Pharmako-AI

by K. Allado-McDowell

During the first summer of the coronavirus pandemic, a diary entry by K Allado-McDowell initiates an experimental conversation with the AI language model GPT-3. Over the course of a fortnight, the exchange rapidly unfolds into a labyrinthine exploration of memory, language and cosmology.

The first book to be co-created with the emergent AI, Pharmako-AI is a hallucinatory journey into selfhood, ecology and intelligence via cyberpunk, ancestry and biosemiotics. Through a writing process akin to musical improvisation, Allado-McDowell and GPT-3 together offer a fractal poetics of AI and a glimpse into the future of literature.

Pharmako-AI reimagines cybernetics for a world facing multiple crises, with profound implications for how we see ourselves, nature and technology in the 21st century.

Genres: Nonfiction, Artificial Intelligence, Science, Technology, Essays, Art

232 pages, Paperback

Published January 1, 2020

About the author

K Allado-McDowell is a writer, speaker, and musician. They are the author, with GPT-3, of the book Pharmako-AI, and are co-editor, with Ben Vickers, of The Atlas of Anomalous AI. They record and release music under the name Qenric.

Allado-McDowell established the Artists + Machine Intelligence program at Google AI. They are a conference speaker, educator and consultant to think-tanks and institutions seeking to align their work with deeper traditions of human understanding.

Reviews

[Dan Tasse · Rating 3 out of 5]

This isn't a "meh" 3 stars, it's a "part 5 star, part 1 star". The author co-writes this book with GPT-3, seeding each chapter with a few sentences, then letting it generate, only occasionally coming back in to gently bring it back on course.

The 1 star: it's dense and not in a good way. I'll read whole chapters and go "huh? what was that about?" And I think a lot of it is nonsense. Unsurprisingly, GPT-3 generates a lot of very normal sounding nonsense. It's like Wernicke's aphasia: it matches the flow of normal speech/writing but it doesn't mean anything. (I wonder how much of this is because I know it's by GPT-3! I bet nonzero.) Even the prompts are kinda pretentious; like it'd be hard enough to make a coherent fairy tale with GPT-3, but let's try to do that before we start talking about semiotics.

The 5 star: ... occasionally it's not nonsense? Occasionally it starts creating a concept, like Quiet Beat Thinking or Meglanguages, and keeps talking about them almost coherently, and that's fascinating. Sometimes it's funny, like when it starts making a table of contents then keeps sticking to that table of contents. And I dunno, it's GPT-3 + Ayahuasca: The Book, so (not to be too much of a Bay Area stereotype, but) of course I'm interested.

Not sure if I invented this criticism or read it somewhere, but it's accurate: This is not Citizen Kane. But maybe it's the Lumiere Brothers? Maybe this is one stepping stone into human+AI art?

[Maribeth · Rating 3 out of 5]

Um, I don't know how to review this book. It was a bit of a slog to read, and often felt like faux-deep nonsense. I would find my eyes reading but my mind absent because following the writing of GPT3 is not really possible. Every couple of sentences alone might make some sense, but chapters full don't.

What was it about? I don't know. The series of short, mostly-GPT3-written essays and poems meander through a variety of new age-y musings on language, AI, hyperspace...I don't know what else.

But then...maybe I just don't understand? Maybe it makes more sense than I thought? A few times, I hallucinated words, my mind auto-filling in, overwriting what was actually on the pages. It is a somewhat psychedelic experience to read. And while at first I thought it was pretty vacant, in the end I couldn't stop thinking about it. Which maybe is the point, to be a provocation as a whole? Help.

[Hestia Istiviani · Rating 4 out of 5]

Kamu pernah nggak membaca buku karena "diracuni" orang lain? 😂 Bukan karena liat review/rating di Goodreads. Bener-bener karena "marketing"-nya pemberi rekomendasi ini jago banget 😂

Pertemuanku dengan Pharmako-AI karena iseng main ke @post_santa . Ketika itu cari-cari bacaan menarik. Di antara banyaknya rekomendasi yang diberikan @maesy_ang , aku auto jatuh hati dengan Pharmko-AI. Sebuah buku dari penerbit Inggris Ignota yang punya fokus genre pada hal-hal eksperimental. Termasuk menyusun buku kolaborasi antara manusia dengan kecerdasan buatan (Artificial Intelligence).

Allado-McDowell "bercakap-cakap" dengan AI language model GPT-3. Lalu menuliskan rekaman percakapan itu menjadi buku ini. Aku jadi kebayang adegan Plankton yang sedang ngobrol dengan Karen 😂

Pembaca akan disuguhkan dengan tulisan menarik terhadap topik seputar kehidupan. Kita bisa melihat siapa yang "menulis" dari perbedaan font face-nya. Kalau sudah terlarut dengan percakapan mereka berdua, hati-hati bisa tertukar mana yang ditulis oleh manusia dan mana yang oleh AI.

Pada salindia 2-7, silakan tebak mana yang ditulis manusia & mana yang ditulis oleh AI 👀

Selama membaca Pharmako-AI, aku nggak merasa tulisannya dipaksakan. Disajikan dalam bentuk esai (& ada yang puisi), Pharmako-AI membuatku merasa seperti membaca Contact-nya Carl Sagan. Dari obrolan tentang "hyperspace" bisa merambah kemana-mana. Termasuk eksistensi manusia di bumi, tentang bersyukur & menjalani hidup, hingga linguistik & berkomunikasi dengan sesama makhluk.

Karena aku memang nggak memasang ekspektasi, aku menikmati setiap esai yang dijajakan. Bahkan aku sendiri takjub bisa menyelesaikan buku ini dengan cukup cepat. It somehow mindblown for me.

Pharmako-AI bikin aku pengin baca-baca terbitan Ignota yang lain (Measy sudah merekomendasiku untuk membaca Unknown Language). Ternyata asik juga ya mengeksplorasi bacaan yang berada di luar radarku 😆

[Trace Reddell · Rating 4 out of 5]

Best absorbed in a big gulp for maximum dis/re-orienting potency, this co-GPT-3 authored book took me a while to get into the swing of things, and I often felt like I was somehow distanced from the content by way of the abstracted networking of the AI writing. But once I realized that this was K's way of digesting and sustaining and reflecting on and digging deeper into the consequences of their pharmacological experience, and once I realized that the AI was being treated as a benevolent non-human machine entity that might be a distant cousin to a plant entity, I felt like I got deeper into the groove myself. The closing poem is really potent, and the section "When I Say Your Name" is a powerful trans-formative moment, and "Quiet Beat Thinking" is just a lovely way to phrase things, I somehow think I've had something like that experience. Echoing thought rhythms of such a personal nature into the echo chamber of the AI and its vast collection of (unfortunately non-disclosed) resources is fascinating, the shortcomings of which are called out when K realizes that they and the GPT-3 have been guilty of using only white male authors as their major reference points. Here's to the poison path. I hope more AI will listen to K than so many of the tech-bro voices out there.

[catherine ♡ · Rating 4 out of 5]

It's really fun to see where the AI speaks and where the human does — sometimes it's clearer than other times.

[Afra Ashraf · Rating 5 out of 5]

Pharmako-AI is a beautiful dialogue between Allado-Mcdowell and the AI, GPT-3, on various subjects such as ecology, ancestral memory, and cyberpunk and reading this made me hopeful and excited for a future where machines are collaborators in the creation of art.

[Fede La Lettrice · Rating 2 out of 5]

• Esperimento letterario intrigante per la sua struttura ma deludente per quanto riguarda il contenuto.

• Il testo è presentato come un dialogo tra l'autore e un'intelligenza artificiale (IA), e questo approccio potrebbe suggerire una riflessione profonda sulle intersezioni tra tecnologia e umanità. Però il risultato è una serie di riflessioni sconnesse, vaghe e ripetitive che mancano di una vera direzione o di un nucleo significativo.

• Nonostante l'interessante cornice concettuale, la sostanza sembra vacillare. L'IA, piuttosto che offrire una prospettiva unica, genera risposte spesso poco incisive quasi come se fosse limitata nel suo potenziale filosofico. L'autore, da parte sua, sembra rimanere intrappolato in una conversazione che non riesce a spingersi oltre una superficie di speculazioni pseudo-mistiche e tecnofuturistiche. Le idee proposte sono allettanti sulla carta, ma nel libro sembrano sfuggire a una reale profondità o coerenza.

• Se l'intenzione era quella di esplorare nuovi confini tra linguaggio e macchina, il tentativo fallisce nel trasmettere qualcosa di memorabile o innovativo. Il libro avrebbe potuto sfruttare meglio la sua originalità strutturale ma nel suo svolgimento dimentica l'importanza di una sostanza narrativa e filosofica.

[Lasse Olsen · Rating 3 out of 5]

I love the concept of this book and what I could have been. It’s just a taste thing where I didn’t really “vibe” (hi, I’m 12) with the inputs from the author, and therefor the output from AI likewise didn’t catch me.

I’ll bet there will be plenty of AI books in the future (that’s both a secret and non-secret)

[Daniel Hughes · Rating 3 out of 5]

The company behind this book had a more interesting premise than the book itself, it’s an interesting read but overall begins to feel very repetitive.. years into the world with AI being more normalized in daily life has definitely aged this book. For the time it was a good read but now it feels like reading hallucinations of a computer on acid.

[chaka · Rating 4 out of 5]

GPT-3 as steering apparatus for a new kind of cybernetic approach to writing. Enthralling. Incomprehensible. Sometimes both at the same time. Hopefully more to come. pretty cool. 4 stars more for the technical aspect but still

[Taylor Peters · Rating 5 out of 5]

Blew me away to forget and remember and forget that I was reading the output of an AI. New, strange, frontiers for language.

[Chris · Rating 4 out of 5]

A psychedelic trip into how AI is simultaneously both the poison and antidote to our current ecological situation.

[Alexandre Renaud · Rating 3 out of 5]

It is a very deep book that leaves you thinking deeply about the implications of the next AI wave. More broadly, it made me think of humans and our rapport with another lifeform, communication frameworks, signal and sensing, and more. It's not for everyone and not giving it a 5.

[Shelby Fielding · Rating 2 out of 5]

K Allado-McDowell's co-authored musings on ecology, ancestral memory, cyberpunk, artistic identity, and literature history with Open AI's (Artificial Intelligence) system GPT-3 (Generative Pre-trained Transformer 3) is inherently fascinating. It's a piece of text that often confines itself to deep faux, straying away from the intricacy of human narrative and rather focalizing the conversation between human-writer and machine-writer on the abstracts of life's difficulties. There's an anxiety present in the text, how the sanctity of human expression feels under threat by the potential of artificial creativity. But also a certain type of beauty, not simply in the language presented and how it dances with these dense topics, but rather the differentiation of thinking. While we often capture the world through an emotional lens, often camouflaging it with false arguments for verification, there is an objectivity to how GPT-3 presents their views of these subjects. My favorite of which is this philosophical question of who does art belong to? The pop-culture centralized notion of artist or fan is one side of the discussion, but does the art belong to the artist or the inspiration in relation to source and output? Is it the seed of the plant that sprouted? There's a cohesion here that derives upon the tropes of science-fiction literature and mutates it into these inquiries of existential existence that demands further investigation. It's a work of significant intellectual heft, only stumbling on avenues I wished it pursued further. Still, it's a dialogue, and not every problem can be dissected in one night's conversation, even one as rigorous as this.

[lala · Rating 4 out of 5]

Okay so this is actually a very trippy book to read. Before the rise of ChatGPT where you can easily converse with an AI, there is this book. McDowell conversed with GPT 3, Generative Pre-trained Transformer 3 is an autoregressive language model that uses deep learning to produce human-like text. Given an initial text as prompt, it will produce text that continues the prompt. So, McDowell will give an opening prompt and GPT 3 will continue it. What makes this book interesting is the opening prompts that McDowell gave to GPT 3. It’s sometimes in the middle of a paragraph, and sometimes just random words. In reading it sometimes you get lost on which part is written by AI and which part is written by McDowell. It was pretty thought-provoking and interesting book to read.

[Persy · Rating 2 out of 5]

“Patriarchy is the poison path of civilization. It is the way that civilizations create a medium of consciousness that perpetuates a reality of violence and oppression. It is the way that civilizations use violence and oppression to perpetuate an unsustainable reality.”

A journey of self-analyzation that is aided by an AI. This was a bit too philosophical for my taste, but not a bad read in general.