The Rhetorizor Problem

In his 1964 novel The Penultimate Truth, Philip K. Dick placed a character named Joseph Adams in front of a machine that should not have been possible for another sixty years. The rhetorizor accepts a topic, an argument, a direction, and produces well-formed paragraphs calibrated to persuade. Adams discovers, as millions of users of large language models would discover six decades later, that a meager prompt produces meager output: the machine is capable but not autonomous, requiring direction, specificity, a human intelligence shaping the request before it can generate a useful response. The quality of the output depends on the quality of the input. Dick understood this dependency in 1964—and understood something more troubling that the enthusiasts of AI have been slow to confront: Adams's reliance on the rhetorizor erodes his own capacity to write. The tool that was supposed to augment his ability gradually replaces it. The muscles atrophy. The craft decays. Stripped of the specific resistance that kept