Put another way, ChatGPT seems so human because it was trained by an AI that was mimicking humans who were rating an AI that was mimicking humans who were pretending to be a better version of an AI that was trained on human writing. This circuitous technique is called “reinforcement learning from human feedback,” or RLHF, and it’s so effective that it’s worth pausing to fully register what it doesn’t do. When annotators teach a model to be accurate, for example, the model isn’t learning to check answers against logic or external sources or about what accuracy as a concept even is. The model is still a text-prediction machine mimicking patterns in human writing, but now its training corpus has been supplemented with bespoke examples, and the model has been weighted to favor them. Maybe this results in the model extracting patterns from the part of its linguistic map labeled as accurate and producing text that happens to align with the truth, but it can also result in it mimicking the confident style and expert jargon of the accurate text while writing things that are totally wrong. There is no guarantee that the text the labelers marked as accurate is in fact accurate, and when it is, there is no guarantee that the model learns the right patterns from it.

I avoid contact with fans. Occasionally, I watch trash TV because I think the poet shouldn’t avert his eyes. I want to know what others aspire to. I’m a good but limited cook. My steaks are excellent, but they’ll never touch what you can get on any street corner in Argentina. Tree huggers are suspicious to me. Yoga classes for five-year-olds—which in California are a thing—are suspicious to me. I don’t use social media. If you see my profile anywhere there, you can be sure it’s a fake. I don’t use a smartphone. I never quite trust the media, so I get a truer picture of the political situation by going to multiple sources—the Western media, Al Jazeera, Russian TV, and occasionally by downloading the whole of a politician’s speech. I trust the Oxford English Dictionary, which is one of mankind’s greatest cultural achievements. I mean the one in twenty massive volumes with six hundred thousand entries and more than three million quotations culled from all over the English-speaking world and over a thousand years. I reckon thousands of researchers and amateur helpers spent 150 years combing through everything recorded. For me, it is the book of books, the one I would take to a desert island. It is inexhaustible, a miracle. The first time I visited Oliver Sacks on Wards Island north of Manhattan, I had mislaid the house number but knew the name of the little street. It was evening, winter-time; the slightly sloping street was icy. I parked and tiptoed along the icy pavement looking into every lit-up home. None of the windows had curtains. Through one window I saw a man sprawled on a sofa with one of the hefty volumes of the OED propped on his chest. I knew that had to be him, and so it was. Our first subject was the dictionary; for him as well, it was the book of books.

Disposable products aren’t merely more convenient than the alternative; they’re also safer, particularly during a pandemic but also at any other time. And they have other virtues: the throwaway society is healthier, cleaner, more economical, less wasteful, less environmentally damaging—and yes, more “sustainable” than the green vision of utopia.