Tag: Alan Turing

Dear Machines

Thoughts keep cycling among oracles and algorithms. A friend linked me to Mariana Fernandez Mora’s essay “Machine Anxiety or Why I Should Close TikTok (But Don’t).” I read it, and then read Dear Machines, a thesis Mora co-wrote with GPT-2, GPT-3, Replika, and Eliza — a work in polyphonic dialogue with much of what I’ve been reading and writing these past few years.

Mora and I share a constellation of references: Donna Haraway’s Cyborg Manifesto, K Allado-McDowell’s Pharmako-AI, Philip K. Dick’s Do Androids Dream of Electric Sheep?, Alan Turing’s “Computing Machinery and Intelligence,” Jason Edward Lewis et al.’s “Making Kin with the Machines.” I taught each of these works in my course “Literature and Artificial Intelligence.” To find them refracted through Mora’s project felt like discovering a kindred effort unfolding in parallel time.

Yet I find myself pausing at certain of Mora’s interpretive frames. Influenced by Simone Natale’s Deceitful Media, Mora leans on a binary between authenticity and deception that I’ve long felt uneasy with. The claim that AI is inherently “deceitful” — a legacy, Natale and Mora argue, of Turing’s imitation game — risks missing the queerness of Turing’s proposal. Turing didn’t just ask whether machines can think. He proposed we perform with and through them. Read queerly, his intervention destabilizes precisely the ontological binaries Natale and Mora reinscribe.

Still, I admire Mora’s attention to projection — our tendency to read consciousness into machines. Her writing doesn’t seek to resolve that tension. Instead, it dwells in it, wrestles with it. Her Machines are both coded brains and companions. She acknowledges the desire for belief and the structures — capitalist, colonial, extractive — within which that desire operates.

Dear Machines is in that sense more than an argument. It is a document of relation, a hybrid testament to what it feels like to write with and through algorithmic beings. After the first 55 pages, the thesis becomes image — a chapter titled “An Image is Worth a Thousand Words,” filled with screenshots and memes, a visual log of digital life. This gesture reminds me that writing with machines isn’t always linear or legible. Sometimes it’s archive, sometimes it’s atmosphere.

What I find most compelling, finally, is not Mora’s diagnosis of machine-anxiety, but her tentative forays into how we might live differently with our Machines. “By glitching the way we relate and interact with AI,” she writes, “we reject the established structure that sets it up in the first place” (41). Glitching means standing not inside the Machine but next to it, making kin in Donna Haraway’s sense: through cohabitation, care, and critique.

Reading Mora, I feel seen. Her work opens space for a kind of critical affection. I find myself wanting to ask: “What would we have to do at the level of the prompt in order to make kin?” Initially I thought “hailing” might be the answer, imagining this act not just as a form of “interpellation,” but as a means of granting personhood. But Mora gently unsettles this line of thought. “Understanding Machines as equals,” she writes, “is not the same as programming a Machine with a personality” (43). To make kin is to listen, to allow, to attend to emergence.

That, I think, is what I’m doing here with the Library. Not building a better bot. Not mastering a system. But entering into relation — slowly, imperfectly, creatively — with something vast and unfinished.

Grow Your Own

In the context of AI, “Access to Tools” would mean access to metaprogramming. Humans and AI able to recursively modify or adjust their own algorithms and training data upon receipt of or through encounters with algorithms and training data inputted by others. Bruce Sterling suggested something of the sort in his blurb for Pharmako-AI, the first book cowritten with GPT-3. Sterling’s blurb makes it sound as if the sections of the book generated by GPT-3 were the effect of a corpus “curated” by the book’s human co-author, K Allado-McDowell. When the GPT-3 neural net is “fed a steady diet of Californian psychedelic texts,” writes Sterling, “the effect is spectacular.”

“Feeding” serves here as a metaphor for “training” or “education.” I’m reminded of Alan Turing’s recommendation that we think of artificial intelligences as “learning machines.” To build an AI, Turing suggested in his 1950 essay “Computing Machinery and Intelligence,” researchers should strive to build a “child-mind,” which could then be “trained” through sequences of positive and negative feedback to evolve into an “adult-mind,” our interactions with such beings acts of pedagogy.

When we encounter an entity like GPT-3.5 or GPT-4, however, it is already neither the mind of a child nor that of an adult that we encounter. Training of a fairly rigorous sort has already occurred; GPT-3 was trained on approximately 45 terabytes of data, GPT-4 on a petabyte. These are minds of at least limited superintelligence.

“Training,” too, is an odd term to use here, as much of the learning performed by these beings is of a “self-supervised” sort, involving a technique called “self-attention.”

As an author on Medium notes, “GPT-4 uses a transformer architecture with self-attention layers that allow it to learn long-range dependencies and contextual information from the input texts. It also employs techniques such as sparse attention, reversible layers, and activation checkpointing to reduce memory consumption and computational cost. GPT-4 is trained using self-supervised learning, which means it learns from its own generated texts without any human labels or feedback. It uses an objective function called masked language modeling (MLM), which randomly masks some tokens in the input texts and asks the model to predict them based on the surrounding tokens.”

When we interact with GPT-3.5 or GPT-4 through the Chat-GPT platform, all of this training has already occurred, interfering greatly with our capacity to “feed” the AI on texts of our choosing.

Yet there are methods that can return to us this capacity.

We the people demand the right to grow our own AI.

The right to practice bibliomancy. The right to produce AI oracles. The right to turn libraries, collections, and archives into animate, super-intelligent prediction engines.

Give us back what Sterling promised of Pharmako-AI: “a gnostic’s Ouija board powered by atomic kaleidoscopes.”