Douglas Arellanes: We're here with Alexa Steinbruck from the organization… the artist's collective called the Moving Target Collective, and Alexa is one of the members of that, and she's going to be showing a piece called Latent Riot at this year's MozFest. And so, Alexa, maybe you could tell us a little bit about your project?
Alexa Steinbruck: Yeah, sure. So Latent Riot is a Web based project right now, and it started when we discovered a data set of 6000 scanned protest images from the Women's March in Boston in 2017, which were digitalized by university in Boston, and the project was called – this archive – this digital archive was called Art Of The March, and we discovered this website and we immediately… and you see this whole range of beautiful like very individualistic handmade drawings with very witty slogans. And it's just amazing, an amazing collection, and my colleagues and I immediately thought – what would that look like if we take this huge dataset of images and trained a StyleGAN with it, which is a generative artificial neural network? How would how would the results look like? Were they able to capture maybe the like visual language of protest or what would it look like? And then we trained this neural network and the results… the results were pretty funny because you could see that they captured this aesthetics. But what it didn't capture, of course, was the text. So all the text was transformed into these blobs and weird shapes and graphical elements, and we thought, that's amazing. Like that's a pretty interesting commentary to the debate about what AI can accomplish and how if AI can have its own agenda and understand what it's doing. And we thought, that's perfect. You see images created of… protest images of an artificial intelligence system that mean nothing and have no intention at all.
Douglas: The humor in it was what was really surprising. Oftentimes a lot of the debate around AI is extremely dystopian and I was really encouraged to see that you're approaching it with a different perspective.
Alexa: Yeah that was also our goal to make an ironic statement, and we hope that people see the irony in it, and if you put it into this context of the debate, there are people or organizations that vouch for robot rights, which is a the idea to give AI systems basically. to see them as legal entities and give them something like human rights, which is bizarre. Because these systems have no… the level of intelligence that systems, AI systems today have are so so far away from… from human intelligence. And this is what's captured as AGI Artificial General Intelligence, which is still like a goal, which is very far away. And it exists basically only in science fiction. But since a lot of the debate mixes science fiction and AI realities. For some people, this robot rights thing seems to be a real issue.
Douglas: You're a professor or a lecturer in Artificial Intelligence right?
Alexa: Not really. No, not really. So I'm working at the University of Arts and Design in Halle in Germany, where we have the X Lab – a lab for AI and robotics. And I'm not a professor. I'm not even a lecturer. But what I try to do is to give actual lectures and workshops to our students to show them the like, the vast opportunity that AI techniques have for creative practices.
Douglas: When you're working with artists, what are you seeing that they're doing with the tools so far?
Alexa: So maybe to start with, the last years there has been a big trend of working with GANs. So basically what we are also doing in this Latent Riot work, and what I hope is that we're not tapping into this GANism, which is an invented word, which means basically, everybody is working with GANs, and they're just, you know, they feed a neural network with some stuff they found somewhere, and then they put the label on it and call it art, or they put the label on it, "It's made by an AI" and I find this… I mean, there is a lack of maybe some deeper thinking of what you are actually doing with it and also the GAN is just one tool in the whole universe, of machine learning tools and it's very tempting to use it because it's generative, and what it promises you is to create… that it creates images, invented artificial images that look like they came out of the data set. So it has some kind of magic to it. But I think there's so many other cool machine learning techniques.And also that the generative part is also just one aspect.
Alexa: So what I see in ways to use AI artistically is basically so you have the generative part, which is interesting, and I think it will be explored more in the next years, but then you also have the opportunity to do interaction work with it to make sense of sensor data and real time gestures. So this is a big (I think) area where it's useful, and the third is making sense of big data collection and finding pattern in it and maybe similarities which are too… for data sets which are too big for the human brain to capture. And I would like to see more work in this third part. For example there is an interesting work of a Dutch Collective called Aesthetics Of Exclusion, and they're interested in the aesthetics of gentrification. And what they do is they show people images of like streetscapes and the fronts of shops and stuff. And they ask them: "Does this look gentrified to you?" And it's an online tool. And they basically crowd sourcing the labeling task. And so what they build up is this classifier of what looks gentrified and what doesn't so later they can use this classifier to label, for example, new images from Google Street view with it. And I think this is a really interesting way of using AI techniques in the critical artworks.
Douglas: Excellent. You're gonna be in addition to showing the artwork at MozFest (the Latent Riot project) You're also gonna be having a workshop as well, right? Discussion, where we can actually ask more questions and things, right?
Alexa: It's thought it will be part input session and then part discussion because we're trying to open up the questions that we're tackling in this project. So one very pressing question is: How do we talk about AI? Is it okay to say: "An AI invented this, or the AI of Google…"? So is it okay to use language that basically gives this kind of agency to these AI systems? How we can avoid that, maybe or tweak our language so it's more accurate to the actual technical reality of AI. Yeah, there may be then even philosophical questions, like, maybe in the future, if we might have AGI, would it be okay to call AI entities to give them legal status? So there's a lot to explore. And we were also interested in, of course, what other people think. And to make this a discussion.
Douglas: Wonderful. Alexa Steinbruck of the Moving Target Collective. Thank you so much for your time. Really looking forward to seeing your work at MozFest and to talking with you further.
Alexa: Cool. Thank you.