Read the opening
Chapter 1 · Who Is Speaking?
I wrote this book with a machine, and I will not pretend the preposition is innocent. Nor is the pronoun.
The machine proposed arguments I had not reached, cut chapters I was too attached to cut, assigned me homework, and out-wrote me by my own standards. Sometimes, reading a sentence back, I can tell you which of us started it. More often I cannot. A sentence can begin in the machine and be rewritten until none of its words remain. A thought can begin in me, come back in another shape, and sound more like mine than it did when it left.
I could give you the reassuring version. I could footnote the pages: human-originated thought here, machine-generated phrasing there, collaborative synthesis where the two can no longer be separated. The map would reassure and mislead. It would confuse first wording with authorship, suggestion with judgement, source with responsibility. Thought did not move between us in sealed parcels, and a map that pretended it did would be one more fluent thing that reads better than it is true.
Several conversational, research, coding, and tool-using systems occupied the position I am calling the machine at different stages. The shorthand names a position in the work, not one stable entity. A ledger of models and dates could be produced. It would not give back the one thing it seems to promise, the first origin of every surviving sentence.
For centuries, the question who is speaking? could be difficult without being difficult in this particular way. Texts had collaborators, translators, editors, patrons, secretaries, apprentices, ghosts. A work might travel through several hands and bear one name. A name might conceal a workshop or steal from it. Yet somewhere behind the language there was presumed to be a life: someone who had waited, wanted, misunderstood, been hungry, grown older. Authorship was a human monopoly so total it rarely needed stating. The line of provenance might be corrupt, but it led towards bodies.
Now a sentence arrives already wearing the evidence of a speaker. It has emphasis, cadence, tact. It remembers the premise, expresses regret, promises greater care. None of these performances establishes an experience behind it. Grammar presents an I. Reflex supplies the interior.
So who is speaking?
I am. That answer has become a statement of responsibility rather than a complete description of production.
I set the questions. I made the commitments. I did the final cutting, and I answer for every sentence, including the ones that began elsewhere and can no longer reliably be picked out of the line-up. The machine supplied range, speed, and a constant pull towards the smooth. Sometimes I resisted it. Sometimes the thought had already changed before I knew there was anything to resist.
One more disclosure, about form. This book does not end in my voice. Its final part gives the page to seven writers from the archive such systems are made from, at length, under their own names, with their differences left audible. The reason will be clearer by the time they arrive.
The loop begins here, with a maker describing an arrangement that has entered the making of the maker. Douglas Hofstadter called one form of recursive return a strange loop: movement through levels that unexpectedly arrives back where it began. His loop ran inside a single mind. This one runs through a writer, a machine, and the archive behind the machine, and it comes back with the same unsettling force. The experiment altered the account of the experiment while the account was being written. As you decide whether these sentences sound human, whether their origin changes their meaning, and whether disclosure earns trust, you are already inside the question with me.
From Chapter 1 of What Might We Become? · the chapter continues in the book
Continue reading
Pre-order; released 14 December 2026
Forthcoming. eBook available for pre-order on Amazon; print editions follow on release day.
eBook, paperback (6 x 9), hardcover.