vague … I wouldn’t say I had a dream about your robot-

Madeline: (laughs).

Aaron: But it was … it was in my dre- … like my pre-dream thoughts. Okay, so you run a studio called?

Madeline: ATONATON.

Aaron: And you made a robot. What is the robot’s name?

Madeline: Uh, my robot is Mimus. So, this robot is a standard, industrial robot, like the same ones that you would find on a car factory.

Aaron: Mm-hmm (affirmative).

Madeline: And, what I’ve done is I’ve programmed it to have more lifelike, personable behaviors.

Aaron: So, physically describe Mimus for me. For someone who has um, not seen a uh, car … a car-construction robot before.

Madeline: So, this is a one-ton machine. Just a big pile of … of steel and motors. Um, it weighs 1200 kilograms, moves seven meters per second, can hold 300 kilograms to a millimeter of precision. So, this machine is meant to do spot welding on car chassis or um, ah, precision painting on … in factories. Um, it’s … it’s not really thought of as a thing that should be used in real time.

Aaron: Mm-hmm (affirmative).

Madeline: These things are usually preprogrammed to do short, repetitive tasks over and over again. They have really boring lives.

Aaron: So, where did you get Mimus? How did you … How does one acquire a Mimus?

Madeline: This … the … the thing … and … You know, you go to the store-

Aaron: Yup.

Madeline: And if you can carry it out, you get to keep it.

Aaron: (laughs).

Madeline: No, so um, I have been fortunate enough to work with amazing partners who have these resources.

Aaron: So, you … you own the studio ATONATON, which is very cool-looking from what I’ve seen-

Madeline: (laughs) Thank you.

Aaron: And you bring home your uh, foster uh, animal-

Madeline: Mm-hmm (affirmative).

Aaron: For the first time at a … at like a … a default firmware level, what is … what comes on Mimus before you start working on Mimus?

Madeline: The … An industrial robot is the … the mechanical parts of it, the-

Aaron: Yeah.

Madeline: The bits and bots of the motors around the joints and the struts in between and then it has a brain. This control box that has um, its own software for controlling those lower-level things like … like motors. So, the work that I do and … and uh, basically just opens up that brain and lets me tinker around in it. So, I put some code on this control box that um, just listens for constant information and commands from me. Um, so, rather than being this closed black box system, I made a little doorway into it that I can send and stream information in real time.

Aaron: Do you have a preference as to whether I refer to Mimus as “he,” “she,” uh, or anything of that variety?

Madeline: Well, I … I call her a “she.”

Aaron: “She”? Okay.

Madeline: Um, that’s the way I … that I think about her but-

Aaron: Okay, let’s … let’s go with “she.”

Madeline: Okay.

Aaron: I want … I want to speak about Mimus in your na- … in your native uh, cadence.

Madeline: (laughs)

Aaron: So, Mimus’s brain is an alchemy of code that was already part of Mimus when you got Mimus and code that you’ve written on top of that.

Madeline: Exactly.

Aaron: So, some of the car-welding Mimus is still there?

Madeline: Definitely.

Aaron: Do you end up disabling portions of the original way that Mimus works?

Madeline: For … for me, I think, the important thing is to embrace this. When I … when I think about this machine I … I describe it almost like you’re working with a creature.

Aaron: Mm-hmm (affirmative).

Madeline: And it’s a machinic creature-

Aaron: Yeah.

Madeline: But a creature nonetheless. So, to work with their idiosyncrasies. So, Mimus’s movement is a little bit jerky.

Aaron: Mm-hmm (affirmative).

Madeline: Um, sometimes, because this thing is not designed to respond in real time-

Aaron: Yeah.

Madeline: Um, there can be some latency in … in me sending a command and the robot doing things. So, as a designer, I sort of, embrace these limitations as a quality of the personality of this machine and I try to um, work with them as best I can to really bring to life this individual personality of this robot.

Aaron: So, tell me what your goal was with Mimus and tell me how that … how that was implemented at a software level.

Madeline: For me, the … the goal of it was really to make an opportunity for people who may have never seen this amazing machine that’s probably um, made the thing that they drove in today. Uh, give them an opportunity to come face-to-face and really cut through some of the … the hyperbole that we hear in the media about robots taking our jobs or robot overlords and let them have a face-to-face conversation with this incredible machine. Now, that being said, I think um, my stance in it was very neutral. Um, I tried to show that this … this piece of hardware that’s really just taken off the shelf um, can be reframed with a little bit of clever software and duct tape to bring it to life in a new way to show, a sort of, alternative vision of what we could do with these machines, if we so desired.

Aaron: The … The word I would use to describe the interactions I saw with Mimus and I … uh, this … these videos are on the Internet, right? Somebody who’s listening to this can like, google Mimus-

Madeline: Definitely.

Aaron: And see Mimus? Okay.

Madeline: Definitely, yeah.

Aaron: When Mimus is interacting … and this is in the exhibition capacity. I … I don’t know anything about the private studio life of Mimus, but in … in the … in the exhibition capacity um, it reminded me a little bit of being at a zoo and some animals you see at the zoo don’t care at all about you and sometimes an animal will, kind of, come up to the glass and be like, “Whoa.” Like, “It’s … it’s curious about me.” Tell me about, like how … how Mimus expresses curio- that idea of curiosity.

Madeline: So, that was something that I … I really tried to emphasize. That this machinic creature is the first time that you might be visiting them and it reminded me of, like, the first time you might see a giraffe in real life.

Aaron: Mm-hmm (affirmative).

Madeline: It’s just … You might … You might see it on television, but when you’re face-to-face with it, when you can hear it,* when you can smell it, um, and sense it, it’s a whole ‘nother beast. Um, so we definitely played up that … that uh, motif in the actual experiential design of visiting Mimus at the design museum in London. So, the idea of … of pulling out and … and of a personality with her was … was to play off this idea of curiosity. Um, that the people who are visiting her … It’s a bit of a spectacle. She’s … she’s loud, a little obnoxious um, and … and what I wanted to show is that she is equally curious about us as we are about her.

Madeline: Um, so, there’s some ways that we can do that in the interaction design. One of the challenges and actually, benefits, of working with this robot is that there’s a really restricted material palette to communicate these emotions. You have her pose. You have her posture. And you have the sound of her motors. And through those three things, uh, we can build a basic body language that’s quite natural to this machine and its kinematics and how it moves. So, some of the things that um, I did to, sort of, elicit a sense of curiosity was, when you’re far away from Mimus, she looks at you from above, so above your head height. And sometimes that can seem like a very frightening thing when this 1200 kilogram beast of a machine is looming over you. Um, it can feel very threatening.

Madeline: As you come closer, she switches and becomes below your head height and, sort of, sniffs up at you, like an excited puppy and jitters a little more.

Aaron: I … I felt a sense of empathy for Mimus. What … What kind of reactions do you get from people to Mimus out in the … out in the wild?

Madeline: I mean, it was … it was really incredible to see. Um, I worked, you know, 10 hours a day, 12 hours a day to bring this robot to life for about two months and then she’s out in the wild and uh, and on different continents.

Aaron: Yeah.

Madeline: So, really like, checking in. I had nanny cams to check in on her uh, remotely, but … but seeing things from social media, for example, and how people responded was pretty wild. Um, so, there was a whole range of emotions that we were able to elicit, which I was hoping for um, diversity of … of reactions. I didn’t have any prescribed goals for it. So, uh, from friendly curiosity was one thing, to um, people bringing little items and gifts and kissing the glass and tapping on the glass and um, kids really loved it right away. We … we favored the algorithm for Mimus to look for shorter people, um, so that it would go to … to children first and … and get them engaged and some … some uh, also some distrust and some creepiness about this overtly dangerous machine that could somehow seem so cute. Like, how easy are our emotions manipulated that we project our feelings onto this thing?

Aaron: Does Mimus learn?

Madeline: In the current state, Mimus doesn’t learn. So, we do some machine learning to do some gesture recognition. So, for example, if you’re really, really excited you can steal her attention away from someone else. So, that gesture that I looked for through the sensor is … is basically waving your arms up and down and hopping up and down so-

Aaron: The, kind of like, “I need to pee kind of [crosstalk 00:10:36]”-

Madeline: A little bit yeah, exactly. Like, “Come see me” and then … then that assumption is like well … well um, I should mention that Mimus also has a … has a bored timer. So, if she’s sni-sniffing you out and investigating you and you stand there and do nothing, she’s going to get bored, just like animals at the zoo and go check out someone else. So, getting really excited is the way to steal her attention as well as keep her attention for longer.

Aaron: What was unexpected about Mimus out in the world that you didn’t see coming um, when she was just in your studio?

Madeline: I think, for me, one of the pleasures that I have is how close the museum staff got with the robot. And they had a little send-off for her when … when the exhibition closed and … and the … the director forwarded me a thread of like, goodbye letters to Mimus. So, there was a … there was a real caring for her, which um, again, and like you know, it’s … it’s poetic and … and lovely, but, in a machiavellian way, it meant that my interactive installation got taken really great care of uh, remotely, but so … so it … it’s a really interesting tread to … to walk to try to elicit emotions without really manipulating emotions. Um, but that … that was a really unexpected thing.

Madeline: Um, also, people who visited her multiple times was really nice and um, and that … that was something that … that was really pleasurable.

Aaron: How did you get interested in robots in the first place? Did it start when you were a kid?

Madeline: Um, not really. I mean, I’ve always liked sci-fi, um, and robotics, to me, is something that I … I’ve only fell into recently-

Aaron: Oh, okay.

Madeline: Um-

Aaron: So, tell me what you were doing before you got into-

Madeline: Architecture. My-

Aaron: Architecture.

Madeline: Yeah. My technical training is in architecture um, and the last year of my master’s at my university I uh, my university got this CNC router, which is a machine that you can uh, connect to a computer and it can carve out material um, with … with a … with a bit that spins out like a … like a drill press on a machine.

Um, so for me that was the first time that I could take my very classical architecture education in 3D model … imagining in 3D modeling environments in a computer and actually translate them out into the physical world and that sense of instantaneous um, translation was really empowering and um, intoxicating that I … I really quickly hit the limits of what I could do with this machine that was really designed for carpenters. Um, so, I was, you know, shoving pens inside of carving tools and experimenting with things and … and all these … using materials that I wasn’t supposed to and what I … what I came to the conclusion is that the … the biggest limit for my creativity to working with this machine is that I had to communicate it through software that was designed for other people.

Madeline: So, that’s when I decided to, sort of, jump into this rabbit hole to learn how to program, uh, learn how to talk to these machines and to see how we can, sort of, blur the boundary between our imagination and our digital creativity and the physical world.

Aaron: How does one start learn … I mean I wouldn’t even know what programming language you would start with if you’re trying to make a Mimus.

Madeline: Um, there’s … there’s a … So, for robots and spec- specifically one of the challenges of working with them is that I … Each robotics manufacturer has their own proprietary language for their machine. So, a lot of it is um, just investing time. And there are also, like, for the industrial robots they’re so far behind in how they share knowledge. It’s usually like, ace you have 15 PDFs with little pieces of information that you have to control F and find random things, and it’s a lot of headaches. It’s … That’s one of the reasons why I built a back-massaging robot was because it was just so stressful to program it I needed some … some tension release.

But, that’s one of the … one of the big draws that brought me here to Pittsburgh is, as a robotics capital of North America it was um, an amazing playground to … to start to experiment.

Aaron: Yeah, I was wondering like, what’s … what’s the robotics community like here? Like, other people come in to see Mimus and they’re like, you can come see my Mimus over here-

Madeline: Yeah. I mean it’s pretty incredible here. You know, you go to a … a coffee shop and you … you hear people talking about, “Oh, you know, my encoder on my joint isn’t really doing well.” It’s, “Oh, my gasket’s viscosity isn’t quite … “ It’s just the … the conversations that happen here are so nerdy and so interesting and so diverse and … and, kind of, odd that everyone is in their … their own um, deep well of knowledge.

Aaron: So, what are your plans with ATONATON? Where does one go with a ro- a robotics studio, if that’s how you define it?

Madeline: We … We do a lot of our work with robotics and I think for me um, what I try to focus on is, sort of, scouting under-explored territories in technology and how technology connects with people. Um, so, a lot of that is translating a lot of the energy and … and intelligence that’s happening in the virtual world into physical, tangible experiences. Um, so we explore many topics outside of robotics, for example, wearables and 3D printing and fabrication and all these things that can begin to break down barriers between our … our imagination and what we can actually physically produce.

Aaron: Tell me about the decision to not have a face. Um-

Madeline: Yeah, it’s … it’s naked. It’s-

Aaron: Naked.

Madeline: In the raw. Like, even for people who work with industrial robots on a daily basis-

Aaron: Yeah.

Madeline: They probably haven’t seen one like Mimus without something on the end of it doing something.

Aaron: Yeah.

Madeline: So, it was a really concerted decision to … to keep her in the raw in … in her … in her element.

Aaron: Yeah.

Madeline: Um, and a part of that is … is sort of cont … to be a little bit of a contrarian to how robotics are dealt today is you have a really cool robot, it has to work with people, so you slap a screen on it and maybe it has some eye balls that looks at things. And to me, that is just such a missed opportunity to really explore the natural lifelikeness of this thing that is articulate and can move in the world and act in the world, um, because we … we interact with things in our daily life, like, for example, our pets, that they don’t look like us, but we can communicate with them. Um, in a really uh, eh, intuitive way, sometimes trained ways, but there’s … there’s that one-way relationship with them that … that we can negotiate with one another in a shared space and enjoy each other’s company.

Aaron: What … What are the most negative reactions you’ve gotten to your project been?

Madeline: Uh, certainly like … like, um, “That’s creepy.”

Aaron: Mm-hmm (affirmative).

Madeline: That, “This is … this is not good.” Then, “This is … this is not the future I want.”

Aaron: Yeah.

Madeline: Which is … th- those to me they’re not negative that … that those are incredibly valid um, emotions to have.

Aaron: Yeah.

Madeline: Um, and that those are necessary to come to the surface so … so we can decide how this is um, as a society.

Aaron: Yeah. I mean, you kind of, have to feel a little bit like you’ve … if someone had a very visceral reaction to what you’ve done, you have to feel a little proud that you-

Madeline: I was going to say-

Aaron: Caused a visceral-

Madeline: Like-

Aaron: Reaction.

Madeline: Like, the negative reactions are like, “Meh.”

Aaron: “Robot’s boring” (laughs).

Madeline: Yeah.

Aaron: Um, so where do you go from here? What’s next?

Madeline: Um, I actually just added a new robot to my family.

Aaron: Oh.

Madeline: Um, so it’s a much smaller robot and … and it … it can travel a lot easier. Packs up into some crates. So, I can actually-

Aaron: (laughs) You designed a robot that can fit into an overhead compartment?

Madeline: Uh, it … just about-

Aaron: Yeah.

Madeline: Just about um, so … so I’m doing a lot of work with that. I’ve actually been uh, working with some film directors about um, how to make sentient machines uh, as useful tools for them and … and some other stuff but … that hopefully will … will come to the surface soon and for me it’s just uh, robotics is … is certainly a passion and I think it’s a … it’s a interesting topic that is under-explored and I’m always keeping my nose searching for the next under-explored territory.

Aaron: Excellent. Well, thank you so much for this interview. This is super interesting.

Madeline: Oh, my pleasure.

Aaron: Um, where can people who want to know more about Mimus or want to know more about your studio find you?

Madeline: Yeah, that’s great. My website is atonaton.com A-T-O-N-A-T-O-N.com and uh, you can also Google uh, “Madeline robot whisperer.”

Aaron: Uh, Ma … and uh, Mimus is M-I-M-U-S.

Madeline: Yes.

Aaron: If you’re looking for … just for … just for Mimus. Thanks.

Madeline: Thank you.