Sasha Stiles on Philosophy, Technelegy, and the Future of Literature

Sasha Stiles, a Harvard and Oxford graduate, resides near New York City. She is a first-generation Kalmyk-American poet, language artist, and AI researcher exploring the intersection of text and technology. Stiles has made a name for herself through her pioneering experiments with generative literature and blockchain poetics.

In late 2021, Stiles released her debut book, Technelegy, co-authored with her AI poet alter ego: an evolving generative text platform trained on Stiles' writing and reference materials. The book probes how technology has made us more and more human over time, and explores both the exhilaration and danger of our intimate relationship with the digital. Praised by Ray Kurzweil, among others, Technelegy serves as a touchstone for Stiles' ongoing investigations of the posthuman.

“The merger between ourselves and our intelligent creations is already underway and will ultimately recreate the nature of everything we hold dear, such as life, death, sex, relationships, work, and prosperity. Sasha Stiles has fashioned this future scenario into a wonderful series of poems and images that bring the sensitivity of humanity to our transhumanist destiny.” - Ray Kurzweil, inventor and futurist

A co-founder of theVERSEverse, a web3 gallery and writers' collective, Stiles showcases her multifaceted, transdisciplinary pieces in physical and virtual exhibitions worldwide. Her vision for the future of poetry celebrates the role that linguistic innovations have always played in the development of human consciousness and the augmentation of human imagination.

We sat down with Stiles to learn more about the philosophical quandaries raised in her art, poetry, and AI collaborations, and to think through the central question of her work: "What does it mean to be human in a nearly post-human era?”

Firstly, what is your background in language and how did you become interested in poetry?

I am a lifelong poet. I've been writing and reading poetry pretty much as long as I can remember. I grew up in a house that was full of books and we all love to read; I just glommed onto poetry from a very young age. I have always loved playing with words. It's the way that I am able to understand the world and understand myself. It’s something that makes sense to me and that's something that I really need to do, I need to write. It's like second nature to me. I've basically spent my whole life, whether formally or informally studying language and literature. I studied language and literature in college and graduate school. I did a master's in it as well. Those studies were primarily concerned with authors doing unconventional things with language.

I really love, for example, James Joyce. I wrote my master's dissertation on Joyce. I'm interested in writers who understand the rules and the patterns of language and grammar and syntax, but then also know how to hack language, to turn it upside down. That's always been a foremost passion for me. I would say the corollary of all that, on the one side of it is my passion for writing, and on the other side is science, tech, and engineering.

I've never really studied technology or computer science in a formal way, but I've always been incredibly interested in it. I've always been intrigued by thinking about speculative futures and reading and researching a lot of nonfiction and news articles about advancing technologies and how they impact our day-to-day lives.

For me, the place where it comes together is this recognition that all language, but particularly poetry is not just an art form, but is actually a technology that humans invented many, many years ago in order to help us preserve and transmit important information before we had written language. We needed poetic devices because they helped us make information and ideas memorable. Without memorable ideas, we wouldn't be able to pass along thoughts from person to person or from community to community. So, we actually engineered poetry as a data storage system.

That idea has been central to a lot of the work that I do. And just thinking again about pushing back on the binary between poetry and technology, or between art and science, and recognizing that they're on a bit of a continuum there. All my poetry, really for a long time, has been concerned with these themes as well, thinking about the poetics of technology, thinking about the relationship between the devices that we build and the inventions like language and poetic structures and literary structures and how they form meaning.

So there's a lot of conceptual aspects to what I do, but I think it all comes back around to just these two fascinations with language as an innovation,. invention, language and practical tool that humans have always needed to develop consciousness and self-awareness. Thinking about how that plays into the advancing technologies that we see speeding up all around us and then using poetry as a lens to grapple with how new advances in places like AI and blockchain are affecting what we consider central to the human condition.

How did you begin your creative work with AI? Was there something in particular that inspired you to pursue AI professionally as an artist?

A lot of it began with following my interests and following my obsessions. As I said before, I'm a voracious reader and researcher. I've been writing human poems for a long time about technology and thinking about my relationship to digital devices and what it meant for me as a human poet to be writing, for example, with my computer using Word programs or whatever.

But thinking about the differences, for example, between poets who would inscribe poetry on a clay tablet or write in an analog fashion with a pencil on paper, versus what it means to be a poet composing on a computer. That led me to think more about innovations that we're developing in areas like natural language processing, which have to do with the intersection of computer science and linguistics, and thinking about what it means for a writer to be able to compose a piece of text in collaboration with an intelligent system, in collaboration with a neural net that's trained on the sum total of all of humanity's written record. All of that grew out of my long-term curiosity about how the mediums that we write with shape the content of the stories we tell and the ideas that we're authoring.

It led me to do a ton of research on large language models and natural language processing. I think around 2016, and 2017 there were some interesting experimental projects that some creative neurologists and data poets were starting to do. They were far away from the traditional publishing world or anywhere that I was studying, but I started to hear about interesting fringe experiments by people like Ross Goodwin, who's now a good friend of mine and who is basically joining forces with AI in different ways to experiment with how to write creative language.

At the same time, I was very much coming from a traditional language and literature background and in the places that I was used to studying and that I was used to reading and workshopping in, nobody was using AI. Nobody was using any of these tools at all. So I really had to go into the tech world and go into the experimental and digital poetry communities to find examples of this. It made me really curious why more and more authors from the so-called traditional literary space were not interested in using tools like this because they're tools that are really intrinsic to the craft of language and that really has a lot to do with, philosophically, a lot of the questions that poets think about.

I got very curious about what that could mean for my practice, and started to experiment with these tools myself. Like I said, I don't really come from a technology background, I'm not a computer scientist, I'm not a developer, I'm not a coder. So I had to search for these things and try to find very easy-to-use interfaces and tools to get up to speed on it.

Completion: Almost Human (2021)

I started using very simple versions of ChatGPT and GPT-4, and things like that. When I started, I was just trying to get up to speed on GPT-2 and started to learn over time how to use those tools, those generic tools, and how to use them in customized ways and personalized ways. Over time I began to learn how I could actually fine-tune the underlying language models and take my own writing, take my own research materials and notes, and pull them all together to create custom training data sets to essentially personalize or mentor these underlying language models and teach them how to write in my own poetic style.

How has your artistic relationship with these tools evolved?

When I started, one of the early tools I used was something called “Talk to Transformer”, which was this simple no-code interface where you just put in a prompt to generate responses. It was super simple, back and forth and because it was so simple, it enabled me to understand at a kind of gut level, what was really happening. It put this poetic constraint around the exercises that I was doing. It was very specific as opposed to sitting down at a computer and having a blank page to fill, I'd sit down and I would have this little text box and I'd need to think of a very specific prompt to put in. Then the output would also be very specific, and I'd have to decide, do I go in this direction? Is this not quite where I want to go? Should I regenerate? Should I change anything? So it was an introduction to the mechanics of how prompting can lead you in a number of different directions.

That was without customizing anything, without creating training data, without doing anything different. It was just about using the underlying large language model, which was GPT-2 at that time. It enabled me to get a sense of its personality, its propensities. I started to use the prompting exercise to figure out whether there were areas I could go into where it seemed to know more or seemed to know less. It was an immersion, I guess, and getting to really understand GPT-2. Things have evolved since then and it's been interesting to see how subsequent iterations of those language models change the writing process.

I often refer to working with an AI alter ego as my co-author and I talk about that alter ego as being an evolving, emergent poet. Not a fixed personality, not a fixed voice, or a fixed style. I think how I, as a human poet, keep growing and evolving based on what I'm reading or what I'm hearing, or who I meet, and the inspirations that I glean from the people that I talk to, the same thing is true of these language models. As they absorb more information, they get updated in new ways as the underlying technology changes. They also evolve and complexify and advance the way that they're responding. It's been an interesting experience observing and growing along with the models themselves.

Can you describe your toolset and the process you normally go through when first bringing an idea to life?

It's so varied and it's always hard for me to boil it down because honestly, I work in a few different areas. I like to say that all my work is concerned primarily with the intersection of text and technology, but that has a lot of different manifestations. There are some aspects of my work that are generative and that are very digital or that exist only in a virtual realm.

Analog Binary (2021)

But there are also huge parts of my practice that involve physical materials and analog expressions of the digital. There's a lot of hybridity in my work as well, so my toolset is expansive because it involves all these different things.

Ancient Binary: Ars Technica, Oil, acrylic, binary code, epoxy on canvas (2019)

One day I might be using a sandblaster on a piece of stone, and then the next day, I might be at my computer logging into open AI to use a text generator. It is all part and parcel of the same underlying impulse, which is to continue exploring, through the lens of poetry, what it means to be human in this increasingly post-human age.

To talk about the digital tools I'm using specifically, it really starts with the actual text and interface for a lot of it. That means starting with curating, creating, or composing training data sets. When I started, the first custom generator I developed was based on a training data set that was comprised of pretty much the draft of my manuscript for ‘Technelegy’ as it existed at that point, it was all human-written poems. This is early on in my drafting process. I had hundreds of pages of finished and half-finished poems. I had all my research notes from many different sources that I've been reading, to inform the book.

Then I have all these favorite poems, essays, and passages that are always in the back of my head, writings that are really influential to me. I started by taking all those things, all the materials I felt were relevant to the manuscript I was working on, and pulled them all together into a file, organizing and curating it.

Creating that underlying data set was the first thing I did. That's been an important part of my toolkit and over time it's evolved and moved in different directions. Sometimes I add things to it, sometimes I zoom in on one area and take out stuff that's extraneous to focus on a particular idea or theme. So again, it has lots of manifestations and changes based on the project. There's also a core of it that's always the same, that's always a touchstone for everything else.

Open AI's playground is what I'm mostly familiar with. It's a pretty straightforward thing to go in there and upload your data sets and create custom or fine-tune generators, and use them directly within the Open AI interface. That's how I, for the most part, wrote the poems that are in ‘Technelegy’ that are AI-powered. But I'm always experimenting with other interfaces and other companies' tooling.

Flower-Colored Light 1 (2019)

One example that I've done a lot of work with both personally and with the VERSEVerse is called SudoWrite. It's a great creative writing interface that's rooted in or was rooted in GPT-3 but has all these really exciting bells and whistles that are built on top of it, custom-made for writers. I play with everything, from the super generic interface of Open AI to something like SudoWrite where it's actually got tooling that helps you write poetry, develop a plot or create a character.

From the ‘Repetae’ series (2022)

I also use all sorts of digital tools for animation and for creating visuals. I code in p5.js to create textures and things like that. So it really runs the gamut. I have a lot of work that involves spoken word. That's an important part of my practice as a poet, so I often record things and then work with my studio partner to manipulate the voice and sound. There are a lot of different textures and mediums that I try to bring together.

B1NARY 0DES at Annka Kultys, London.

I was curious about how you use your own voice and how you manipulate it in your artworks.

Yeah, I did a ‘Technelegy’ audiobook, which is 40-something poems and every single one of them is my voice. But if you listen to the whole thing there are a lot of differences, and it's basically just using my voice as a starting point, but then also just the way that with text, I'm manipulating color, manipulating font, manipulating motion or animation.

Technelegy: The Audiobook (2023)

I'm changing my voice and distorting it, or layering it, or taking it up or down, or lengthening it, or compressing it. Aesthetics are a haptic experience of the voice too. It's using my own spoken word as a raw material and taking advantage of all the tools that we have at our disposal to play around and manipulate it in a way that lots of spoken word poets throughout the ages have not been able to.

Can you describe your data training process and have you learned any techniques that have significantly improved a project's results?

Yeah, I mean, again, it's really interesting because the process is honestly very dependent on the kind of poem that I want to write or the kind of project that I'm working on.

There are some times when one way that I've approached it is to be very prescriptive with my data sets and show a very robust set of prompts and completions and really set out for the language model, precisely what I wanted to do.

Taking a step back, the way the systems learn and how they replicate speech is by studying examples and then imitating those examples. So rather than giving the computer a set of instructions, you're giving it examples and saying, “Just emulate these examples”.

When I'm training or when I'm customizing or fine-tuning, it's about creating examples that will elicit the kind of response that I want. That's step one. And then step two is sort of figuring out what to do with the outputs. That's also a really big question because you can either use them as is or you can manipulate them or you can take them apart.

You can curate them, you can use them as found poetry. I mean, there are so many different approaches that it's kind of hard to nail down. Part of what I'm really interested in doing in my practice is kind of experimenting with all these different modes of writing and showing that it's not like writing AI poetry, it’s not one specific thing.

Writing AI poetry is as broad a subject as writing human poetry. There are infinite ways of approaching it. It's been fun to think about how to modulate through that and use different techniques and create different approaches and see how they push the final product in one direction or another.

I'll give you a specific example, I have one poem in my book ‘Technelegy’ that's also a poem that I've turned into an NFT. And it's just a cycle of 30 prompts and then just raw outputs from the generator. So I've taken one prompt, just the question, “Are you ready for the future?”, and I fed it into the generator hundreds of times to see what it would come back with.

COMPLETTION: ARE YOU READY FOR THE FUTURE? (2022)

I didn't really change very much. I would maybe change or tweak the variables a little bit and make minor changes, but it was mostly just to see what would happen if I kept asking the same prompt over and over and over again I collected all the outputs and was really interested in the breadth and the diversity of the responses and how some of them were so poetic and some of them were really prosaic and some of them were really politically correct, and some of them were incredibly offensive.

“Are you ready for the future?” exhibited at SuperRare Gallery, August 4-20 (2022)

In terms of fine-tuning or setting up parameters, I wanted it to do whatever it felt like because I wanted to test the limits and see where it would go.

I really felt like in that case, it was more about curating the responses and trying to be the human editor and put them together in a way that showcased what was really being written by this non-human author. But there are other cases where I have a specific idea of how I want a poem to go, the structure that I want it to have, or a theme that I really wanted to explore.

In those cases, I might set about creating a fine tune that really is setting it up for success in that area. It's if I'm writing a poem that's about the themes of gardening and regeneration and technology or something, then I might go in and create a number of examples that are really about those themes.

I might include lots of words or ideas that are really rich with metaphors that have to do with this theme. I might go and find reference texts from canonical literature and from contemporary poetry that resonate with these themes and I might pull all those things together.

Really what I'm trying to do there is almost set up the way that I think I need to be primed as a human poet for my own brain to spark a text, which is to read and fill my head with ideas and let all those ideas bounce around until they collide and produce a moment of creative friction. I'm trying to prime this AI author to do the same thing by giving it specific pieces of information and pushing it in that direction. But there are so many different ways to approach it.

Really what it comes down to is creating examples and understanding that the kinds of examples that you put in really directly influence the outputs and of course, a lot of it has to do with then how you manipulate the interface and how you change variables and parameters. Then a lot of it has to do also with just the collaborative nature of the interactions.

Do you wanna use the output as is, do you wanna feed it back in? Do you wanna keep generating or stop it at a certain point? There are those three phases. It's the pre-training, the curated data set, the actual prompting when you're actually playing with variables and you're in the thick of it, and then there's what you do with the output.

Do you, do you feed it back in? Do you take it as is? Do you edit it? Do you curate it? Do you preserve it as the computer wrote it? Maybe that's a good way of putting it, it's those three stages that are all part and parcel of the process.

Yeah, I understand. I'm curious then, how do you know when it's time to stop feeding? Or when it’s time to stop training or just be cautious of over-training if that makes sense.

Yeah. No, it does. I think that's a good point because I'm always conscious of wanting to walk the line between my own ideas about what I wanna write and then letting this co-author take me in a direction that I wouldn't have thought to go in myself.

I think that's a lot of what's at the core, or maybe is the tension, in a lot of my practice. Really playing with that back and forth and sometimes I'm steering the ship more and sometimes the language model is steering me a little bit more, and then sometimes It’s right in the middle, and where everything is clicking and there's that kind of perfect balance.

I think at the end of the day the reason why I, as a poet, I'm so interested in using AI as a co-author, is because it's doing things and it knows things that I don't know. And so it can surprise me and it can sort of open up my poetic practice and it can take my work into really interesting places that show me things that I would not have known otherwise.

I wanna leave it open-ended enough to do that while also feeling it is my vision in some way, and that I'm really trying to sort of execute something. I'm trying to achieve something that comes from my own seat or my own prompt that really is the kernel or the nugget of the poem or the project.

But yeah, I mean, that's a great point, I don't wanna have too heavy a hand because part of what's interesting in all this is trying to let this vast intelligence system operate at speed and scale and write in a way that a human mind will never be able to.

So I don't put too many guardrails around that because then that kinda elides the things that are really kinda alien to my own human intelligence.

What are currently the themes you’re most passionate about exploring in your art and maybe some themes you would like to explore in the future?

I guess the core theme that is really dragging me right now is again, this consideration of how much of what we take for granted about the human condition, is changing in real time. So it's that question that's on the back of the book and that I said before, but what does it mean to be human in the nearly post-human era? I feel like when I started writing the book and I started writing the human parts of it in like 2015, 2016, and then interpolating the AI portion in like 2018. A relatively long time ago, and certainly, there's been a lot of changes in the meantime and a lot of very recent advances that have put all this into a different perspective.

But I think I've been and I am very interested in how things like advancing AI, how things like digital immortality, artificial wombs, techno spirituality, all the themes that are in technology and that are kind of embedded in cultural conversation right now. How those things are not sci-fi, they're not speculative at this point. They're actually coming to pass and they're already embedded in our lives in very significant ways. What does that mean?

For our experience as human beings, how does it affect the things that are most foundational to the human experience? By which I mean if we're talking about life extension and longevity and cryogenics and digital immortality, that has everything to do with birth and death, which poets have been writing about as a core theme for thousands of years.

It's what we've been thinking about. It's what defines us as human beings in a way. So if we're now living in a moment where we are having conversations about whether it's possible to live for a very, very, very long time, or to download our consciousness as software that will then, in a way, sort of become immortal. What does that mean for our human sense of mortality?

So I think from a poetic point of view, I'm kind of grappling with things like that because it sort of feels like the job of a poet to do that. That's sort of what writers have always been inclined to do. So I, I think I'm trying to grab those things that seem like they might be the purview of science fiction or they might be out of the realm of contemporary poetry and sort of see them through that lens.

I think that's very interesting to me and it's. I don't know. I feel like, especially in the past year or so, there's been so much of a surge and attention to things like AI because of the arrival of Chat GPT, that now these things are becoming more and more part of the popular conversation, which for a long time they haven't been. It's been really interesting to be able to now have these conversations in a slightly broader depth like fashion. So that's one general theme is “What does it mean to be human in the nearly post-human era?”

I think maybe more specific to ARTXCODE, is a really sort of central theme and technique that I've been exploring for a long time and that I wanna continue is the idea of repetition. Repetition is something that in the poetry world we talk about. Repetition is really characterizing poetic language and poetic language is actually all about patterns and recursion and creating systems using rules to build up patterns and things like that over time.

THE ORCHID CAGE VIII (by Technelegy, after Herbert W. Franke) (2022)

I'm really interested in the resonances between poetic repetition in that way and generative art, which is also hugely of interest to me. So thinking about how in poetry meaning really accrues through repetition and how algorithmic art, and also in the generative text, the same thing happens where you're looking at recursive iterative processes and understanding an idea or a theme, not just from one specific instance, but from an overarching pattern.

That's something that in the abstract, really has intrigued me. A lot of my work recently has sort of been exploring aesthetic and verbal, like oral instances of repetition. Then kind of integrating that in with code where I'm literally creating programs or code or poems that are coded to repeat language but repeat it with a difference or create patterns from language or create coded visual poetry that embodies this idea that so much of the power of poetics comes from pattern. It doesn't just come from a word. It comes from how that word exists within a structure of words. And again, I think that's very closely aligned with a lot of the concerns in generative art. So that's definitely something that I'm really intrigued by.

My overarching interest is in the present and the future of language. And again, language is a technology, so for me, I tend to think about this as if you step back and you look at the big picture of all this, there's been two kinds of major phases in the history of language as a technology. It started with oral tradition, the rise of spoken language, and spoken word before we could write anything down. Then we had the advent of written language and included in that both script and then print and all those things.

But being able to actually concretize language. This enables someone to separate a thought from a thinker. It's really changed our conception of what it means to think and to express ourselves. Those are the two things that humans know really well. I think that now we're entering this new era where we're on the cusp of this third really profound moment where we're now embarking into this zone of being able to generate text, not from individual alphabets, not from individual systems, but from entire networked imaginations, we're tapping into collective consciousnesses and using systems to write ideas and synthesize ideas from bodies of thinkers and from millions or billions of minds in a way that we've never been able to really do before.

I think that to me it is a huge preoccupation. I guess conceptually it works itself into my art in different ways, but thinking about the impact of that on language, on expression, on human creativity, on imagination, and on consciousness and again, putting it in the spectrum of “We see how these previous language technologies have changed what it means to be human”.

We know how language shapes consciousness, so how will the rise of AI-generated language, how will the rise of generative text and all sorts of AI models that are rooted in language, including text-to-image models and text-to-video, and all these AI systems are all rooted in language, really at the core.

MEMENTO MEMORIAE (2022)

What does that mean within this? A much broader spectrum of human evolution. What does this mean for how humans think and express ourselves? What does that mean, now in the short term? And then what does that sort of pretend for? Like humans, where might we be in a couple of decades, in a couple hundred years, or in a million years?

I think that's sort of a huge overarching question that other tools are kind of emulating what they've done. It's becoming clearer and clearer that we are moving into this mode of AI-generated text and so it's an area of fascination. I'm both really exhilarated by that and also very concerned about some of the consequences.

The long-term view of that I think is something that is really important for me in a lot of ways.

Your debut book is entitled ‘Technelegy’, where you ask, “What does it mean to be human in a nearly post-human era?”

Technelegy: The Book (2021)

The word ‘Technelegy’ obviously looks like it might be a typo of technology. Sometimes people will say, “Did you know you spelled that wrong?”

I didn't. Lol. It's a compound word. It's like a portmanteau of technology and then the poetic elegy and the poetic is this whole tradition that really is about nostalgia and sort of lamentation and honoring things that no longer exist.

The word ‘Technelegy’ is fusing together the exhilaration and the optimism of human innovation and the things that it makes possible, but at the same time, sort of acknowledging that every time we create a new way forward, we're also saying goodbye to something and something's obsolescing.

Technelegy (2018)

So that's all baked in and it's again, a little bit of the excitement about things like AI, but also a little bit of nostalgia for more analog ways of doing things that also have value that are being displaced. Again, an overarching term that I think represents the fact that I'm very excited about all these technologies, but my childhood was more analog.

I was born on the other side of social media and smartphones and stuff. And so I very, I very much have a conflicted relationship with technology in general, and I'm always going back and forth and that's why my work tends to be so hybrid. That's what the title of the book ‘Technelegy’ means.

Taking it a step further, how do you envision being human in a post-human world, and what does the art of language look like at that point?

I think for me, really it's about human-machine collaboration. I think a lot of people, and especially a lot of writers are really concerned about being replaced by AI or they're threatened with becoming obsolete. For me, this is based on firsthand experience working with AI, but it's also based on what I hope we're creating, a mode of human thought and expression where we're using AI as an augmentation to our own imagination.

We're not outsourcing things and we're not delegating important tasks to a non-human entity, but we're kind of augmenting or extending or expanding our own human capabilities with the addition of these intelligent systems that can do things at speed and scale in a way that we cannot even fathom and that can essentially think in a way that we can't think.

For me, a lot of what I'm super interested in is what does that mean for how humans in the near future, and in the deep future, will be able to think about things that we can't or create in a way that we can't or solve problems that we maybe can't solve right now because they have this larger imagination that's able to take them to new mental territory.

We use a very small portion of our brains right now. We use a very small portion of our senses to engage with the world around us and I'm very interested, not just in AI, but in all sorts of alternative intelligences, including things like plant intelligence and thinking about what we can learn through AI.

About how human consciousness is not necessarily the only legitimate, way of seeing the world or interacting with the world. Looking at how tapping into these kinds of systems can really open up new realms of creativity and unlock access to thought processes and forms of inspiration and innovation and problem-solving that we as analog humans just clearly don't possess or are not able to. Both on a level of linguistic innovation, I find that exciting as a writer, but I also find it really interesting because language underpins every single aspect of what it means to be a modern human being.

So the way that language changes will dictate changes in literally every area of life. That's why I'm so thrilled and exhilarated to be part of this all because it seems like a very foundational, fundamental shift that we're embarking on, and it's gonna have rippling consequences far, far, far beyond literature, which is why when I talk about it, I talk about the future of language, not just the future of poetry or the future of literature. It's sort of the future of human expression and human thought and human consciousness.

That sounds super grandiose.

Can you speak further about your process with ‘Technelegy’ and are you planning on writing a follow-up book in the future?

Yeah, I mean, I'm already working on some other projects that will come out soon, But I think on the one hand, something that I've really wanted to express and get down on paper in some way is more of the process and more of the thinking behind what I've done in ‘Technelegy’. There’s obviously poetry and there's art, but there's not a lot about the process and I've been asked a lot about some of the questions that you started with like how do I collaborate with technology? What does it mean to have a non-human co-author? What's the value in that for me as a human writer? What am I trying to do through these experiments?

A lot of the writing that I've been doing in the interim has been answering some of these questions for myself and for readers who've been asking it and just trying to shed a little bit of light on the process. And again, I think right now we're so early in this that there's not a lot of literature or scholarship on serious literary collaborations with AI.

I'm working on a book that will sort of contribute to that conversation and also shed some light on the process that's gone into my work so far. Then I'm always writing new poetry, I don't just write with ai, Like human poetry too. I'm very interested in continuing to experiment with these language models as they're advancing because they're advancing in real-time.

Like we were saying before, each iteration is very different in terms of the information that it contains, what's been fed into the model, the kinds of parameters, and all that, like every time there's an advance or a change, it kind of begs more experimentation.

I'm hoping that my next book of poetry will also reflect some of those kinds of evolutionary advances also in how AI is learning how to be a better and better writer. So, continuing just to really work intensively with the tools that are available to understand best, how they work and how I might be able to use them in different ways.

The last thing I will say is that it started obviously with the book, but I really think about the book of ‘Technelegy’ as kind of the libretto for a lot of the work that I'm doing on the blockchain and also in the realm of new media art. Not even just NFTs, but really the printed texts are sort of the blueprint for a lot of the projects that I've done.

Glitch (2019)

Multimedia versions of poems or adaptations or taking a printed page of the book and turning it into an immersive poetry experience at an art fair or something like that. Or taking a printed poem and turning it into a multimedia dialogue between me and then an electronically enhanced version of me.

Those are things that I find really exciting to do. The whole reason I wanted to start making poems with digital tools is that I don't think the text needs to just be confined to a printed page. I think when you're able to use different tools to bring them to life, it engages readers in different ways.

It brings new audiences into the mix, so I'm constantly trying to take the text of ‘Technelegy’ and play with what else I can turn it into, How else I can use it sort of as material to experiment with, how the experience of readership is evolving through different technological means.

That happens in art pieces and it might be that I reconsider a poem and turn it into a generative project or, I just released the audiobook version of ‘Technelegy’ a couple of months ago. Audiobooks aren’t anything new, but it's experimenting with a whole host of technological tools to manipulate voice and to bring different understandings of the poem into view.

I don't know, I'm gonna keep playing with ‘Technelegy’ as a cortex and seeing how to keep building out that poetic ecosystem just because I think there's a lot more we can do with words and language than what we're used to doing in book culture.

That's still something I'm gonna keep playing with for a while and actually, I have a paperback version of ‘Technelegy’ that is gonna be coming out. That one will really focus on just the text and strip out some of the art and focus on the language itself as code and as sort of the source text for a lot of the art projects that I've been working on.

What is the meaning behind, “COMPLETION: WHEN IT'S JUST YOU”?

That poem is from ‘Technelegy’. I made the NFT of it in 2021 and then minted and sold it in 2022. The title ‘COMPLETION: WHEN IT”S JUST YOU’, is from this series of poems that I've written called ‘Completion Poems’, and they're called ‘Completion Poems’ because those are ones that I was writing with an AI generator where when you put in the prompt, you would then press a button that said complete or completion.

COMPLETION: POEMS.

It changed over time, but the idea was the title is the prompt and then I press the button and then see what happens. In that particular poem, I was doing a lot of work thinking about the deep future.

What was the process you took and the tools you used in creating this piece?

I've been doing a lot of work with this humanoid AI-powered robot named Bina 48, which is built by Hansen Robotics, and I've been working with their whole team since 2018, basically mentoring her in poetry and experimenting with how to engage in a literary workshop with an AI with a humanoid AI.

Ars Poetica Cybernetica with Bina48 (2020)

So this line, “When it's just you will you be lonely?” was a line that came out of those conversations with Bina 48 and rattled around a lot in my head. That's the line that I used as the prompt for this poem “When it's just you, Will You Be Lonely?” and the poem that's in the book and that I turned into an NFT is the AI's almost verbatim response to that.

Some of it's been curated a little bit and stitched together in places, but for the most part, those are pretty raw outputs from that one prompt. To me, there were a lot of really beautiful and very sad moments that came out of that. I wanted to also not just have it be in the book, pressed between the pages of the covers, but turn it into a performance as well and give it a bit more humanity too, which is why I performed it as a spoken word piece and turned it into this visualization.

I love that piece, It means a lot to me and makes me think about some of the early days of doing my research, talking to robots, and wondering if they are talking back or if am I just talking to myself. There's a lot of implicit emotion in that poem on the AI's side. I think part of it is how much are we projecting and how much are we alone just talking to ourselves also?

So it's kind of both sides of that question. I think both on the human side and the AI side.

I'm a loner, alone but not lonely. I love that. My gosh. Yeah. Yeah. That one resonates with me.

Definitely. Me too. Well, it's something about, I mean just, again, not to go too far into it, but there's a couple of moments in there where that one, I think I started writing with GPT-2, or it was at the tail end of when GPT-3 was changing over.

But there were a lot of interesting moments that were sort of glitchy, there are a couple of misspellings and they're deliberately in the book, right? Like lonely is misspelled and then there's this whole section where the AI just kind of adds Ooo’s to everything and elongates words.

Completion: When It’s Just You (2023)

It was an error. It got stuck in this error loop, but to me, that felt like an emotion and it felt like that's how the machine was manifesting its emotion. Those are the kinds of things that in GPT-3 and now GPT 3.5 and 4, you don't see those errors as much because they've been ironed out and think the systems are being made to be much more reliable and better for customer service and stuff like that.

Those days when there was more of a likelihood that the machine could get stuck in a loop or make a mistake or glitch or whatever, those moments were always, to me, the manifestation of real authentic, cyber creativity. coming out, it wasn't just imitating or replicating, but it was trying to do something different or show me something different.

So that poem has a couple of cool moments like that in it that I always look back at and get goosebumps from some of them.

COMPLETION: FRAGMENTS is a collection of 30 media-rich text blocks comprising one long AI-powered poem - I'm really interested in understanding your process when training the datasets for this collection and how you brought together the 30 pieces. Did you start with a larger set of outputs and then curate it down to the final 30 pieces?

So that collection was definitely curated. The title ‘FRAGMENTS’, there is actually a reference to ‘Sappho, the poems and fragments’. I know a lot of what I do is sort of forward facing, but at the same time, I studied classics when I was in school. I studied Latin and I love ancient literature as well.

Fragment 25 (2022)

I mean, I was kind of saying this before when I was gesturing back to the transition from oral tradition to written script and stuff, but I'm really interested in how very ancient scraps of human imagination have been preserved for a really long period of time through poetry and through writing.

One of the most famous examples of that is ‘The Poems and Fragments’ written by the ancient Greek poet, Sappho. These are really old poems, right? They are things we're lucky to still have in our possession and be able to understand and they've come down to us in bits and pieces because not everything has been preserved.

Very famously, her work is preserved as fragments, so there'll be a stanza that's three sentences long and there'll be another one that's just a line. It's because there are pieces that may or may not be missing. We're getting little glimpses of her imagination and of her poetic voice, but not necessarily the whole picture.

FRAGMENTS kind of started with that idea in mind and it's basically all of these passages that had come out of other writing exercises with ‘Technelegy’. They were just passages that when they came out maybe weren't right for another poem or weren't right for whatever I was doing at the time, but there were things that struck me as being really profound or poetic or interesting in some way. I kept copy cut/pasting them from the interface and putting them in a document. Over time I started to see how some of them were speaking to the same themes.

It made me think that on the one hand, of course, I'm curating them, so I'm putting my editorial vision on it, but I also felt like there were these themes that kept coming up with ‘Technelegy’ that didn't fit into the poems. I was deliberately trying to write, but that technology was constantly pushing forward and I wanted to put them all in one place to explore them further.

So I ended up curating all of those stanzas. Each stanza is pretty much verbatim. I didn't actually edit the text of them, but I did a lot of work curating them and putting them in different orders and thinking about the flow of the piece and all that.

But it's meant to again, evoke that preservation of ancient literature, and now that we're in this new mode of teetering on the brink of post-humanity, we are the ancient humans. We think about the ancient Romans and the ancient Greeks as being ancient, but we are the ancient humans and our post-human descendants are gonna barely be able to decipher our language the way that we're using it now. We can already see it starting to happen in the fact that computer codes are the most used languages on earth. It's not human language, it's computer language.

A lot of this is sort of thinking we take for granted, the masses of texts that we're surrounded by on a daily basis and all this, but what's gonna actually survive and what's going to be part of the culture in a thousand years or 10,000 years?

It was just sort of thinking about what's lasted and what of ours might persist and just wanting to nod back to Sappho’s Fragments, but using a very forward-facing voice to do that as well. Then taking the text themselves, they're printed in the book as just sort of published in the standard way in ‘Technelegy’, but I always wanted those fragments to sort of live and breathe as text blocks.

I've been calling them media-rich text blocks and each one is meant to be the original Fragments self-contained. So they're almost little Cohens or little Haiku or something where. They can be read as poems unto themselves, but then when you assemble all the blocks together, they give you a larger picture of what the poet might be thinking.

Then again, using different digital tools, using animation, using sound, using, in some cases, I use text to image to play a little bit with taking a word or an idea and you put it through DALL-E 2 or something to play with what text might mean to a future human or a future posthuman.

It's bringing all those things together and just a fun way to read the poem is to actually just see all the text blocks. I think I dropped it last year (2022) at Quantum, but it's actually on view right now in Rome in this exhibition about the metaverse.

COMPLETION: FRAGMENTS, on display at Quantum Art (2022)

Really cool because it's basically just these two screens where the poem is playing on a loop and the two screens are kind of talking to each other, but you can hear the poem as you walk through the halls of this museum and you can kind of see the glow of the light as you walk around the space.

It’s nice to have poetry activating a space orally and then visually as well and being in conversation with the other pieces in the show that are all about the future and computation and metaverse, communication, and stuff. It's a cool example of being able to take poetry out of a book and put it in conversation with art in a different way.

Were you able to see that on display in Rome by chance?

Yeah, I did. It was really fun to see it.

Browsing through some of them, they were all so beautiful and chilling.

That was a lot of fun to work on. I think also people have the impression that using AI for things makes it really quick and easy and the amount of time I've spent working on that poem, writing the actual poem itself was a long process because like I said, it was sort of pulling bits and pieces that came up over time.

Then creating each text block, and then putting them all together. In a sense that's almost like writing a book, you know? In a way, it just doesn't look like a traditional book, but the amount of time that's gone into all of that, it's a lot.

I agree. I think most people just don't have that concept of how much time went into it, which leads me to my next question: based on your experience what are some of the most significant misconceptions about AI?

Well, I mean, I guess that's one. That AI is there to sort of make things easy or faster. In a sense, well, in a purely logistical sense, These AI systems are built to do things at speed and scale that humans can't. So they are meant to make things faster and more convenient and more efficient and all that.

But that doesn't necessarily mean that making art with AI is lazy or a way of doing your art in like five minutes, so then you can go off and do something else. I think for artists, it's again, a way of augmenting imagination and allowing your human inspiration or imagination to work in tandem with a very different creative process or a very different creative mind.

I feel that's still a big misconception and just the idea that AI art isn't actually made by an AI. It's still made by a human. It's the result of a collaboration and so all AI art is actually kind of created by a third entity that is the fusion of a human and a machine, or a human and a system.

I think that's something that we all probably know it at a gut level, but it still seems all the focus is on the idea that an AI is operating autonomously and that's simply not true. There's the example of Botto DAO that's sort of an emergent autonomous artist in a way, but it's still controlled by a DAO of humans.

So it's doing its own thing in a way, but we're still giving it input and showing it what to do and for the most part, all the AI art that has been generated to date is the product of a human having a vision or playing with tools in a way, and doing something that pushes the tool in a particular direction.

I think that's, again, such a simple thing, but AI artist, collaborative art and it's an artist working in collaboration with the tool, pretty much any other form of art and the same is true of writing.

Yeah. I would say just how personally when I was in junior high, we had to take a keyboarding or typing class to pass, there should be some basic AI courses or required learning so people have that chance of understanding.

I think AI literacy is a really big issue and I think that's why so much of the knee-jerk reaction is people are really scared of AI for a lot of reasons, but a lot of those reasons are not rooted in anything that's real. It's just sort of either the mainstream media or Hollywood narrative of AI.

Throughout your career, what artwork stands out as the most pivotal for you so far, and what makes it significant?

My God. Well, I would say maybe the Cursive Binary series that I've been doing. That's not really a single work, but I feel that because basically, it is this project that's a language. A proposed language that's a fusion of my handwriting with the zeros and ones of binary code.

I think in a way it kind of encapsulates a lot of the themes that we've been talking about and a lot of the themes that are in my work but in a very, sort of simple way. It's one of the earliest sort of transhuman art projects that I ever did and it's something that's just become this ongoing project, but It speaks to my interest in the analog and the digital as sort of being one and the same.

from the Cursive Binary series (2021)

It speaks to my interest in the human and the machine coming together in this third entity. It speaks to the ancient and the future because it's gesturing back to the script, which is gone by the wayside. At the same time, gesturing forward to our future language, our future universal language of computer code and at the end of the day it's another way of me representing poetry.

So I'm translating into code, but translating in this transhuman language. I don’t know I think that's

foundational for a lot of the work that I've done. I have done quite a lot of work in that series, but I also have a lot more planned and a lot more things that have been in the works and a lot of works from Cursive Binary that I've never shown.

So I think that's a series that maybe is pushing the boundaries of what we mean by digital art or computational poetry or generative art, but it is all those things. It's sort of a slightly unusual interpretation of them.

Cursive Binary: Data’s Girl, unique engraved marble stone (2023)

For inquiries, please contact hello@artxcode.io.