 Oh, there's so much food up there. Oh, there's so much food up there. They should be tested. Except for his gallerosis. Oh, there's so much food up there. Oh, my God. Oh, I have someone else. Just got a really big, really big box. Oh, but then I started reading it as a good thing. I was just, Oh, no. They've been preparing the warts. It's two hundred. It doesn't matter who is who. It's multi-national. It's an other day. See, I don't know, should I go, I'm because I'm ancient. Okay, it's like, you have to go like this. Oh, okay. Okay. What? Okay, okay, okay. You know. No. No. Definitely. Yeah. Sit together. Sit together. Okay. Sit together. Yeah. Yeah. You're right. Right. Right. Right. Right. Right. Yes. Right. That's also the 2CG company in June in Arizona. Nothing's happening in Arizona. Yeah. That's what I'm saying. I don't know, but it's also during the day. Yeah. Yeah. Yeah. Yeah. Yeah. Yeah. Yeah. Yeah. Yeah. Yeah. Yeah. Yeah. Come on, everyone. Yeah. Yeah. Yeah. Yeah. Yeah. No. All right. Yeah. Yeah. Yeah. I'm not sure. Yeah. Could you move back in? Yeah. I just have to read those for you, and we need the kids to look at the other names and ask them how to read them. Are they going to see you? I don't know what to say. I don't know what to say. So it looks like you need to do something, right? You want to write a statement for the community, right? Yeah, yeah. You need to help. It's just like you don't need to help. You need to help us out. I would have a professional, I don't know if it's going to make a lot of money. That's actually a good thing. If you're a performer, you live on whatever the show is. Seven gigs. Yeah, I have seven gigs canceled last year. Oh my god. Yeah. This is pretty much everything right up to that. Oh, always that. Yeah. We're going to have... Yeah. That's kind of what you're saying. This is how I started, you know? And then you do this, which is back in the days. Now I used to get stuck. Yeah, hold the tape and you're out. You have to put them on. And then you push... I'll make that touch the sound. Do you want to touch it? Yeah. Let's touch it. Yep. Sorry. What to do? Yep. I know. Good evening. Good evening. Oh, okay. Good evening. My name is Muriel Borst-Harrant. I'm from the Kudorapic Nation from Animations. And I am your artistic director of Safe Harbors, Indigenous Collective, here in New York City. And today we were asked to do the land acknowledgments. And what that means to us as Native people in urban areas or in New York City is we were asked to do the land acknowledgments in urban areas in New York City, especially. I would just like to give a little history about land acknowledgments. And one of the stories I was told was that Manhattan has always been a place for intersections of very different Indigenous people since time immemorial. And part of that was that a long time ago, the island of Manhattan itself, Manahata, was where people all met. And so there were all these different fires and different nations that are surrounding this island. And I always think like 500 over 500 years ago we met and one of the warnings that we got was that colonization was coming and we all met here to be prepared for it. This particular land acknowledgment that I'm making has been vetted through the local communities and through several of the different leaderships in this area. And I would just like to read that for you. Manhattan has always been a gathering and trading place for many Indigenous peoples, where nations intersected from all four directions since time immemorial. It was a place to gather and sometimes to seek refuge during times of conflict and struggle. The staff here, Safe Harpers, Culture Hub, and La Mama pay respect to all their ancestors past, present, to their future generations. We acknowledge that this theater and this work is situated on the island of Manahata on the island. Traditional lands of the Muntzi Lenape, the Canarsie, the Ankhachank, the Shinnecock, the Hadeshoti Confederacy. We respect that many Indigenous people continue to live and work on this island and acknowledge their ongoing contributions to this area. Thank you very much. I'd like to introduce to you my better half, Kevin Terrent, who's the lead singer of the Silver Cloud Singers and Dancers. Thank you. Good evening, everyone. I'm very happy to be here today and asked to come here and kind of do an opening that they were recognizing the Indigenous people of this land and that we're here and we're still here. We're not just in the past, we're still here today. And I'm very happy that they recognized that and that we're here to do this. And what I'm going to do is sing a song for you. And this is an actual song that was composed by the group right after 9-11. It's called New York, New York. And I think it's a good song to sing tonight for all of you out there and all of us that are here. You know, this is where we're from, this is where we live, and this is where, you know, where it's at. So I'm going to sing this song and it's called New York, New York. So I ask that you think of good positive vibes as we sing this song. And, you know, it's all for you. Thank you to Silver Cloud Singers and the Safe Harbors Indigenous Collective. And we'll see them Saturday, hopefully. Yeah, for a workshop called Native 101. Thank you. Yeah, I think Billy and I are just going to say a couple words to contextualize tonight and why we're coming at you live only. And then we're going to have a night of experiences and fun. Billy. Hi. So my name is Billy. I'm the artistic director of Culture Hub here in New York City. And we are very thankful for all of you that are joining us online. Obviously, you know, it's a very difficult time, I think, for everyone. I think that we're realizing more and more how globally connected we are. We have spent a better part of this last week and obviously the months leading up to it, preparing for this event. And at the very last moment, it became clear to us that we needed to find an alternative mode for presenting ReFest this year, considering what's happening in the world. So as you may or may not know, Culture Hub has been very focused on live streaming and collaborating over distance from the very beginning. We've been around for about 10 years. And I would like to thank our founding partners LaMama, where we are based right now. This is where we're at LaMama currently in New York City. And also the Solence to the Arts, which is our other founding partner, one of the first contemporary art schools in Korea. And I don't know, it's hard to know what to say. It feels a little bit strange. This was an exhibition that was designed to bring artists, activists, technologists, and the general public together to have conversations and to think about our role in reshaping the future. That's what this event is meant to do. Obviously at this time it's not realistic for us to come together in physical spaces. And so we are trying to do our best to find ways that we can still convene people using digital tools. So we're going to be streaming for the next all of tonight. Also tomorrow again from 6.30 to 9.30, is that right? And then again on Saturday probably between 2 p.m. and 7 p.m. You can find the streams if you want to share with other people, your friends. We would greatly appreciate it. We're hoping to slowly build viewership online because many of the artists that are participating have put in such an effort. A lot of the artists, the people that you'll see in the room are the artists of the festival for this evening and otherwise we're not open to the general public here physically. The stream is going out on HowlRound. So a shout out to HowlRound and thanks to them for helping spread the word about the festival and the stream. There's also a chat function on HowlRound. So you can go I think to the main page of HowlRound.com. Is that right Deandra? So if you're in this stream on HowlRound you can see it. Okay, great. So we really would strongly encourage people to utilize the chat and to communicate through the chat. We will do our best to monitor that chat throughout the evening. So if you have questions or you have thoughts or you just want to use the chat to feel connected to other people watching this stream, we hope that you'll do that. It's also going out via Facebook on the LaMama channel as well as the Culture Hub channel and the HowlRound channels. So you can also I think communicate with friends there as well as on the Culture Hub website on the Culture Hub Watch channel. So feel free to share any of those avenues for watching this stream. We greatly appreciate it if you spread the word. I think I'll pass the mic to Maddie who's going to talk a little bit about this year's theme and the activities for tonight. So tonight we're going to start with a performance by Lan Zhang which is, well, okay. First of all, the theme of Refest is regeneration and it's about exploring generation as a concept through intergenerational collaboration, which we realized was going to be sort of difficult given the current circumstances bringing young people, older people together. So that's interesting, but we're still finding ways to do it. And exploring what generative artwork is looking like today and also thinking about regenerative design as a concept for how we can live together, not just sustainably sustaining a certain level of engagement but creating systems that actually regenerate themselves, thinking about biological models that, you know, like I think photosynthesis is a regenerative system. So, yeah, the conversation is going to be big and wide and far reaching. What, Max, it isn't regenerative? It's kind of, kind of. Well, you talk about it later. So we'll have a performance, then we'll have a discussion with little bits of performance, and then we'll have a couple of artist interviews, and then we'll end with another performance. I don't want to say any of the names, but they're all listed on the lineup and I'll announce them as we go. Yeah? Sounds good. Okay, so for our viewers at home, we will take just a couple of moments to set up for our first performance. So maybe this is a good time to go to the bathroom or grab a beverage and we'll have a performance in a couple of minutes. He is the photographer. His name is Brendan. This is Sennheiser number three. Doesn't sound like their sound. Check, check. Oh, there we go. That's definitely my mic. Is everybody good to go? Deandra is good? Okay. Okay, so I'm introducing this performance. That comedy is a computational comedy performance that aims to interrogate our perception of humor through a live procedural generation that reflects the condensed themes and identities in the American comedy landscape. With text input such as words and phrases, virtual comedian modules will complete writing sentences using word references from the transcript compilations. Human actors are choreographed to deliver the output verbally. Lan Zhang, the artist whose native language isn't English and who has struggled to become culturally competent, wants to use this unexpected way of programming to reach the American humor pedestal. The process, however, reinforces the failure of the artist's ideal pursuit of cultural competency and the comic absurdity of the pursuit itself. In the beginning, God created the first thing you said when you dropped this shitty ass baby off at my goddamn house. And the earth was in LA. I was so excited. And darkness was upon the trail the next day and try to prove they are just as big as a glutinous slob as the rest of us. And the spirit of God moved upon the slope. And God said, let there be where I am today if we're not from my support group. And there was no click. And God saw the guest bathroom that it was an earthquake. And God divided the way it is. And from the lady in the porn, and God called the front desk and the darkness. He was trying to get my attention. And the evening and the morning were the cage. And God said, let there be weird at all and let it divide the White House Correspondents Dinner. And God made the meaning of this and divided the answer which under the rug for a couple months. From the waters which he would use as a canvas and it was so funny. And God called the raccoon where you just punch her in the both eyes and knock her over trash. And the cracker in your mouth and they'll find each other in there. And God said, let there be flagellating themselves because it makes everything awkward. All right. And let me address you directly and I try to win you over with logic. And God called the only vice. It's not we all have something. And the gathering together of the waters. Called me third rate and so called comedian. And then a respective comedian. And God saw that it wasn't a violence as a as violence as a public health issue. And it was so meaningful to me. Genesis chapter one, Genesis one, one, one. Shall I compare thee to a Laker game? Thou art more lenient of a friend. And more pain, right? Rough winds do shake the darling buds of dragons. And summer's lease heath all too psyched. But he's all fired up about pride. Some time to hot the eye of Los Angeles where Gadolajara meets Korea and often is his butthole. And every fare from holding is my shit at work by chance or something. But thy internal summer shall not gonna nor lose possession of that day will be when I put on my socks. Because putting on my socks, that means I have to. Here's what I have to do. Nor shall death brack thou wondrst in his family. And then later on, I get a job writing on a TV show for Cedric, the entertainer. When in eternal lines to time, you tape your kid's dance. Tape one second of it and then at 20 minutes of just your own asshole. So long as men can do this not eating thing or eyes can do it. So long lives life this and this gives life to stop breastfeeding after eight and a half months. Shakespeare, shall I compare thee to a summer's day? Hey Jew, don't make it the most beautiful amazing perfect special moment ever. Take a shower at their houses full of bucket with hot water. And you take a smaller bucket and you pour the water on. Remember to live in particular town. Then he has to get a gray Prius. Then you can start to one of the weddings and I asked the groom. I said, hey, why did you decide to get married? Hey Jew, don't know. You were made to the minute you know like my sucks life way better at 45 looks because the situation is okay at 45. Then you can start to think that I'm not that I'm a nice person, but I don't know man. And any time you just have to guess and check your institution by sniffing its ass. Hey Jew, refrain. Don't carry the word at the center of all my other problems upon your fingers. Bled from playing the cello thinking about your butthole licking jokes. For well, you know that it's just a mystery. You know that it's a mystery. Who gives a shit by making his daughter's 15. He's like, she's gonna start having sex. Na na na na na na na. Hey Jew, don't let me run around when I was a kid. You have a job and get the fuck out of here. Remember to let him know a Z's. I got more Twitter followers than the president. Then you can start to drive too far. Hey Jew, don't gotta take a sad song about the joy of soy with Justin Timberlake. Remember to let her life takes all of her shit and then she just becomes a sad ass tree stump with the defated titties. Then you'll begin to make it in the bag. Bag, bag, bag, bag. Hey Jew, the Beatles. I guess we'll just slide right in over here, right? Hi everyone, I'm Maxine LeCohen. I'm a former artist in residence at Culture Hub. And I'm here to help moderate a discussion with three amazing artists and brains who work with generative text and language and technology in both similar and different veins to the amazing work of Lawns you just saw. And all three of them are gonna show you a little bit of their work and what it means for authorship and style. And then we're gonna talk about it a little bit. So first up we have Allison Parrish. Hello everyone. So I have some slides, I don't know if everybody can see that the slides are there. So I'm gonna talk about, I call it phonesthetic style because we were sort of nominally tasked with talking about style and machine learning today. But this, you know, it was a different presentation, I just put this title on it. So I want to talk about machine learning model of phonetics for creative phonesthetics that I've been working on lately as part of my practice. Wait, were we supposed to introduce ourselves? I think we were. I'm Allison Parrish, I'm an assistant arts professor at NYU's interactive telecommunications program slash interactive media arts program and I am a computer programmer and a poet. So machine learning model of phonetics for creative phonesthetics. I want to take you through the model and what it can do and I have a piece at the end that I'm gonna show you and read. So this project started with research on phonetic similarity. I wanted a way to assign coordinates to words in such a way that words with similar sounds would be near to each other if you visualize them on a scatterplot like this. So that you could say that like the words octopus and apocalypse are similar in sound compared to inky and kinky. Both of those pairs are similar but they're not similar to each other as pairs. There's an easy way to get started with this, which is the CMU pronouncing dictionary. It's a freely available dictionary of words and their phonetic transcriptions. My initial work with this was based on a statistical model of the way that phonemes work in these words by breaking the phonemes up into their lower level features and then building a statistical model of that. I was able to come up with a way of changing a word into its transcription into a number that represented the sound of that word. That model I used to compose articulations which came out, it was published two years ago by Counterpath Press, a full-length volume of poetry in which the book is composed of a long prose poem in which each paragraph consists of multiple words that are selected from a large corpus of poetry and the words are glued together based on how phonetically similar they are. So I'm not going to read this, I might read the first parts so you get an idea. These are all lines of poetry from Project Gutenberg and a freely available corpus of public domain texts. So you get connections like sweet hour of prayer, sweet hour of prayer, it was the hour of prayers, and the hour of parting, hour of parting, hour of meeting, hour of parting this with power avenging his towering wings, his power enhancing in his power, his power thus the blithe powers about the flowers, chirp about the flowers, a power of butterflies, so forth, so forth. So it's a kind of poetry in which phonetic cohesion is the only compositional strategy. Where this model sort of broke down is when I got interested in sound poetry and nonsense poetry. This is an example of Baroness Elsa von Freitag-Loringhoven's data sound poetry, it's a piece called I, Gas, Sangren, Jalem and these are all made up words. They were words that were made up for the purpose of this poem. My model was unable to cope with this because it was based, it only worked for words that were in the CME pronouncing dictionary. So if I wanted to be able to compose poetry like this using a computer, which I do for some reason, I need a machine learning model that can work out how words sound, not just words that are in the dictionary but words that you make up. So neural networks to the rescue, this is a block diagram of the neural network that I've been working with, that I made the architecture for and trained on the CME pronouncing dictionary. Basically it is a sequence of two sequence-to-sequence models, one that knows how to sound out words from spelling and the other that knows how to spell words based on how they sound. And so you put a word in and then it actually predicts the spelling of that word based on how it sounds, which is kind of weird, you wouldn't really want to do that in practice because if you already know how a word is spelled, you don't have to ask a computer how it's spelled. But the interesting thing about this model is that you can tap into different parts of it. So you can take the feature representations in the middle of the network or you can use the process when it's predicting what the next phoneme is going to be. You can sort of tap into all of those parts and perform creative manipulations of the underlying data structures of the neural network to produce predictions that are more interesting. So among the things that you can do with this model, one is what I'm saying is denoising sequences of random characters by sounding them out. So you can put in essentially random sequences of characters and then it gives you like a phonetic pronunciation of those, like the NBA has for like the names of players. So you put in just any random sequence of characters and it tries its best to spell them out. Another variation of this is inventing new magic words by taking the phoneme feature vector and then adding random noise to it and then doing inference from that randomly generated noises. So this is like variations on the word abracadabra. You have abracadabra, abracadabra, abracadabar, abracabida, amricadabra, abracadobo, abracabadara, amricatrabar, ablabadara, abracahada, aronakada, agracabparabag, ebrabigraburgmba, actrocapta, afibabagribidi, ebraxbragabad, olvamunvamhamnahavma, and ofovolder. Abracadabra is traditionally an apitropaic word. It wards away danger. So I think I've just warded away all kinds of danger by doing this, by saying these words out loud. Another thing I've been working with quite frequently is the model's ability to interpolate between words. So if you find the underlying phonetic feature vector for two words, then you can average that and then use that as a starting point for where the vector, for where the model will spell the words. You can invent new words that are between two other words. So halfway between paper and plastic is pasite, halfway between kitten and puppy is putpy, halfway between birthday and anniversary is Arthur Day, and halfway between artificial and intelligence is entelophysal. These are new words that are invented by the model. Another feature of this model that I've been working with quite a bit, I've sped up this video, but I don't think I sped it up enough. What this is doing is it's manipulating the probabilities of the underlying phonetic features at the point of prediction. So what you can do is you can say, I want a particular part of the mouth to be used more when you're spelling out this word. So you can kind of make it spell as though it has a head cold, spell as though your mouth is full when you're saying something or things like that. Or it can also do meme stuff. I don't know if I have in this video. Well, anyway, this is a fun thing. But one of the big projects that I've been working on with this is compasses, which is a chapbook in Andreas Bullhoff's sync series that came out last year in which I used this model to write poems that start with quartets of words, words that occur in natural quartets that don't necessarily have a natural ordering to them. And then I used the model to find the words in between each of those words. And then I averaged together all of those words and put that word in the middle, giving you sort of this composition based on phonetic similarity among these words. So you get North, Earth, East, Suet, South, Waust, West, Worth, Ayworth, Noon, Through, Three, This, Six, Nick, Nine, Neon, and Thine, Cyan, Miney, Magenta, Meilat, Yellow, Balo, Black, Blean, Maite, and finally Google, Agilzen, Amazon, Acebound, Facebook, Aspool, Apple, Ogle, Aspool. So that's what I've been working on. That's my contact information, and that's all for me. We're going to try to change displays real quick. Oh, that was really quick. Okay, awesome. Hi, I'm Katie Giro. I'm a computer scientist and also a writer and a poet. I'm doing a PhD in computer science at Columbia University, and I want to talk to you about a project that I've been working on that also has to do with changing text based on different ideas of style, like a totally different idea of style than the way things sound. So I started thinking about this about a year, maybe a year and a half ago, and I was inspired by all the really cool things that were going on in computer vision. So they were able to do this really cool style transfer stuff that maybe you've seen. This is actually one of the first papers that was written about this, and it's almost five years ago now that they started doing this. And the ideas in the top left-hand corner, you have this photograph, and then you could take a painting and change the style of that photograph to match the painting. So in this particular image, you take the photograph from the top left, and then here are three applications of three different paintings. The painting's the little thumbnail, and it's trying to just affect the style. And if you look at this for a long time and you think what it's doing, it seems to be taking the texture of the painting and applying that to the content and global structure of the photograph. And actually, this stuff has gotten way more sophisticated, it's super cool, and now you can just do it in Photoshop, like no big deal. And I was like, this is really cool, but what would this look like for text? Can you do a similar thing for text? What's the texture of text that you could move around? And I did this, I started working on this with my friend Chris Kezzi at Columbia, and we really started thinking about syntax as this thing that was like a surface-level feature. Like it wasn't the content, like you don't want to change actually any of the words that give the sentence meaning, you want to change something else. And so here's an example of what we did, or an example of what our system does. So here's a famous line from philosophy, the life of man is solitary, poor, nasty, brutish, and short. And we said, okay, well let's just rewrite this sentence to have way more personal pronouns. Like this is very philosophical and lofty, let's make it really personal and narrative. And so you tell the system, rewrite this sentence, keep as many of the words as possible, like insert five personal pronouns. And it says, my life, it was a man, my solitary, poor, and I would have nasty her with it, her initial short. So this is doing something, right? Something interesting. And the reason why we got on this whole like syntax, pronouns, this kind of thing is kind of inspired by this work that was actually done in the 60s by these statisticians. So there was this big historical problem about the Federalist Papers, which were these essays that were written by some of the founding fathers, they were trying to convince people to ratify the Constitution, and there were a group of three people who were writing them, but they all published it under a pseudonym. And so for a long time, historians were trying to figure out which essay was by which person. And they were using this like, using other evidence, there were a set of the Federalist Papers that they could not find any evidence for, but they really wanted to know. And so in the 60s, these two statisticians were like, I don't know, maybe we can solve this problem for you. And they thought, okay, you can't look at the content words because each essay was about something different, one was about following policy, one was about economics, whatever. So you couldn't say like Madison liked to write about X, because they would all write on these different topics. But they thought maybe there are these other words in there, like of or with or the. And the authors might have like certain predilections towards using those at different frequencies. And so what they did is they took the Federalist Papers that they knew the authors of and they basically counted words. They were like, how many times did Madison use the? How many times did he use of? And then they looked at the Federalist Papers that were under dispute and they said which author does it most match in terms of the frequencies of these meaningless words. And then they could say with some estimation of certainty, because they're statisticians, we think these Federalist Papers have this chance of being from this author. I think the result is that they were all written by Madison, like all the undisputed ones, which is kind of a funny result. There was some interesting follow-up in the stats world about how legitimate their numbers were, but they actually wrote a whole book on this which you can go get from the library and read. It's riveting. There's a lot of math. But this is part of the world of stylometry, which has this idea that these kind of meaningless words that you don't think are very indicative are actually really indicative of authors or styles or genre. So what Chris and I did is we went and we just took a list of things that we thought we could work with. And so this is our list. The first four are actually syntactic things, which ended up being structural, clausal things, so it's like the number of adverbial phrases, the number of fragments, which actually ended up, we still haven't really gotten that to work, because actually like rearranging parts of the sentence while keeping the fluidity of the sentence is not something we figured out yet. But the rest are actually literally just counts of words, like how many conjunctions in the sentence, how many determiners, how many helper verbs, negation, that kind of thing. And so we also use a neural network where we give it a sentence and we say, here's the number of... here's the counts of all these different features, and then you can change those counts and say, rewrite this sentence with a new set of features. So here's a famous opening line from the romancer. The sky above the port was the color of television tuned to a dead channel. And you might say, okay, this is cool, but it doesn't have enough punctuation. Like, just give me more. And so here's the rewrite. In the sky, the port of color, television, was tuned to the dead channel, that. And so it's doing pretty good. I mean, it kind of gets what's going on here. You might say that you want like a lot more helper verbs, you just like end pronouns. And then it really writes it as, it would be the sky of the port color of the television. I have tuned you on the dead channel. So it's got a couple ideas of what's going on. This is a sentence from Sylvia Plast, the bell jar. I took a deep breath and listened to that old brag of my heart. I am, I am, I am. And you might say this is way too personal. Get rid of all of the personal pronouns and let's like increase the punctuation if at all possible. I think it's also increasing the number of determinants. So then it says, it took a deep breath and listened. That would be old. That would have to be an output of the heart. So it's cool. I mean, it's getting like some of the vibes. So what's kind of interesting about this is that these ideas of this, these like syntactic features and these counts of these like small kind of meaningless words don't like really feel like style sometimes. Because I think if you like sit down to write something, you're not like got to up the prepositions. Otherwise no one's going to know it's me. But I think it's possible that what's going on here when we think about style in this way is that these kind of counts of these weird features are like a trickle down effect. And what you're actually seeing is people really are making these high level decisions. Like if you are writing a Gothic horror novel, maybe you're writing a lot of descriptions of weather and you really care about like placements of objects, right? You're being very imagistic. And then you are going to use certain kinds of prepositions more often than others if you're writing a lot of descriptive scenes. Or if you're writing, you know, certain lengths of sentences or with certain syntax like clausal constructions, you're going to use certain kinds of punctuation more so than others. So it's possible that this is actually a reflection of what writers are thinking about. But who's to say? I mean, I think I'm still not convinced that I really know what style means when you talk about text. So I'm excited to figure it out. Blair's going to figure it out for us. That's amazing. I'm not so sure I'm going to figure it out, but I'm certainly going to talk about it. Okay, so I am Blair Simmons. I am a lot of things. I have a playwright and an artist that uses computation. And I teach and I don't know. I really like watching television. So I am going to talk about modeling conversational relevance. Because I come from a theater background, I was in the middle of a different project that became this project. And I was taking a source text. And I was mixing up all the words to create a new monologue. So I was taking the instructions to inflate a balloon. And I was mixing up all the words to create a beautiful monologue. And then I thought to myself, I could just write a computer program that does that. And then I was like, no, no, not right now. We'll do that later. So a year later I decided to do that. I wanted to do it for my master's thesis at NYU. My advisor was actually Allison Parrish sitting right here. So I'm honored to be part of this conversation. And what I really wanted to do was I wanted to create a play that I could sit in as an audience member for. I wanted to have my theatrical performances surprise me the way that it surprises my audience. And I wanted to create a structure for that. And so to do that, I had to start asking myself, or actually more accurately, Allison was asking me, what makes up a play? What are the features? What are the elements that you can hone in on and create a model for? What I ultimately landed on was dialogue. And I landed on dialogue because it was a particular style of theater that I was trying to emulate. And this was like the theater of pinter. And I really wanted to create dialogue that was people talking past each other and using the same words for different meanings. And they repetitive language. I really wanted to create this. And so in order to do that, I essentially needed to solve the question of conversational relevance. And I went about using some machine learning techniques like Markov chains and other things. And those all didn't really work for me because they didn't have the memory that I needed them to have to create a scene that made the sort of sense that I wanted it to make. And so I'll show you an example of the types of scenes that my actors read. So here's a scene. The actors can actually take however many lines they want, but I'll just read it for you now. Can I add up a column of numbers now? We'll use a computer later. If you should be designing software, you cannot use a computer any more. Should I consider designing software in a headache? When you're fixing a computer, I have a headache. I'm fixing a computer. Oops. You have an exhaustion from the... I think so. So here what's happening is the way that I decided through my style that I decided to create dialogue through words and phrases that essentially are connected to each other in some way. So we have column of numbers and computer. So we kind of see how that could be created. You know, those could be related. I said created. But designing software makes sense in terms of computers. Computer designing software, headache. And so somewhere in the navigation of words and phrases, we ended up from designing software at headache. Perhaps maybe designing software gives you a headache. You could make that. And essentially what I wanted to do is I didn't want to make those connections between those words and phrases obvious, but I did want there to be a sense of a connection so that a conversation could be happening on stage and in the minds of the audience without me deciding what the conversation was going to be about. Yeah, and I'll read a second one. We'll buy a house later. It's like you pay for it sometimes, but don't cut your hair without me. All I do is buy a house. I cannot. You pay for it when really... Can't you cut your hair? I can't pay for it. No, you should improve your image. All you do is cut your hair. Don't cut your image anymore. I will get a job when I want to get a job, but don't pay cash, really. All you do is get a job. Oh, very well. Give me the charge. You want me to pay cash? You've got a charge here. Did they bring you the art? I don't know how we got there, but I only have one paint. Damn it. There was art here. Wow. All I do is illuminate. I'm sorry. That is part of my paint. Or isn't that the Miniat? To be clear, I don't know that word. Good. So that I can illuminate. I think so. A Miniat personality thinks the color was always well emblazoned. Wow. You've got a color there. And so what's amazing about this is we have some semblance of sense, but the sense is made actually in the reading and in the perception of the thing. And I think that's really all I wanted to share other than I wanted to say that we're dealing with semantic networks. I just wanted to add that little thing in there. And in order to do this, I actually wanted to start the discussion portion immediately. I had to make a lot of stylistic decisions about what theatrical dialogue was to come to this. This isn't what all theater sounds like, but it's a type of theater. It's almost like a parody of a type of theater. And so, yeah, that's how I feel. And I feel done, also, in addition. Great. This is working. Okay. So my first question for all three of you is all three of you are effectively collaborating with systems you've created to a degree, right? Or not. But I guess what I'm curious is how your style as a programmer or a creative technologist intersects with your style as a writer or a poet. I can say I have sort of a tendency to enjoy nonsense. So I take a lot of pleasure in the types of language that comes out of my program. That's a really, oh, yeah. I think it goes a lot in one way, but not the other. I'm not sure, like, I don't think this particular project where I thought, like, a lot about prepositions and how often they're being used, like, changed the way that I use prepositions. But I do think that when I was looking at the academic research on style and the things that they were doing, I did find as a writer I found them, like, lacking some kind of weird and having blind spots. So, like, a lot of the academic research on style and text will do things where you're like, what do you think style really means? So, like, there's a style paper that's all about changing the sentiment of a sentence. So it's like, oh, let's try to take a movie review that's negative and rewrite it as a positive movie review. And this is, like, maybe an interesting thing to do in the world. But it seems weird to me to call that style, right? So, I think that is an intimate part of the meaning of the sentence. And I think that's where, like, sensitivities around thinking about things as a writer or as someone who cares a lot about literature makes me reflect on it a bit differently. I don't have a creative writing practice that's separate from my computational practice, so I can't address this issue. No, but I think that is a good answer in a way. I thought it was a good answer. What are you saying? A more abstract question that I've always wanted to ask all three of you, so I'm going to take this opportunity to do it in front of people. Why do you do what you do? I must know. Particularly, like, particularly, like, I'm interested in what triggers made you obsessed with this. I think that's, like, a really... So, like, a Freudian approach. Yeah, let's go there. Alison and Katie both hard pass on that one. No, I'm perfectly... I have, like, a set pat answer for this, I guess, because it's, you know, no offense, Max. It's a question that's often asked, like, with some measure of incredulity. Like, why? Why would you do this? No, I say it with excitement. I ask it with excitement. But other people are incredulous. Do you want me... I was letting you go first as my former student. I feel like I need to make space for the new Vanguard. Definitely. I think I know your answer, and I think, ooh, mood lighting. And I think my... Did I look too sweaty? Was my laugh popping too much in the mic? No, I think that my answer is really theater-specific and project-specific to this, because I was trying to solve that particular problem. I think I did touch on that. I really wanted to see a piece of theater that I created without knowing what would happen. That was, like, a particular problem I was trying to solve. And then, once I started upon that, I became obsessed with the fact that I could create my own, like, Steinian experience in theater. And this project was funnier than I thought it would be. I think in my mind I was writing, like, very serious short plays, and then I got addicted to the laughter. But I think that... But your answer is so much better. For me, it's a confluence of just a number of things that I sort of progressively got interested in as a kid. One was reading JRR Tolkien at an early age and getting interested in invented languages and how they seemed to have an expressiveness that resulted from their... First of all, from their, like, external characteristics, but then also, like, once you dig into them, they're all these interesting structural linguistic things that Tolkien was doing. And I went to UC Berkeley to study linguistics specifically because I was interested in this art form. In high school, my creative writing teacher just as a one-off had us read as a wife has a cow, a love story by Gertrude Stein. And that's, like, sort of coming at the same issues as Tolkien weirdly from a different direction of, like, language being able to operate in, like, its own systems. And then I also was, you know, I love computer programming, like, as an activity, as a form of expression, and just as, like, you know, it's like knitting or quilting for me. So, like, the craft aspect of being a computer programmer also folds into that. And then, like, you see it through the lens of the history of the poetic avant-garde and, like, all kinds of experimental poetic practices over the course of history, not just in the past hundred years or so, I've dealt with algorithmic methods. And I think that's a really exciting intellectual heritage to be a part of. So, yes. Yeah, I think maybe I'm a bit similar to Allison where I'm really interested in language and I also love computers. And I think a lot of just what I want to do has to do with, like, being able to play with language. Like, I think really when I showed the style, the image style transfer stuff, that stuff that goes on in computer vision is always very inspiring to me, because I'm like, oh, they get to do all this cool shit. Like, I don't know. Somehow the visual artists are having way more fun. And sometimes when I try to explain my work in academic contexts, I'll use, like, Snapchat filters as an example of something that's, like, really empowering to people. Like, people just have so much fun with, like, making their face into a dog and all this other kind of stuff. Like, in a really, not like in a demeaning way, like, it's legitimately, like, empowering and cool and fun, and people feel like they can play in a way that people are, like, afraid to draw a dog, because they feel like they can't draw. And I think especially for the style, I was really, and I still, like, just what does, the text is not like images. Like, there are so many differences. And as writers, you don't get to use the same kind of tools that visual artists do. But I think a lot about, like, but what if we could, or, you know, when you bridge that analogical gap, what do you get on the other side? And what does that say about how language is different than the visual arts? Yeah, that's a great point. Like, with visual, tactile media, or other kinds of, like, non-literary media, I guess, like, you can describe the things that you do with gestures. And, like, I have this gesture, like, I want to do this to language. I want to, like, punch it and stuff. And you can't do that with our, with our composition tools. But with computation, you're open to these different verbs that you can use to operate on text. And I think that that's super exciting. Yeah. So, I'm going to use the microphone so people can hear me. We've had a little room of people who have been watching this discussion and the performance. And I wanted to bring them in now so that they could maybe ask some questions and be a part of it. And, yeah, the reason we're doing this is because, obviously, this is closed to the public aside from live streaming, but we wanted to experiment with how we could gather people online to watch and engage. So, yeah. Can it happen? We're excited to meet you. Oh. Hi, Mimi. Navi, is that an invitation for us to throw out questions at this point? Yes. Can you hear me? We can hear you. Oh, sorry. Cool. I'm not in the room with you so I can't tell if I'm being heard or seen. Introduce yourself. Hi, my name is Benton. I'm also a culture hub alum having done, I don't know, a few projects with the team thus far. And I just really enjoy the talk. I've been a fan of computer assisted writing for a long time. I used to have a book of computer poetry and interestingly enough, for some reason, there was a character named Benton who kept reappearing in the in the poem, so I don't know where that came from. But I was curious if you could give, you talked about how, you know, in the visual, I'm a visual artist and I am likewise amazed at this wealth of tools that just anybody could use and actually personally I believe it's really advancing the art of moving visual image. But what is an example of like the text equivalent to a Snapchat filter? Is there something out there like that? What if I want to write, I don't know, a letter to a friend, but I want it to sound like Shakespeare or whatever. Is there such a thing out there? I mean, one of this is just spell check, right? But I'm serious about that. That's absolutely like an algorithmic or machine learning model that we use every day to communicate with our text, right? And I think, you know, both of the projects that Katie and I showed were both about that operation of text goes in, you move some sliders around, and then you get a different text out. But that specific example will like Katie address that. Yeah. Well, so first of all, there is like a two Shakespearean thing. Because there's like, because there's modern translation of Shakespeare, like a pretty easy thing to do. They have a lot of like parallel aligned data, which is cool. But I think there's like a sub-question which is like, I don't think there is the Snapchat equivalent in the sense that people don't use it in the way that people, like we're not all sitting here on our phones texting with a little thing that uses Allison's thing that makes it sound like we have a head cold, even though that would be supremely fun. I mean, it's like, it's like, it's like, it's like, it's like the mobile version of that. But I mean, just like, super briefly, I think part of that, and this is what makes this work so interesting, is that we have a really different relation to text than we do to images. Like there's something about manipulating a photograph of yourself that seems like fun and playful. And something about like manipulating your images is weird in a way that I think it's not weird for photos. Do you guys use Grammarly? No. I've used it only for like research, like to know what's up. I have it on my work emails, and you can check to see how professional, how formal your email is. And that to me feels like a very pretty much like everyday use. And I feel like we're pretty close to your email is too female. Like you're using too many exclamation points. Or the other way around. But there's a question tree of like why that hasn't taken off. Like it would be pretty not, it's not so unreasonable that like Gmail might have baked in like this is a sad email. I mean arguably they do. Like there has to be like some kind of model, statistical model of style that's going into the smart replies and stuff like that. Which I always turn off completely in any writing interface I use. I don't use spell check, I don't use auto complete, I don't use anything that's suggested by a machine learning model. Specifically because it like feels like it's writing it for me. Which I think like addresses like speaks to your authenticity issue of like being truthful to your mom or your mom. So that's the question. I had a question. Oh, there's just like no feedback. I'm in a bubble. So it in listening to the three presentations on text in particular. I realize that working with text generatively is always an act of regeneration as opposed to when you can do generative visual art and it's completely abstracted and you're not necessarily quoting or depending on resemblance to what has come before to engage human humans in the output of what you're doing. Meaning especially in Allison's examples with the words um the sort of what makes them work or what makes it so that our brain actually latches onto them is just this like odd feeling of familiarity and recognition and also that anything in English that has the word ass in it somewhere is kind of funny. You've discovered my secret. Yeah. Um, but so I was curious like for all three of you in your research in your work like if you've um kind of rediscovered things sort of refallen in love with things that you had forgotten about just just because it feels like you're constantly looking and manipulating and working with material that is very familiar actually and old um I don't know just wear your own sort of predilections for things that you like factor into the choices that you make for what you put into um into the algorithms that you're using. I feel like mine I uh I feel like I could explain more parts of my project in order to answer this question um because I have sentence structures that I have written that exist as sort of a branching logic but they're my sentence structures and they're filled in with the associations of words of other people and other things and so I have a bit of an amalgamation of other people's associations and my sentence structures um like what I have decided is a valid sentence structure within this context and so I have the sorry um so I've made some decisions about what kind of sentences and can be said back and forth between between people on stage but um but it's the content and the order of sentence structures that is surprising so yeah, so mine's a bit different in that way I think one of the interesting things about doing this kind of work which I think you kind of spoke to which is a lot of it's old and the reason why it's old is because you want to use things that are in the public domain um so I never get to work with recent text like I would love to work with recent text that sounds amazing I would love to work with novels that have just been published um I think that would be like so much in some ways like so much more interesting and so much closer to this idea of the Snapchat filter because it would be so much closer to contemporary culture and people you can do this people do this with tweets but it's like I think it's like ethically questionable whether or not we should be using um tweet text in research um even though people do it all the time but people do all kinds of things all the time that aren't great uh so mostly you know I'm working like you said like I'm working with things that are at least several decades old and sometimes I'm kind of just bummed about it like it does I'm like oh yeah this is great like it's not my contemporary right like I would love to do something that was a little closer to home or maybe I wouldn't actually who knows maybe it would actually be like kind of terrifying or would like um sometimes you don't want things to get too close yeah I'm not I'm not sure what to add to that other than yeah I use I use lots of like um public domain text specifically for that reason because I want to like have easy ownership over the things that I make um and when it needs data you then that means you have to work with public domain data but like you know I am interested in in a particular set of artistic concerns that are that are really only addressed by this small subset of artists and like I'm I'm not at a point where I feel like I have to be apologetic about that like it's it's it is totally niche and it is like about a very small set of concerns so is there one more question from folks watching or in the chat yeah can I ask the question yes please hi hi hi um so I'm just wondering so um I just want to know about like your experience interacting with all of these like text tools and like your your art practice do you feel like um like all these like much like machine learning models or or these like um trend models that you're dealing with do you feel that they somehow um in reverse like kind of like influence your style like since we are talking about like style for the whole conversation do you feel that that some like that has some influence and on your like artistic create creating process I'm just wondering I'm interested in Blair's answer to that question and actually like in your other work as as a play right have you found like sometimes your you get stuck is like the wrong word that's not even close to the word that I wanted but do you like is that something that you find you like use that as like a set of techniques that you can use to solve other problems hmm um I guess I'm a bit confused I'm a bit confused with the question so I'm going to try to answer it to the best of my ability um the algorithms that are out there um have I actually I think I experimented with them and rejected them I think was kind of our experience in my thesis is I I played with them I found them to be unsatisfying to my needs and uh and I said I'm actually I'm going to not use machine learning was was the choice that I made for for the aesthetic that I wanted um instead I have chosen sort of the random features within computation to to create something that is mutable and changes a lot um and I do think that that actually that actually I think comes from from my my playwriting preference um and not actually out of um new tools in fact like my my program as Allison knows is like a mashup of a bunch of different libraries and it's a mashup of a bunch of different data so that um I could I could create the type of play I wanted to create so I actually think I forced the technology that I had at hand on to um that that outcome that I wanted but but that said what was your significant other's reaction to the first prototype I had to change it because my significant other said that um it sounded like I was talking on stage the whole time um so I had to kind of I had I actually went through a lot of different plays that I liked and I and I kind of stole their sentence structures instead of just writing from me so I so I went and I I did some some some public demand stealing um so I guess it's not stealing but um but yeah if that does that answer your question um um sort of yeah and so just to um explain my questions a little bit because um like my own experience like interacting with um machine learning models or this like auto generating um softwares sometimes I feel like um my style if I'm just talking about like me writing a paragraph or trying to like read something like I have my own style and but somehow when I'm interacting with these models I feel like um consciously or unconsciously I'm influenced by this sort of um machine machine aesthetics so like let's say like the words or like the result I got sometimes unconsciously influence what like the outcome that I come up with so just wondering like what's your feeling or do you feel like reluctant or what's your feeling when interacting with them okay I think I can answer this um yes so machine learning generated text definitely has an aesthetic um what's kind of interesting is it like changes over time is like the popular machine learning tools change over time and yes I totally also feel like you that interacting with them I'm like this really isn't what I'm trying to do um some people like it there's actually a writer called Segal Samuels who wrote a piece um I think in Vox but maybe I'm misremembering where she talked about using this particular language model called GBT2 to like help her write her next novel and she liked the way it captured um certain like tonal textures really and so it would just kind of it was like plot wise drivel but like tonally it had a lot of good ideas so some people find ways to make it work but I will say that the only tool of mine that I've ever built that I actually use in my own writing practice consistently um is a thesaurus so I built this thesaurus for myself um they use something called word embeddings and basically you could give it a source text you could like give it everything um like a bunch of books that Charles Darwin wrote and then you could like search the Charles Darwin thesaurus and so you search for the word work and it'll give you like the most similar words like in the context of everything Darwin wrote and it's kind of fun and weird because it's like not always semantically correct but it's a bit of like a brainstorming thing um there's a a lot of research about you want things to be like not too far but not too close and it does something kind of like that like a regular thesaurus is is a bit too close sometimes and I think the to your point of like style and aesthetics the reason why this is the only thing that I've ever used consistently is because it's at the word level um somehow to me looking for a single word in a larger sentence doesn't seem to disrupt my own ideas about what something should sound like in the way that being given a whole phrase or sentence does ironically this is the only work I haven't managed to publish in a computer science conference is like the one thing that was useful for me um not sure what that says about my career I just want to add one more thing to that discussion um which is that um the your question as as phrased um implies that these writing interfaces are unique in that they affect the way that we write and that is not the case like every writing interface whether it's like pen and paper or Google docs changes materially the way that we compose and the way that we think and the style of the things that we make and things like that so I don't think there's anything new about these technologies from that perspective I think there's a unique um opportunity afforded to them and like being able to make tools that are powerful and there's also unique danger in that like you know we we all use a typewriter which Google can inject whatever words they want both into our documents and into our brains right um maybe other people don't agree with that it came out sounding really paranoid when I was saying but I think it's absolutely I think it's absolutely true if we don't already have sponsored autocomplete then we will very soon um you type coke and it says did you mean Pepsi um so that's that's something to keep in mind that like every writing interface has a politics every writing interface changes the way that we write but also people have been doing like an analog version of this for a long time so um you might be writing something and think oh I wish my style was like a little bit more like weird and heady and intellectual and so you go read something that's a little more weird and heady and intellectual to like get you into the zone um like as Eddie Smith has talked about this that there are certain books she'll go to when she wants her writing style to move in a certain direction and she feels like she can pivot it just by reading um I had a poetry instructor who said before she could write anything she would have to write several pages of just whatever's on her mind and it was because whatever she had been reading was stuck in there and she felt like she had to get that out before she could get to her own voice so this idea of like taking aesthetics from other places is like Alison said like not unique to computers um it just feels it feels different somehow and there are differences but there are also a lot of analog um precedence uh I wanted to ask one if you wanted to answer this question or any of the other questions that we've talked about um I guess I'll just add my answer to the current conversation we're having um I guess in the context of my practice I'm mainly trying to hack into the space of comedy um there are a lot of different mediums that I could have chosen including I was suggested by many people that you just go take an improv class so you can become more comfortable with presenting yourself um and at the same time I was also doing different experiments um with different um machine learning tools including gpd2 um but somehow I feel like um the results that I was given very much alike what you experienced um make me feel like um it's like totally out of my control there's a lot of things that I'm hoping to be more selective about um that I couldn't really have that freedom so eventually I think there's a lot of um pick and choose kind of decide um what really works for you but at the same time I still don't know if what I'm working on would actually get me anywhere um in the realm of comedy but um I guess at the end of day what I'm trying to say is that um yeah there are all different ways um for creators, writers um programmers to um find the platforms that um might work at the moment might not um you might get into challenges but um you know it's a process it's a journey um yeah great well thank you so much everyone sorry we went a little over but thanks for all your time it was really great and thanks everyone who was streaming and watching us and asked questions and uh I'll close by saying as the non computational writer on stage contrary to what you think I love all this stuff so you know don't think that we all hate it okay now we're gonna set up for the wind catcher um and while we're doing that we're gonna talk to a couple of artists um and so first we're gonna go to generative photography and uh Felipe wants me to find my light um hold on let me find it I got it I got it um yeah so let's look at generative photography and this is a project um generative photography and this project and regeneration so this project was born from two very different spectrums of work um one inspiration is from the pinhole camera the orders image making camera um and one is from GANS so generative adversarial networks um we both are students at um ITP which is a media arts program and we do a lot of work with machine learning and generative visuals and while we were doing that we realized that it was others very cool and generating um a lot of different visuals it was lacking the sense of we actually we didn't know that it was lacking the sense of emotional connection um until we started taking a class at the photography department where we learned about the pinhole camera and with the pinhole camera um being one of the orders um image making techniques what it does and how it works is it um it's like a long exposure camera so in every image you take it's at least like five minutes and it compresses all those images and all those different frames into one frame um and what you see is essentially almost like a painting it doesn't even look like a photograph um and we were very like inspired by that visual um and tried to combine these two very different um technologies together try to talk about how yeah so basically what we're showing here on the left you can see these are the prototypes of the pinhole camera and they come in different sizes and basically a pinhole camera so this is a pinhole camera and the way this is actually the lens if you can see the little um pinhole here and um what we do is we put paper in the back of the camera it's just a like an empty cardboard box and close it and the way it works is we then open the lens and let the light in and wherever the light hits the paper it becomes black and wherever it doesn't it's um still white and over time like I said over five minutes um we then get that one frame which is um like basically all of seven minutes um on one single paper and then we try to simulate this process but you are using a digital process so basically what you're seeing now uh these photographs are the analog and the digital prints of the result then coming from this which is a video generated by again and then what we did is to simulate this process of the things that physical world the thing that happened in the physical world where light hits the photoreactive paper constantly and then we try to recreate that by rendering a digital canvas with the digital frames from this video and then by doing over time exposing over time we sort of get this digital pinhole photograph and then we print it out and then take it to the darkroom and then the analog the digital analog photograph making process and through this way we can sort of give this emotional impact or the aesthetics or textures that you can see or serve in analog photographs but this content is entirely digital based so you can see like these tiny detailed like bleeding effects and also if you come here and then these variants in the gradients which is like kind of hard to reproduce by just like doing the digital way and also this one so yeah that's like how we try to recreate these emotional impact like where you actually see an analog photograph so there's like so much nuance in the great scale and the black and white I think now we'll move over to Winnie's work so let's travel over through the exhibition to her piece okay here's Winnie and you can introduce yourself and talk about your work Hi I'm Winnie Yeoh I am an artist and a designer and I'm currently a student at NYU's ITP program I'm originally from Hong Kong and so I've created a series of data-driven projects that primarily explore the relationship between distance and narrative in response to the protests in Hong Kong and so in each of these projects there are three main ones I work with a different protest artifact to sort of explore my distance here whether it's my physical distance being 8,000 miles away from something that I care deeply about the distance in pro-democratic and pro-establishment narratives or the distance in media interpretation so I could go through each of them so the first one is called On My Mind this is a data portrait where I recorded all the protest information I have consumed through social media from September up until today so here on the X-axis you have the dates and on the Y-axis you have the time so it starts from September and this is from midnight to 11pm and the brightness of the boxes corresponds to the number of information I've read in that hour so basically the brighter it is the more information there is in that hour and this is sort of my attempt to bridge that anxiety and the transparency that I feel being so far away and it's as though the only thing I could do is consume as much as I could so I could scroll through and I think an observation that someone had as they interact with this piece is that it seems like I would consume a lot and then I would pause and then consume a lot again and that's quite true especially as the movement kind of goes on and I kind of get burned out and then there are these two pieces that are created from these police press conferences and I scraped the YouTube videos which is actually produced by these pro-Chinese media they have a simultaneous English simultaneous interpretation that they added on to these press conference videos and then I used a command line program to scrape all the transcript and it generated this 600 page transcript that I created a generative text with the transcript isn't perfect but it didn't bother me too much because I feel like there is a parallel between a authoritarian narrative and a machine generated text and so with the text I also look at the most common 300 most frequently used words and with that I made this fridge magnet piece called magnetic bullshit where audience could make sentences starting with the words Hong Kong police and it's quite interesting that I showed this piece last month in Hong Kong and it was interesting to see how the audience aside from making sentences they were also rearranging the magnet pieces and making them into protest slogans and then the last piece called interpretive narratives I was really inspired by Alessandra Bell's work and I was really interested in seeing how a lot of times when we read the same information we could have very different interpretations and I was interested to understand that process so I read a lot of articles about the same incident which is the seizure of the university polytechnic university and so I compiled my own article from a pro authority and pro democratic newspaper and so what I asked the audience to do is that they could highlight the three most important sentences and then they could select some free keywords and also select an image that they think is most appropriate to the article and I also asked them whether they've been following the protest in Hong Kong and if they've lived in Hong Kong and after they this exercise they could see all the responses by other participants and I tried not to ask too many identifier questions so that I could remove myself from judging the answers to people's identity yeah so I was trying to just understand what people think would be the most important whether they would select sentences that are because so what I realized is that these pro democratic media they would focus more on describing the individual and how these students escape the campus whereas the pro authority media would describe more about destruction so I was interested to see which sentences people will pick up and it was particularly interesting the first time that I sent out the survey I actually asked people only people from Hong Kong and China to fill it out but there is actually even divide between the number of people picking the sentences that are more about destruction and the sentences more about the individuals so I'll read some of the sentences that the participants here today have made Hong Kong police behind something unlawful Hong Kong police attack men over a public protest Hong Kong police cause filing demonstration Hong Kong police needs questioning within crime Hong Kong police suspect present sexual litter Hong Kong police report yellow gas Hong Kong police charge music violent Hong Kong police and illegal international trains Hong Kong police hope possible gun room thank you Winnie okay next up we're going to have a performance and let's just check in to see if they're ready have you guys done your sound check so I got this inspiration from the saying that everything has already happened and what we are doing now is the repeat of what I have done and many times people are saying that what you have done is just like chasing after the wind seems like all the process and work we know what is going to happen so I'm just thinking and wondering what is the thing that I really gain from this whole process even though many things have already been down and the human repeats the history over and over again I started with this concept I'm trying to find a relation a correlation between this concept and also the acoustic importance of wind and air in general so I then started to play around with all these objects that I found trying to find interesting sound out of air and wind in general and very interestingly I realized just using mic and put it into different different volume of the air and that creates this vibration the frequency of the vibration really depends on the volume and also the shape of the container it has so I really want to incorporate that into the performance cause by taking this mic into the jar since there's really nothing there but we actually gain something that may even be different depending on the different objects you use then I try to take into fan as another part of it just because of the wind that it can generate from and weirdly it works out in some way so I create this kind of place that maybe no one can recognize making work as a more general process of a human moving around their body I have made some rules for each object I use so I follow the rules and just use that as a metaphor to express the idea of we work and then we always wondering, maybe me I'm always wondering what is this I get out of it I think it's just encouragement for myself and also maybe for some people I guess yes thank you I think that's it look at the audience they loved it cool, I think now we are going to go into the lobby to talk to Yesul song about rice simmer and we're almost at the end of our live stream at night Yesul I'm going to hand the mic to you so you can talk about your work I actually want to show it first show it first this work is oh hi my name is Yesul and I'm an artist and researcher I'm a young faculty at ITP and I'm an artist fellow at Tish and this is called rice simmer so show how it works first so I'm moving right from here to the plate to slowly dim up the light as you can see it is going to take quite a while and how I started this project is there is a human and there is a device or machine and what is in between is an interaction and we usually really intentionally think about it and for example for turning up a light it's so just intuitive and natural to us that we just flip a switch and that happens and I sort of want to take a deeper look into the moment of interaction and want to turn into a meditative and cultural experience and I was imagining as a slow dimming study which has a series of work the one big project and this is one the first project and yet now I see as I'm moving the rice one by one it is slowly dimming up the light and I was showing this at a show and I saw people who are becoming very sensitive to really small voice and changes and like even when it's very loud around people became very attentive to like one rice at a time and then taking a time and taking a meditative moment so I have been really enjoying people interacting with it and then after the it can go more brighter but I also demonstrate after this people take this tassel to clean up the dish it turns up the light again and this is like indication to situation that some certain cultural background is affecting or influencing some very simple tassel everyday tassel that is not related to cultural background at all so I sort of wanted to indicate that moment very subtle way and how it works is I'm using electronics from behind and below the plate there is a load cell which is sensing the weight anything on the dish so it is very sensitively sensing the weight that is placed on top of the plate thank you great now we want to roll some of the video so now we are going to show you a video live stream audience from Rakhima Gambo who is an artist with the Magnum Foundation Rakhima Gambo Rakhima Gambo Rakhima Gambo Rakhima Gambo standing on the back in pigo 3.2c the supporter is balancing himself on his necks and hand is in a crawling posture the partner on the other hand is standing on the back of the supporter and at the same time spreading his hand out like a wing the violin arrived put out the joint artist know where to go you are at the starting you are at the starting point you are at the starting point you are at the starting point you are at the starting point you are at the starting point you are at the starting point you are at the starting point you are at the starting point you are at the starting point you are at the starting point you are at the starting point you are at the starting point you are at the starting point you are at the starting point with vehemence on a might he burns, causing myopia, respiration, warmth, weakness of body, logo. So that's it for us tonight in New York City for Refest Night One. We have two more nights and days of programming. Tomorrow night, Friday, Friday the 13th, we have a discussion with the Santana Project performances by Eron Moreno Ayala and Nora Rodriguez. And we'll have more artist interviews. And I just wanted to acknowledge our collaborating partners who we're presenting Refest with, Center for Constitutional Rights, Magnum Foundation, NYU's Interactive Telecommunications Program, and the Safe Harbors Indigenous Collective who are with us tonight. I also want to shout out our crew who have turned into a broadcast studio in less than a day. So behind the scenes, we have D'Andra, and Songmin, and Jerry, and Filippo, and Carlos, and Jean, who is behind the camera, and Billy, and Livia taking photos. So thank you all. See you tomorrow. Spread the word, please, that we're doing this. Peace.