transcript

So, can you please introduce yourself and what you do- occupation, study that kind of thing?

I'm Gian Sanghera-Warren. I'm a student. I'm a dancer. I'm a movement artist. My work, I think, mainly is about process and touches on kind of notions of memory, archivism, documentation, and kind of like tactile sensation. And then right now, at the moment, I'm using artificial intelligence software as collaborative tools in order to create a new choreographic practice that still touches on some of those themes, but in a new way.

Awesome. That's so exciting. So in terms of this new work that you're working on, it's called as - I'm aware – ‘The C shape of your hand like peanut butter on the roof of my mouth’?

Yeah. So that was- essentially this is kind of a yearlong research process. And as part of that, I made a work called ‘The C shape of your hand like peanut butter on the roof of my mouth’, as a kind of initial research for this project, if that makes sense. So that was kind of the first work I made that kind of existed in its own right, but also was a way for me to investigate these ideas. And then now I'm kind of starting again, but refining this process. And I'll make a new work, hopefully, that uses some of the same tools, but may be completely different. But yeah, that was a project that existed.

Awesome. Yeah, that's so cool. What made you kind of interested in using AI (Artificial Intelligence) for that? Is there something that drew you to it?

Yeah, so really I'm quite like a tech novice. I'm not at all particularly science-y, or know that much about machine learning. But what I did want to investigate was memory, and notions of personal and collective memory. And I read a book called ‘Pharmako AI’ by K Allado McDowell. And essentially this book is a collection of essays, prose, and almost like- touching on memoir. Written in collaboration with GPT-3, which is a language generating software. And essentially this book is written in chunks of the human author, and the GPT 3 responses, but reads like a fully formed book that could be written by one person. And I just thought it was incredible as a kind of like creative outlet and way of making. And then it kind of drew me to AI as something that would allow me to access this notion of personal and collective in a very immediate way. Because as a sole creator, and as an individual, there's only so many ways I can touch on this notion of collectivism, and it's kind of like a shortcut to that.

That's really exciting. I think a lot of people are seeing this kind of - everyone's calling the “AI revolution” - all this kind of increase in artificial intelligence, and the increase of the accessibility of that. A lot of people are very, very scared, and frightened of it. And I think there's a lot of fearmongering that perhaps has an input on that as well. But in in terms of using it as this sense of collective and this sense of collaboration, what was it that drew you – because you're talking about a kind of archival work and using that kind of memory and that experience, something that doesn't necessarily have that body, that mind as an archive. What drew you to artificial intelligence in that way? In that way of collaboration, and incorporating that kind of archive?

Yeah, so, there's a podcast by Holly Herndon and her collaborator - I forget his name [Matt Dryhurst] - called ‘Interdependence’. And in an episode- they talk a lot about kind of like tech, and AI, and creativity - Holly makes music, kind of like really experimental music, and uses AI and various other software within that process. And they talk about AI and use the metaphor of kind of, AI as a mirror, that it doesn't really exist in its own right. AI cannot create anything without us. But essentially what it does is hold up a mirror to us, to humans. It’s trained and developed on human data. And yes, that data can be biased, and that data can be not always positive, but it is true to an extent. And yeah, they have this kind of image of like, you can hold up a mirror to yourself and you can do with that what you please. So you can just stare at your reflection, or you can - I think they say you can practice a dance move in the mirror until it gets better. There's so many ways of using this technology. But to me, this concept of like mirroring humanness, mirroring behavior, and society, is interesting and also, in a movement context, kind of absurd. It's slightly ridiculous. That yeah, as you say, I'm investigating these, ‘embodied notions’, particularly for my last work that was all about sensation and memory, with a technology that has no ‘body’. And I think that kind of like, juxtaposition, it is interesting. I think the other thing is it’s a process that I'm attempting. It's not necessarily a process that is true, or happening, or successful; in that I can try and investigate these things with AI, and there are probably a million issues ethically, and conceptually, with that. But to me, some of that's okay, it's just an attempt.

Yeah, 100%. But you were saying about it being sort of sort of bodiless, sort of formless and how that interacts with dance, which is a very kind of form-based art form. What was it that drew you to something that is bodiless, and devoid of bodily form, in something that is very much a “human” kind of art form?

I think there's a few things. I think ultimately there's this idea in modern dance that's been around for a while, of this ‘body as archive’, that everything we do, and every experience we’ve had, kind of exists within us somewhere, and we can bring it to the surface. And I think that interests me in a way that the relationship between mind and memory, and then movement, exists, but isn't always so obvious. And then that relationship also exists within technology, or within our digital footprints -that everything we do is out there somewhere and could be recalled at any moment, but often isn't. I think there's a relationship there that's quite obvious, but maybe not tapped into that much.

And then also I think it's notions of, like, I consider my work queer, but not because it's necessarily about queerness, but because it's through a queer lens and because it practices queer as an ideology. And I think there's something about queerness and embodiment and how often it can be a queerest experience to escape that embodiment, and escape notions of form. And I suppose AI, or just kind of nonphysical software, and environments, are a way to access that too, is to kind of escape the trappings of human body.

In terms of queering though having that formlessness, being without body, or without ‘humanity’ and escaping from that, like you were saying; you see it as through a queer lens, as rather than kind of queer that has come to mean a political standing - I say come to mean, do you know what it? It always has been. But in a more abstract way of queering from a formless perspective, rather than queering from a “practical” perspective - that's not quite the right wording. But, with these kinds of accusations that are being thrown around, like you were saying, of Artificial Intelligence tools beginning to exhibit biases, and political or ethical, kind of social leanings, did you find that that was something you interacted with in your project, or something you wanted to avoid?

This is a good question. So I think on a basic level, part of the queerness of my work, or this work, is investigating ‘embodied notions’ with something that doesn't have a body. And to me, that's queer. But I also would acknowledge that this thing doesn't have a body but is trained on human data. So it does have the ideologies, the biases, the flawed aspects of humanity. It carries those too. And I wouldn't say I was trying to escape that. And perhaps, yeah, part of what is also queer to me, is kind of embracing that, that resistance. To me that's kind of camp. It's camp to use a machine to make dance about touching people. and I guess embracing that, to me, is kind of queer. But I wouldn't say that - as somebody who is very new to this technology - I am not the person to explain or philosophize about the ethics of AI particularly. And that was something that I, from kind of the beginning or when I began to research, got to a certain point and was like, I don't think I can do this in this project. I think that's for someone else. I know, yeah, there are definitely ethical –like- risks, but there are, probably with most new things; and I think we'll figure that out, but I'm not sure I'm the person to figure that out. So I think, yeah, my approach was kind of like: acknowledge, and maybe play with some of these issues, but also maintaining a slight distance to it, if that makes sense.

Yeah, maintaining like you were saying, kind of playing with it, that playfulness, I feel is also pretty inherently queer.

Yeah, for sure.

In terms of how you collaborated with it, what did you do ‘practically’, on a on a simple level, what did you use, how did you use it, and how does that compare to a ‘normal’ choreographic or creative process?

Yeah, I think every choreographic process is completely different. To me, this felt like a process I might have done anyway, to an extent; or a way in which that felt kind of comfortable for me to work with, but it's almost like adding an additional layer. So I think I could have made this work without AI. And a lot of actually - there, there were certain things that I tried that didn't work out, or there were certain things that was like, I'm incorporating AI when it would probably be much more efficient to do it without. But I guess it adds this, A) layer that I don't have access to as an individual, but B) lots of people talk about this software as a kind of ‘combat-tool’ for creative block. Specifically GPT-3, because it's a language software, can work where I'm writing a certain thing, and I can't figure out the direction that it should go in next, so it can help generate some possibilities for you. So, I kind of hope that maybe it will take me in directions that I wouldn't have done by myself. But essentially, in a practical way, I realized I wanted to use GPT-3 because I'd seen it being used, or heard about it being used, in writing, in song-writing, in these kind of other formats, but not in a dance context. I quite like working with language through choreography anyway, so it kind of felt like comfortable ground to begin in.

So I wrote essentially, with GPT 3 I wrote lots of things: lists, and poems, and prose, and I realized that I needed a specificity to what I was talking about. And that's where this idea of bodily sensations, these personal memories of touch and experience, and how they can be queered, or how they can be queer, kind of came to the surface. Again, for all these reasons that we've talked about, this kind of like juxtaposition of ‘embodiedness’ and lack of body.

So I wrote essentially like an essay, as a kind of section of prose about queering bodily sensations. With GPT-3, which is a language generating software, you provide an input, GPT-3 responds, kind of mirroring, reflecting the tone that you've provided. But you can also vary the temperature of the generations, which varies how kind of wacky and ‘out there’ the responses are. So, yeah, together, I guess, we wrote this text that talked about pins and needles; it talked about eating cherries; as these experiences that kind of have nothing to do with queerness, but to me can be queer. And I read this at a writer's group, some of the language I'd been generating. And it was like- I was like, ‘oh, I don't know, I'm not - I write lots - but I'm not, kind of like, “an author”’. But the response was that this spoke to an idea of queerness that they hadn't seen in literature elsewhere. And I think that's through the process of doing something with the machine and the kind of contradiction in that.

But with this text, I worked in a kind of language-based practice, which I often do anyway. We kind of dissected it, deconstructed it, ripped it up, took words from it, made new poetry from it, and generated movement from that. Which kind of created the language, the movement vocabulary, for the whole piece, and the chunks of movement that become repeated, and morphed, and changed. And throughout the process, GPT-3 was like a second opinion, or a collaborator, that I could go to to guide the process, but not dictate it. I would say it's still very much like my choreographic practice, but with a little assistance. So whether that's- for the end of the piece, we take the same movement that we've worked through, that accesses personal sensations, and then we turn it into something more erratic, more distressed. And I would go to GPT-3 and write lists of uncomfortable sensations and get GPT-3 to respond with more sensations. That I'd be like, ‘okay, delete - hate that one, like that one, that one's interesting’ - this kind of like, mining of information. I also used GPT-3 to generate guided meditations that we’d do throughout the process, as a way of accessing these sensations, and to develop a shared movement practice with my dancers.

The other software I used is Dall.E, which was kind of like ‘all the rage’. And through that, instead of generating image from nothing, I would take photos of the dancers, or parts of the dancers, in improvisation scores or in movement, and then get Dall.E to generate variations of these positions. And using the ‘out-painting’ tool to kind of create a context for these small bits of movement that didn't really exist, and then integrate these into performance. So it was a way of structuring the work, through these positions that felt choreographed by the software, as opposed to just me, but still in collaboration with the dancer’s own movement and bodies. Yeah, if that makes at all any sense.

It does! at least to me, yeah, it's so fascinating. I love it.

In terms of, you kind of mentioned, that there's moments where the AI tools became a bit of a kind of hindrance if anything, kind of you would have been maybe more efficient without it. And you also spoke about the kind of ownership over your choreographic process, as it is yours, in terms of the result product, or kind of what you created out of that, how much did you feel that your process was changed through using AI? Did it push you to different areas? Or did it not, did it hold you back? Is what was created different from your usual thing?

I think it's really hard to know. I think the work I created still very much speaks to my aesthetic. I don't think I used the software in a way that massively informed the nature of the movement. I think it's probably worth noting that AI has been used in dance, for many years or since its creation, but often in a way that - through my research, I found, and maybe this is me not looking hard enough - but predominantly it's used either in production, like using AI as systems of lighting, interacting with music, interacting with sound, interacting with visuals and that as a kind of like Artificial Intelligence. Or it is also used to generate movement. [Wayne] McGregor has done lots of things with motion-sensing the dancer’s movement, and then choreographing from that. And this isn't what I've done at all, and something that A) I probably don't have the facilities for, but also doesn't necessarily interest me. I wanted to create a work that still felt connected to humanness and felt that it could be made by a human. And I think I did. And for that, I then needed to incorporate a stage of translation in the process. And kind of through the work, I've realized that maybe this is actually what my work is about, it’s about translation. And that comes in that, anything that I'm using AI for is one thing; and to integrate that into a dance practice, takes work. So I generate text, but how I create movement from that text is very much my personal decision, or the dancer’s. There has to be this kind of like ‘in-between’ phase.

The same way with using Dall.E. Dall.E, or the function I was using, it wasn't creating movement, it was just changing images that I had, and then I chose how to incorporate that. And I think that meant that the work that was created, still very much felt like mine, or ours, like the dancers and I. It still had this kind of intimate sense. And therefore, I think potentially could have been made without any of the software. But there were definitely choices, or directions, that the AI generated that I potentially wouldn't have gone down.

And even in - the work I'm starting now is slightly different - but I think the uncontrollable nature of it, is what's interesting. In that when I do these generations working with GPT-3, it is this process of mining. And the responses I get, lots of them I don't like, I don't find them interesting. I also weirdly feel that although it's reflecting my tone, and potentially like this is a self-critical thing, I always feel that the generations, it wants to go more and more kind of ‘poetic’ and kind of like, this slightly pretentious direction, which is probably what I'm giving it. And sometimes I'm like, this is gross. Oh my God. Poetic nonsense. And then I have to be like, ‘no, I want you to be more practical. I want you to be more precise’. But it's also this negotiation of sometimes I'm like, OK these are two sentences I wouldn't write, and I think are a little a little too far out, but maybe that's where it needs to go, and kind of like negotiating with that. I’m not sure if any of that answered your question whatsoever.

No it did, it absolutely did.

Right, so to finish up, how do you feel like your project was received? In terms of your day-to-day person, how would they react? And then, kind of, people who are in that kind of ‘world of the arts’ and are perhaps critiquing your work, how were those interactions, how did they go?

I think like, as a dance artist, I'm lucky in the respect that this still feels like slightly untread territory, or less saturated maybe, than a visual art context. So I think the reception was that generally it was kind of, like, interesting. Maybe in five years’ time it won't be, and that's fine I'll change, I'll do something else. But I think, yeah we kind of touched on it earlier, but when you just say the words AI there's kind of like this alarm bell type thing. But I think within this practice, it's very clear that some of the risks that people are worried about are maybe not present in that I'm still working with human dancers. I'm still doing live performance. I'm not kind of replacing anybody with robots. And I think that's the fear, is that it's about like people losing jobs or losing kind of autonomy. But I think the autonomy is very much in the dancers, in my performance. And also, I'm not kind of shortcutting or stealing - well, I hope - like anybody's style or form, I think that that would be quite hard to do using AI in a dance. Well, not necessarily, but in the way that I'm using it, that's not present. So I think some of the kind of dangers aren't as easy to fall into. I think there's definitely a level of people not understanding AI, what it doe,s or what it's for, which is also fine. I wanted to make a work that still stood as a work that you could enjoy without knowing the mechanics, and hopefully that was true.

But I think my constant struggle with making performance is that I'm like 90% interested in the process and 10% interested in the output. Maybe less than that, maybe it's more like 70/30. But yeah, it it's really in the studio and on my laptop that I'm most excited. So how do I convey that in performance? It's really difficult, without everybody having to read my thousands of words that I have to write about it, which I would love, but I just don't think anybody would be that interested in. And I do think that some of the AI input helped me show that, in the kind of way that I use Dall.E to structure the work. I think even if you wouldn't watch the piece and know that this image was generated by a machine learning software, there was a kind of like, formatting of the work that felt like there was a connection to the formatting and structure of kind of language and image generating tools, I hope. Yeah. Does that answer the question?

Yeah, it does, that's great. You talked a bit about this research kind of continuing. Are you continuing to use AI or any similar kind of software tools? Is it something that you want to continue using in the future? Also, what are you doing next? What's going on?

Essentially, yeah, I'm continuing this practice. My research for the year is to try and find new choreographic practice that incorporates AI software. So I think of this performance as almost the draft or like, experiment number one. And I'm going to do experiment number two and hopefully refine the process that I built the last time and maybe go into some new directions or just use the same processes, but to create a new output. So, I'm just kind of going back into the studio now. I'm using also hopefully the same dancers, potentially with some more, that I used with the first piece. So I have this kind of like, similar process, similar collaborators, but hopefully the work will be different. And I think to accomplish that, I'm kind of broadening and changing the content of the work. So, the last piece was about, yeah, this kind of like archive of personal bodily sensations queered and made collective. This time I'm kind of opening it up slightly. A lot of the work last project was about identifying and accessing really specific sensations. So, I don't know, getting into a hot bath when you're cold or yeah, I talked before about kind of like pins and needles, or- these really like specific moments, and how they kind of have existed in the body, or we have concepts of them that can be drawn out. This time I'm more interested in kind of the macro of that. So I'm looking at bodies in - the whole body - in states of activity and passivity, which I think is ultimately like, all movement sits on a range, and all kind of like embodied beings, sit on a range between this active and passive state. Snd perhaps there's a middle ground or there's active ways of being passive, and passive ways of being active. Which I think again, I'm touching on notions that are very present in the, kind of ‘cannon’ of modern dance. The last time it was kind of like body as archive, which is still relevant, but this time it's like active passivity and passive activity. And I think these are two things that kind of like have been touched in contemporary dance lots of times. So I'm trying to provide like a new voice on that.

The way I'm accessing this at the moment is kind of strange and specific. I went to a gallery in Leeds a few weeks ago, and in this exhibition, there was a small box that had a, I think an oyster shell, or potentially a mussel shell, and in it with these tiny little pearls shaped like Buddha. And it explained that molluscs create pearl as a defense mechanism. So when grit or parasites enter their shell, they coat them, cover them in pearl, kind of how our bodies might create white blood cells, or kind of like absorb and destroy these diseases that it finds. So then what happens is if you place a specifically-shaped item into a mollusc shell, it will cover that item and that shape in pearl, and then reproduce that shape. So somebody, some scientists probably, had put in these tiny little Buddhas that had then become pearls shaped like Buddhas through this process. And to me, I was like ,this is both active passivity, and passive activity. It's this shell, this kind of like being, actively defending itself, but as a as a kind of byproduct of that it's creating this shape that it has no concept of. And that's kind of my entry point into this notion. So, I'll do the same process likely, I'll write about it with GPT-3. I'm kind of investigating the notion of the body as mollusc. Which we’ll access through, hopefully, guided meditations generated by the same software, and then using Dall.E again, as ways to kind of mutate and experiment with movement.

So yeah, that's what's next. And hopefully that would be a piece that yeah- I'm also looking at- kind of, I spoke about it before, but how AI I think has been used lots of times with dance in production. And that's not something I particularly did with the last piece, other than I used the dancer’s voices and played with using software to alter that. But I'm interested in how I can reintroduce AI after I've used the process to create, how can I reintroduce it to perform? And hopefully that will help me with this notion of conveying process in performance, so I'm hopefully going to work with kind of software like Stable Diffusion to create visuals that could hopefully be incorporated in performance.

Nice. That's exciting. I think that's pretty much it. Thanks so much.

You’re welcome!

Where can people find you and your work? Can people come see things? What do you want to promo? That kind of thing.

So hopefully the work will be performed in the Bonnie Bird Theatre at Laban sometime in May. Date’s unconfirmed, and it will be open to the public, so anybody could come along. And then potentially there's opportunities to perform it later. I think this is the practice that I'm going to continue in some form.

I'm setting up a collective called ‘Heavy Wear’ with my friend and collaborator who's in the work, Alex Bielaire. And we're hopefully going to work on kind of combining our two practices. So Alex works with visual arts a lot, and has this kind of like interdisciplinary approach to dance performance, and whose work is incredible but also very different from mine, much more kind of like emotionally charged. And hopefully we're going to kind of combine these processes and create work. We're looking to create a project based around the concept of plasticity as a kind of material thing, but also an ideology that interacts again with queerness, and kind of like world-building. So definitely keep tuned on that.

In terms of like my promo - I post things on Instagram occasionally. So ‘gian_sw’ is my @. And I also have an Instagram called ‘thebodyasanarchive’, which I've been kind of using as a bit of a sketchbook to document the process. I'm not particularly up to date with it at the moment, but that's also a little bit of an insight into at least the process of the last work. So yeah.

Nice. Awesome. Amazing.