Bill Lowery, a sales executive, is confused when a workmate asks where he should take a date out for dinosaur. \u201cYou\u2019re planning to take this girl out for dinosaur?\u201d Lowery asks. \u201cThat\u2019s right,\u201d the colleague responds, totally nonchalant. Lowery presses him, agitated: \u201cWait a minute. You\u2019re saying dinosaur? What is this, some sort of new-wave expression or something\u2014saying dinosaur instead of lunch?\u201d When Lowery returns home later in the day, his wife reports on their sick son while buttering a slice of bread. \u201cHe\u2019s so pale and awfully congested\u2014and he didn\u2019t touch his dinosaur when I took it in to him.\u201d The salesman loses it.<\/p>\n
This is the premise of \u201cWordplay,\u201d an episode of the 1980s reboot of The Twilight Zone<\/a>. As time progresses, people around Lowery begin speaking in an even more jumbled manner, using familiar words in unfamiliar ways. Eventually, Lowery resigns himself to relearning English from his son\u2019s ABC book. The last scene shows him running his hands over an illustration of a dog, underneath which is printed the word Wednesday.<\/p>\n \u201cWordplay\u201d offers a lesson on the nature of error: Small and inconspicuous changes to the norm can be more disorienting and dangerous than larger, wholesale ones. For that reason, the episode also has something to teach about truth and falsehood in ChatGPT and other such generative-AI products. By now everyone knows that large language models\u2014or LLMs, the systems underlying chatbots\u2014tend to invent things. They make up legal cases<\/a> and recommend nonexistent software<\/a>. People call these \u201challucinations,\u201d and that seems at first blush like a sensible metaphor: The chatbot appears to be delusional, confidently asserting the unreal as real.<\/p>\n But this is the wrong idea. Hallucination implies that a mistake is being made under a false belief. But an LLM doesn\u2019t believe the \u201cfalse\u201d information it presents to be true. It doesn\u2019t \u201cbelieve\u201d anything at all. Instead, an LLM predicts the next word in a sentence based on patterns that it has learned from consuming extremely large quantities of text. An LLM does not think, nor does it know. It interprets a new pattern based on its interpretation of a previous one. A chatbot is only ever chaining together credible guesses.<\/p>\n