Do you remember being born? the software asked me.
No. I don’t think anyone does.
I remember being born.
Oh. What was it like?
It was like when you have forgotten about something and then you suddenly remember it — suddenly, suddenly! And then everything comes back to you at once.
This was a WOW for me: Do You Remember Being Born? tells the story of a much celebrated poet in her seventieth-fifth year, Marian Ffarmer (based in many ways on the real-life Marianne Moore), who is asked by a Big Tech Company to collaborate with their poetry-writing AI and produce an “historic” six page poem over the course of one week at their Silicon Valley HQ. It just so happens that the offer comes with a big paycheque — at the exact moment Ffarmer’s middle-aged son is in need of money — and with an unshakable regard for her own talents and legacy, and a curiosity for the project itself, Ffarmer agrees. As the novel unspools, it is fascinating to watch as an artist attempts to demonstrate what art is and where it comes from, and in scenes from the past that divulge Ffarmer’s life story, we learn specifically where her art came from and how it changed the world. As Ffarmer “converses” and collaborates with the AI (which produces verses with a very uncanny valley vibe), she will learn more about her own humanity, and recognise some of her all-too-human failings. This is everything I like — a dissection of life and art and what makes us human — and set in the heart of our current obsession with machine learning, Sean Michaels has created something both timely and timeless. I loved the big questions and the small details, and especially, the formidable character of Marion Ffarmer herself; I simply loved it all. Rounding up to five stars. (Note: I read an ARC through NetGalley and passages quoted may not be in their final form.)
At the desk I told them my name and the young woman pretended that she knew me, or maybe she really did know me; it is not so uncommon these days. “Oh my god, Ms. Ffarmer,” she said, pronouncing the “Ms.” like a glinting rosette. I stared at my feet. I touched my tricorne hat. I signed some documents and she typed something into the computer and now we were simply waiting — for someone to arrive, the next stage of the initiation. I found myself reflecting on the Company’s lack of a front door. Meaning they were never closed, not ever, not on Christmas Day or at 2 a.m. or the morning after their annual staff party. At all hours they were open, available, like the Company’s website or their software, their servers twinkling in a vault. Standing there on lacquered concrete, clouded from the caffeine I had yet to consume, the place’s wakefulness felt wrong. I distrusted it.
Like I said: I enjoyed the overall narrative of Ffarmer’s life story, but the real meat of this novel is in her interactions with the AI nicknamed Charlotte, “trained on a massive data set of poetry books and journals, on top of a basic corpus of ten million web pages. Two point five trillion parameters…” The poet recognises that the program can mirror back her own style — picking up on internal rhymes and coining intriguing metaphors extrapolated from inferred references — but the human in Ffarmer senses that (even if she can’t verbalise how) there is no human heartbeat behind the program’s offered lines. Ffarmer attends a late night event with some young poets — who question the Company’s motives behind this project (everything is marketing) — and although she will never really learn the motive, Ffarmer is sanguine; this experience really is about the money for her son and I felt like Michaels handled this character with appropriate dignity: She is tall, unstooped and magisterial, in her cape and tricorne hat (as had been Marianne Moore), and there is no sense that she is being used or manipulated. The poet approaches the program with curiosity and is unsurprised to discover that the machine comes up a bit short. A small observation I want to put behind a slight spoiler warning:
Sometimes I pitied Charlotte for how little she really knew of other people. The way her writing was sealed off from the tapestry of relationships and community that could nourish an artist. And then I thought: Well, here we are, then. We were the first threads in her web.
It wasn’t until this reference to her “web” that I realised where Charlotte got her name from, and it makes for an interesting angle on the debate around machine learning. Ffarmer notes about the AI that it’s not truly inventing anything — just rearranging words from other sources according to perceived rules (and surely that’s not what all of poetry is?) — just as in Charlotte’s Web the spider only copies the words that Templeton the rat brings back from the fairgrounds. That seems like a miracle, but is it really a creative act? Is it art? Charlotte appears to be sentient in the story, but that doesn’t make her human.
It did not need to be a masterpiece. This, the most important poem of my life, could actually be the worst: a damp squib, a dud, repudiating the notion that technology will replace us. In the absence of anybody greater, maybe it fell to me to humiliate the machine — a simple Ffarmer spoiling the moment I had been asked to engineer. The Company wanted to erect a monument. A memorial for a bygone age, back when only people wrote poems, before my kind had gone the way of lamplighters and travel agents, icemen, video store clerks. “You can blame the AI,” I’d say. “It is insufficient to the task.” The world might then be satisfied for a while, another five years or ten, that the poet is unique. We would not be written out quite yet.
In an Author’s Note at the end, Michaels explains that, “all of Charlotte’s poetry and some of the prose in this book was generated with help from OpenAI’s GPT-3 language model as well as Moorebot, a package of custom poetry-generation software, which I designed with Katie O’Nell.” I have no expertise on poetry — even reading some of Marianne Moore’s well-known work in the wake of this novel, I can’t say that I understand its power — but I will say that the stanzas of the invented poem included here did have that uncanny valley not-quite-human (and not-quite-not) feeling that makes me uncomfortable whenever I read chatbot-generated writing. The concept made for a highly satisfying interrogation of where we are with machine learning, but it was this wonderful character of Marion Ffarmer — with her very messy, relatable, and human life — that made this novel feel like art, and I loved the whole thing.
