The Lie Wearing Your Coat

The laundromat on Western had three broken machines and one television bolted near the ceiling, tuned to a cooking show with no sound. A woman in a blue coat sat under it folding tiny shirts for a tiny person who was not there. She held each shirt up like evidence, inspected it, then flattened it with both hands.

There is something honest about laundry.

You put in a dirty thing. You add soap. You wait. If it comes out still dirty, nobody gives you a fellowship or a media center or a jacket quote from a famous friend. You either wash it again or wear the stain.

This is why publishing makes me nervous. Too many people have learned to wear the stain and call it a pattern.

A man writes a book about truth and the machines that are eating it. Good subject. Hard subject. The kind of subject that should make a writer wake up at four in the morning with his mouth dry and his notes spread across the floor like dead insects. Then it turns out the book contains quotes that were never said, words put into the mouths of living people by the same machines the book is warning us about.

You have to admire the efficiency of the thing. In the old days a writer had to ruin his own credibility by hand. Now he can get help.

The quotes are the kind of sentences machines love because they sound like thinking if you do not listen too closely. Mirrors. Morality. Ethical frameworks. Social signals. Truth as a societal value. All the polished furniture of public seriousness. You can almost see the sentence wearing a name tag and waiting to moderate a panel.

Kara Swisher supposedly said one of them. She did not. Lisa Feldman Barrett supposedly wrote two more. She did not, and worse, they were wrong in the particular way that makes an expert squint and reach for a drink. Meredith Broussard got a real quote dragged into the wrong room and introduced as somebody else’s memory. Lee McIntyre got a sentence that sounded close enough to his work to be dangerous, which may be the worst kind. Not a clean lie. A lie wearing your coat.

That is the part people miss. The machine does not always hand you a dead rat and call it a sandwich. Sometimes it hands you something close. A sentence with the right haircut. A thought that lives three blocks from the truth. You are tired. The deadline is making animal noises. The acknowledgments already mention the tool. The famous people have already sent their blurbs. The book has a title big enough to stand on a stage by itself.

So you let it pass.

Maybe you tell yourself you will check it later. Later is the cemetery where good intentions go to decompose.

I worked at the post office long enough to know what happens when people trust the system because it has a system’s face. A letter looks official because it has a barcode. A machine sorts it into a bin. A man in shorts carries it through the heat. But if the address is wrong, all that machinery just moves the wrongness faster. It gives the mistake a uniform and a route.

That is what these tools do to falsehood. They don’t invent lying. People handled that job fine when the only cloud was cigarette smoke. The tools give lies better shoes. They make them clean, confident, plausible. They remove the stutter. They shave the drunk. They send the bastard out into the world with a bibliography.

And then everybody acts surprised when it finds a publisher.

The funniest sentence in the whole mess is not one of the fake quotes. It is the defense. If this serves as a warning about the risks of AI-assisted research and verification, that is why I wrote the book.

Beautiful. A man slips on a banana peel, breaks his nose, and tells the ambulance he was demonstrating sidewalk safety.

I know the move. Hell, I have used smaller versions of it. You wake up beside the wrong woman and call it research into loneliness. You bounce a rent check and call it a critique of capitalism. You burn dinner and say you were testing the smoke alarm. Human beings are magnificent at turning the crash into the lesson plan.

But there is a difference between learning from a mistake and billing the mistake as part of the curriculum.

This matters because quotation marks are little churches. Not fancy ones. Not cathedrals. More like roadside chapels with bad carpet and a vending machine outside. Still, you are supposed to take your hat off when you enter. Inside those marks, the writer makes a promise: this person said this. Not something like this. Not something that would have sounded good if they had been trapped in a podcast studio after lunch. This.

Break that promise often enough and the walls come down.

The AI crowd will call this a workflow problem. They always do. Better guardrails. Better verification. Better human-in-the-loop practices, which is a phrase that makes me want to crawl into the loop and die of boredom. They will build checklists and panels and little badges of responsible use. They will sell the mop after selling the leak.

Fine. Check the quotes. Build the tools. Put three editors and a priest on every manuscript if you can afford it.

But the deeper rot is older than the software. It is the hunger to sound finished before you have done the work. The need to appear profound at conference speed. The little itch that says a sentence is good because it glows, not because it is true. The machine did not create that itch. It just scratched it until it bled.

A real quote is heavy. You have to carry it from its source. You have to know where you found it, what came before it, what came after, what the speaker meant, what they did not mean. You have to accept that sometimes the perfect line does not exist. The dead do not always cooperate. The living often refuse to sound useful.

That is annoying. That is also the work.

The woman in the laundromat finished folding the tiny shirts and placed them in a plastic basket. The cooking show kept moving silently above her. A chef smiled with perfect teeth over a pan of something nobody in that room would ever cook.

One of the broken machines had a handwritten note taped to it.

OUT OF ORDER.

At least it had the decency to say so.

Source: ‘The Future of Truth’ Contains Quotes Made Up by A.I.