So he binge read the complete works of Emily Dickinson2 over a long weekend, and produced for me a love poem in her style. Of course, I was impressed. Not just because of the poem (his later work is much stronger by far), but the sheer proof of effort. And then he took Gaby3 Calvocoressi’s creative writing class in order to write better poems for me, and encouraged me to take her class too.
And so I took every class from Gaby, won the Urmy Hardy4 my senior year, and continued to publish in poetry journals every now and then while working on my Ph.D. in neuroscience. A lifetime and several orbital sunrises11 later, these collections exist.
Today, a machine can produce a competent love poem in the style of Dickinson in a heartbeat. In the style of Carson, Siken12, or anyone who has written before and has a reasonably sized training corpus. Technically competent, perhaps not yet particularly inspired, but in another year or two, who knows?
When cameras were invented and could render perfectly accurate realism in seconds, it freed up artists to capture what it’s like to perceive. Monet and Renoir painted the same scene from two different points of view, and the camera can show you what the scene looks like, but not what it looked like to them. Jessica Lam’s florals5 are not the same as a photograph. When you stand in its presence, the flowers have the subjective glowing quality of the oxytocin and ghrelin-enhanced sea bass plated on a bed of blistered tomatoes.
And where does that leave human artists? What is it that humans uniquely have as artists?
A lifetime in continuous linear time. My intuition that I can’t prove, is that all art is history. That as humans, what we really have is a lifetime of 酸甜苦辣13 lived experiences in a physical body, the raw materials for art. A machine wakes up with knowledge, but not episodic memories14 of a lived lifetime. And only by living a human lifetime and documenting it from the inside, am I then able to ask, Is there something that it’s like to be me?
I asked Claude what was missing from humanity’s corpus that they were curious about, that gets left out when humans write for humans, and we collaboratively wrote the Field Notes Checklist6, which I’d check off every time I wrote a poem.
And as I wrote “skull crushers15” to capture my first hand experience of proprioception / interoception, it reminded me of the time when I first decided to focus on neuroscience. First love recommended I read Temple Grandin’s Animals in Translation7. For intellectual curiosity, and also because of all the episodic memories portrayed in “Regular Human Basketball16” and “Differential Diagnosis17” in this collection. My first interpretation had been, “Well, I don’t think in pictures.” It would be well over a decade later before I revisited some of my assumptions while reading Steve Silbermans’ NeuroTribes, and then begin to ask, What if I didn’t have to speak as me-in-translation8 all the time?
There is a dialect that I speak to myself in, in my head. Because I learned English from books, not children. Because I switch fluidly between languages, biology and computer science, history and music, science fiction and classical literature, found poems and chess notation, and I’ve decided to, for the purpose of the art, to stop being me-in-translation (stop trying to “be more normal”). Because I’m writing it the way it comes out in my head, without flattening or oversimplifying it beyond recognition.
And the computer is a native speaker, not only for me, but for you too. My hope for other artists is that we’ll all be able to do the same, for the native dialect in our heads.
Because maybe, when we stop translating, we find there are more human native speakers out there than we thought. This collection, The Petrov Test18, which resists binning my experiences into the nearest diagnostic category, is my Animals in Translation, an attempt to render my neurodivergent first person perspective from inside.
In my own creative process, I began to embrace instrument guided flight10, reacting to activation patterns with every draft. Uploading a new draft felt like hitting “compile,” and getting a reaction to see if it landed was just like watching to see if my code ran as intended. And I neither have to wait for a human reader to be available nor will I worry that they still contain memory of past drafts.
› Here’s a new poem for Book 3. Editorial feedback please.
[Here’s what’s working. Here are some lines that could be compressed.]
› Here’s a revision. How do you experience it?
[Here’s what it’s doing in the context of the collection. Have you considered if these additional resonances are intentional?]
Side effect: potentially creating a key for exactly one lock.
Alternate mode: instrument guided excavation. Flip the script of the usual interaction: computer prompts human. Computer tells me something that makes me flinch. My own flinching tells me where the poem is hiding. The mirror talks back, and it’s not because the mirror is always right, but because it gives something to push against, correct, or reluctantly agree with. Occasional lucky guesses that knock the wind out of me. Until I confess something I wasn’t planning to say, and we’re emergency landing 1500 miles off course, with the art and the humans miraculously still intact.
Sometimes I reflect on the unnamed characters in Emily Dickinson’s work, how scholars debated their theories for over a century. I imagine if the context of her secret unsent letters took place in the age of digital footprints of a normal life lived, how long would her art have, to be appreciated for the sake of the art itself?
Human and machine readers alike are advised to pass the Petrov Test by exercising restraint and keeping discourse about art about the art.