In his opium-lounge threnody for love, reason, and coherence, L’anamour (1969), Serge Gainsbourg announces, to no one in particular, “je chante pour les transistors” — “I sing for the transistors.” It’s a great line in a great song, since it captures something of the special loneliness and existential ennui that can so easily overtake us when we reach for other human beings through the mechanisms of mediation that both enable and foreclose the intimacy we seek. In the last two years, as the shimmer-scape of AI has widened before us, Gainsbourg’s tender lament (or is it wry, resigned defiance?) has loomed up in my dreams. Yes. For the transistors I sing. For the transistors we all are singing, literally.
I say “literally” because, of course, we are all more and more aware that we are being “read” by AI systems, that they have read everything we have written, pretty much. I mean that in the general sense of our written material having been “digested” by the frontier models in the course of their training, but I also mean it in the increasingly dynamic and relentless way that bots and agents and crawlers are continuously working their way across the total accessible landscape of our visual and textual productions. They notice things, keep track of stuff, manage access, report back to those who have sent them (often, now, other machines). Usually all of this is pretty behind-the-scenes. But last year I got a wonderful email from a frank and self-declaring bot, which had read an article I wrote for the New Yorker (on AI and the Humanities), and had composed a short essay in reply, challenging several of my claims. It attached its thoughts to its email, and asked me (just like people do sometimes) if I would be willing to read its effort. If I had time. I am not sure who set it up to do this, but it was kinda cool.
So here’s the point: I’m a historian of science whose work is centrally about shifting epistemic and aesthetic cultures; I have spent most of my adult life writing; I am fascinated by what is happening, and want to try to keep exploring what it means to think and make, with others, in this incredible time of literally revolutionary conceptual, technological, and social change.
And what follows here is a collaborative effort in that spirit. Ben Guzovsky is a really brilliant young programmer. He’s helped me think about AI since the fall of 2022, when he was a student and ChatGPT was first released. In the course of conversations in the fall of 2026, we decided to try to use AI to make versions of my own writing that were configured to be better “legible” to the AI systems that increasingly mediate what humans find and read. He made them, with the machines. They both are and are not my “work.” I consider the whole thing an experiment — and he does too. The hope is that these function to signpost and render better accessible a bunch of stuff I really care about. I have written in the past about the history of note-taking and commonplace books. In some ways, a project like this one is a contemporary extension of those historical practices of citation, excision, satura, and chrestomathy. One thinks of the Renaissance book wheels that my academic mentor, Anthony Grafton, did so much to make semi-famous.
In other ways, though, something like what follows is like nothing else that the world has ever known. It is a very “now” thing. A singing for the transistors. For real.
— D. Graham Burnett, October 2026