The Curiosity Loop in Practice: The Translator Who Stopped Playing Defense
Chiara has translated novels for twenty-two years from a book-crowded apartment in Bologna — the kind of translator who will spend an afternoon on a single sentence because the author hid a joke in the rhythm and the rhythm is the joke. In March, her agency sent a new rate sheet. Machine translation would now produce the first pass on commercial titles; translators would be paid to “post-edit.” The rate was forty percent lower. Her reply, composed at speed and never sent, ran three paragraphs and included the phrase a machine cannot hear the music.
She was right about the music, as it happens. But being right is not the same as being finished, and Chapter 1 of the book offers a practice for exactly the moment when expertise feels cornered: the Curiosity Loop — four movements, repeated as needed, that turn a defensive reaction into an experiment.
Notice
She started where the loop starts, with the reaction itself, observed without judgment.
Tight jaw. A letter written in one breath. I hadn’t felt this exact anger since a publisher shortened one of my titles without asking. Which tells me the anger is about territory, not really about money.
Question
Then the assumptions, challenged one by one.
I’m assuming clients pay me for translation. Do they? Or do they pay for judgment — for the thousand small decisions the reader never sees? I’m assuming the machine’s draft is a rival. Is it? I’ve never actually read one all the way through.
Experiment
One small action to test the question. She took a chapter of a novel she loved — one she had translated years ago, so she knew where the traps were — and ran it through the machine. Then she edited the output, honestly, hour by hour. The result surprised her twice. The machine handled the scaffolding — dialogue tags, description, the plain carpentry of prose — faster than she ever had. And it walked straight past the chapter’s best moment: an ambiguity the author had built on purpose, which the machine resolved because machines resolve. But the second surprise cut the other way: in one clause, the draft caught a reading she had gotten wrong in her own published version, twenty years ago. The old proverb says traduttore, traditore — translator, traitor. Apparently the machine could betray a text in new ways, and occasionally keep faith with it where she hadn’t.
Reflect
What surprised me: my fear was about identity, not income. What assumption broke: that the draft competes with me — it competes with my first draft, which was never the part I was proud of. New question: if the scaffolding is free now, what exactly am I selling?
The answer became her new rate sheet. For commercial titles, she now accepts the machine’s first pass and sells what she calls the voice pass — the judgment layer, priced as judgment rather than as word count. For poetry and for the literary novels that are the reason she does this work at all, she declines the machine entirely, and says so in the contract, and clients accept it because she can now articulate precisely what they’re paying for. She translates more books than she did a year ago. The loop didn’t talk her out of her convictions; it moved her from defending the whole territory to knowing exactly which ground was worth holding.
The loop begins again, she notes, every time the technology improves — which is to say, often. That’s why it’s a loop.
Chiara is a composite — drawn from conversations with translators and other craft professionals whose fields AI reached early, with details changed. The unsent three-paragraph letter exists in every profession.
Try it yourself
The next time a new AI capability lands in your field and your first reaction arrives pre-written, run the four movements — Notice the reaction in your body before you argue with it, Question what you’re assuming, run one honest Experiment on real work you know well, and Reflect in writing on what surprised you. Then let the new question start the loop again.
Comments