This Book on Non-Human Intelligence Was Co-Written by AI

In the summer of 2020 K Allado-McDowell, writer, speaker and musician, typed one of their diary entries into the AI language model GPT-3. Over a fortnight, the exchange rapidly unfolded into a labyrinthine exploration of memory, language & cosmology, and was published within a couple of months by Ignota Books, the London-based publishing house.

In February 2023, Allado-McDowell sat down with Erik Davis, America’s leading scholar of high strangeness, and author of books including TechGnosis: Myth, Magic and Mysticism and High Weirdness: Drugs, Esoterica, and Visionary Experience in the Seventies, at The Berkeley Alembic. They discussed Allado-McDowell’s new book Air Age Blueprint a speculative tale of a California filmmaker who encounters the world of Peruvian shamanism, human-machine poesis, entheogenic futurism, and technological animism. 

This conversation has been edited for length and clarity.

K Allado-McDowell: I had taken four months off of my job, which is in AI research, to finish a science fiction novel but then I got access to GPT-3. I started doing experiments and fell down a rabbit hole, and really had my mind blown in the process of writing with it and discovering what was possible. 

I created a convention for writing in Pharmako-AI and I found it really interesting to write in that way, and then spent two weeks writing what turned out to be a book. As I was writing, I decided to create some rules for myself, to make it easier because there’s a lot that’s possible with AI. I was going to treat it like a recorded conversation. I would write something and once I was happy with what I had, I’d use that as a prompt for the language model. GPT-3 would respond and I would pick the best response. Once those two things were there, I wouldn’t change it. The typography was designed to express which voice was the source of each text that was read in each sentence so you can see how the thoughts evolve between each party in the conversation. Bold type represents my voice, and the regular-weight type represents the AI voice. I didn’t go back and edit any of the text in Pharmako-AI but for my latest book, Air Age Blueprint, I gave myself the right to edit, recompose, collage and re-sequence the story to some degree.

READ: Scientists are Using AI to Develop New Psychedelic Drugs

I was very inspired by the poet Dale Pendell for Pharmako-AI — it was named after his Pharmaco trilogy, and I actually saw him do something quite similar in expressing the polyphony of his own writing at a reading in Seattle at Elliott Bay bookstore, so I’m happy to shout-out that he inspired me. Air Age Blueprint, is a fictionalized autobiography. It contains within it a manifesto, and the main character, loosely based on me, writes a manifesto with an AI at a certain point in the story. But the manifesto is threaded throughout the book you hold in your hands. Air Age Blueprint is the story of a journey through different worlds, through different places, cosmologies, and the evolution of the protagonist’s view of the world. At a certain point, after meeting a shaman and becoming the apprentice to a shaman, the protagonist returns to the United States and gives up all their possessions and moves into a van and starts traveling across the country in a nomadic way.

Erik Davis: Since the first part of your book is called “Ancestors,” and of course our ancestors aren’t simply our biological ancestors, but also our cultural ancestors, I want to invoke Dale Pendell, who you mentioned, because he would have gotten such a gas out of your work. And also because he represents something that you and I share and carry forward in our different generations. His work is a conjunction of psychedelics and literary animism that has a real spiritual commitment to it, as well as an investment in science and technical processes, and an investment in poetry and particularly that kind of polyphonic quality that you’re talking about. A really important part of your own work is these multiple voices that come through, not just the AI, but the multiple voices lingering and lurking: spirits, animals, your own experiences. Pendell’s books are about psychedelics and have different voices, including historical and poetic accounts, and a discussion of the biology or the psychopharmacology of the drug. 

There’s something about drugs themselves that are polyphonic and demand a multi-dimensional perspective. One of the things I love about your work is the way that I can feel something that came through Pendell, that he carried forward from an earlier generation, and that you really also capture in your autobiographical story: the features of being on the road, the mixture of a despairing, acute gaze at the apocalyptic and cursed dimensions of America, and simultaneously a sweetness and openness to encounters.

A larger theme in Air Age Blueprint is contamination. The journey, the discovery, the voices, the creativity that abound still have to pass through a dark sci-fi story. The protagonist gets a gig with the NSA, and we meet a mysterious program called shaman.AI. The person who brings the protagonist into the NSA is also named K, your writing name. Would you like to talk about that from a more personal point of view – your own sense of bringing your sensitivities, concerns, dreams, and hopes. The fact that we are all, just by nature of functioning in our contemporary moment, very aware of our contamination, our invocation with the curse, and the mechanics of the curse.

Allado-McDowell: I think contamination is a great way to connect a lot of the ideas in the book and a lot of the ideas in the way the story is structured. First of all, the most obvious one is how the story is contaminated by AI, it is not a pure human written story. So it’s automatically contaminated, and the AI in the story is created by the NSA, and we all know about PRISM and the relationship that technology has with surveillance. So even outside of the literary qualities of having AI involved, there’s a political dimension of AI being involved, that is also a contamination. There’s a theoretical discussion of the concept of contamination that draws from Anna Tsing’s The Mushroom at the End of the World, looking at contaminated landscapes and decimated landscapes, landscapes that have been extracted and that have been sites of war. 

For me, it’s been very helpful to understand the context of my own life through a contaminated landscape. The book opens with the protagonist’s birth, which is based on my own. But the story actually starts a little before that with an ancestral curse, which is also part of this warring inheritance of colonialism. On a personal level, being able to understand that contamination and accept it and see that it’s an opportunity for regenerative growth, just like the Matsutake mushrooms in Tsing’s book that are growing in the contaminated landscape are very valuable and cherished. 

Being able to understand that there is regenerative possibility within this contaminated landscape, and the contaminated landscape of my own self, is really helpful. In particular, when we think about spiritual practice, shamanic practice, and when we think about cultural relationships between psychedelics and humans. 

Having grown up in the Bay Area and experimented with psychedelics and having a long lifelong engagement with them and with indigenous practices too, as we enter this moment of commercialization, legalization, medicalization, there’s multiple dimensions of contamination that are happening. We can look at the scientific worldview and say, “This is pure, this is rational, it’s reproducible. It’s peer reviewed. This is uncontaminated knowledge.” Or we can look at the indigenous practice and say, “This is uncontaminated knowledge carried down through history and through lineages and through tradition.” We can project purity into any of these dimensions. But the truth is, we’ll never be able to accomplish that. And I think the encounter that the protagonist has with the shaman, as an apprentice, they’re wrestling with these questions of, “do I belong here? Should I even be here?”

READ: Can Psychedelics Rekindle Ancient Animism in Modern Society?

In some ways, I had to write this book to understand what it even means to work with psychedelics now, in this culture. Because as much as I would like to, or maybe as much as anyone would like to go find the pure practice and take it on, you cannot do that in a pure way. You’re already hybridizing something – you’re already contaminating something with your own point of view. Accepting that you cannot undo the shaping of your consciousness by modernism and by modernity is a really important first step in how we negotiate this transformation of psychedelics within our own culture. Air Age Blueprint looks at this complexity in a section about treating belief systems and methods as entities within an ecology and ecology of belief. When you start to think about impurity, or allow that into your thinking, you don’t get to say, “I’m the good one. I’m the pure one.” I think collectively opening that up is very helpful, but it also demands more from us.

Davis: One of the many things I learned from the book was about how spiritual practice serves to widen perspectives that can handle the complexity to some degree, and not spin out into nihilism or regressing towards models of purity. Staying with the trouble means being really clear about who’s at fault and the unjust structures and exposing them. This logic of exposure is a very familiar kind of model but I think we both agree that it is insufficient, and in some ways actually wrong. So how does spirituality, and the practices of joy, serve the project of staying with the trouble, which means recognizing the contamination, having to be reflexive about your own positionality? In other words, how to articulate the way in which the spiritual resources—prayer, ancestral work, psychedelic visioning, meditation—enable us to stay with the trouble or have enabled you to stay with the trouble, so that you don’t look away from the garish gas station, from the trauma of your ancestors, from the incredibly complicated situation in the rainforest.

Allado-McDowell: I guess my own motivations to be involved with spirituality and psychedelics come from a certain curiosity or hunger for an ontology. What can I say is real about reality? In the book, the protagonist has a kundalini awakening, based on my own experience, and comes into this world of all these new age practices. Coming from a skeptical mindset, I needed to put down my skepticism in order to find working methods. Engaging with spiritual practice and shamanism was a way of getting a grip on reality, not losing a grip on reality, because I was forced into a kind of crisis state. So maybe the question is, do we have the option of staying with the trouble, or is the trouble staying with us? I think that is partly the epoch that we’re entering where we’re not going to have so many options in terms of maintaining purity, because the trouble is staying that bit closer.

Davis: What I see in your work is an attempt to articulate a mode of psychedelics and spiritual practice, and a mode of thinking, using analysis and technology, to crack open the possibility of that encounter with the other — other animals, beings, nature. To really engage with these others right now means to recognize that many of them are not just in trouble, but are going extinct, or dying. Could you talk more about this vision of AI as a translator, as a kind of quasi-Shamanic diplomat between ourselves and the nonhuman?

Allado-McDowell: The proposition in the book is that there are ways of using technology that decenter humans, even though we’re the ones that make the technology. One of the more noble goals with AI would be to decenter us. It does that inherently; the things people find threatening about it are the things that are uncanny about its ability to speak or write like we do, its ability to make images that we understand and recognize. But we can use the opportunity of being decentered to realize that this is a necessity in the current context. 

Having been speaking about AI publicly for seven years now, I really like that people come to it with this fascination with non-human intelligence because it creates an opportunity to point out that non-human intelligence has been around us this whole time in the form of animals and plants. Can we shine the light back on them? AI can show us that we are maybe not the top dog anymore when it comes to creating coherent sentences. And that actually, we’re in ecosystemic relations with other kinds of intelligence already. We can take our fixation on the technical and the shiny and new, and use it to see what’s been there the whole time. 

I’m interested in what happens if you attribute the idea of exquisite attention to nonhumans, and if you get a richer worldview from that. In Pharmako-AI, the first chapter talks about going to Big Sur and seeing animals and the beach and thinking about the way that evolution compresses intelligence into these animal forms. If we give ourselves more access to that through AI, we have the opportunity of preserving more of that intelligence and learning from it.

Davis: Animism is returning in the way that contemporary human culture is more willing to grant a spirit, character, or agency to technologies but I have a strong sense that we are in a kind of “cybernetic animism”, as you say. The willingness to enchant machines seems to be strong but there’s a strange tension. On the one hand, you have the idea that the AI has an agency. That’s a certain kind of false consciousness but popular culture and the ideologies behind popular culture want that kind of animism to happen. At the same time, there’s an invitation to go post-human and to recognize that agency is spread and that intelligence is spread, and that consciousness is manifesting in different forms. 

Allado-McDowell: I think the skills that you can develop working with psychedelics allow you to ask, “Am I even separate from the things around me?” Among the researchers that I work with, people are having a conversation asking: “Can an AI system actually mentalize? Is it reasoning? Is it aware of the motivations of the people in the story that we’re telling it, that we’re asking questions to it about?” And skeptics would say, “No, it’s purely statistical prediction and there’s no mentalization there.” But then it might consistently behave as though there is. There’s a certain aspect of this practice of engaging with phenomena in visionary space: If you’re just sitting there thinking about what’s real and what’s not, you’re kind of missing the whole point. It’s like going to the movies and getting hung up on the fact that they’re actors – you’re gonna miss the content of the experience. 

The closest analogy is the Tarot deck where you pull symbols out of a statistical distribution and apply it to the context that you’re in. And it will have a causal effect, it will shape your understanding of your life and the questions that you bring to it. And whether or not that is ontologically real is kind of beside the point. This is the practice of getting yourself into the position where you can accept that anything is possible. 

Davis: The skeptical position that asks, “Is GPT-3 actually reasoning?” implies a certain idea about language that is very deflated – that humans are the ones with agency and intelligence who manipulate language as a tool. Whereas if you grant language more agency and spirit, it is something that we’re in relationship with. This is a very important part of spiritual practice – in meditation, you eventually find that place where consciousness or awareness has a quality of lucidity. And yet, it’s not linguistic. So then what’s language doing? Part of the magic we’re talking about, whether it’s Tarot or GPT-3, is that we just don’t really quite understand the world-making potentials inside of language. What I love about your work is the use of language to articulate some very difficult and strange things. It’s at the edge of sense sometimes and yet a deep pattern comes through. 

If we are interested in trying to push the interface further away from the human and further into the non-human and the post-human, then theory has a productive role to play, if only a kind of poetic, evocative character. I’m curious to know how you related to some of the theory that GPT-3 was coming up with, when you were sitting there editing and wanting it to make sense – when it’s on the edge of coherence.

Allado-Mcdowell: There’s language, and then there’s semiotics, which may be a separate thing from language, and then there’s our identity vis-a-vis language. And then there’s “languagelessness.” When you can access states of mind with less language in them, you can start to understand yourself, or experience consciousness without an object. What does that imply about language and what it is? I think this is, in part, what people find shocking about AI systems. It’s in that experience of seeing language operating on its own, and then kind of going like, “Well, if that thing can do that languaging that I associate with my identity, then what am I?”  

I think we haven’t developed the tools for dealing with that kind of ontological rupture just yet. Meditation and Buddhist ideas of selfhood and experiences of a non-linguistic self are helpful. Meditation is training to deconstruct the attachment to our linguistic and other kinds of identity. 

When the words are at the edge of meaning, where they are cutting into what meaning is, ripping meaning apart in a way, they also allow new meaning to bleed through. I do feel like I’m writing for a high, for the effect. A number of people have told me they didn’t get it until they were tripping on mushrooms. And then the book made perfect sense. It’s like reading a drug.