AI wizards and prophets
By 
Charles Mann wrote a great book a few years ago, about the debate between techno-optimists (“wizards”) and ecological conservationists (“prophets”) about how humans should approach scarcity.
One reason the book is so valuable is that Mann doesn’t presume to solve this ongoing debate: instead he presents it in all its complexity.
It seems to me, as I twirl the moustache of understanding, that a similar sort of book could be written (by a human) about the debate between the wizards and the prophets of AI. Some AI-like bullet points:
*There’s a rather frightening story in the WAPO about the explosion of AI-generated artifacts, including e-books, scientific papers, and musical compositions.
*I had a lunch with a very senior colleague yesterday, and was startled to learn that he’s using AI to do pretty much all the preliminary research for his current projects. He realizes that this produces a lot of initial inaccuracy — AI is constantly generating fake quotes from real authors, which seems rather problematic in an academic context — but he still finds it useful for an initial overview of subjects.
*The prestigious literary magazine Granta appears to have inadvertently awarded a prize to an AI-generated short story.
*Another thing somebody should write is an essay tracing the relationship between the LLMs and the idea of an infinite text generation mechanism. Borges explores this in an essay, “The Total Library,” which is a precursor to his famous story “The Library of Babel.”
[Thomas] Huxley (who is one of these men) does not say that the “golden characters” [Cicero speculated that tossing golden letters a sufficient number of times would eventually produce an exact copy of a famous text] would finally compose a Latin verse if they were thrown a sufficient number of times; he says that a halfdozen monkeys provided with typewriters would, in a few eternities, produce all the books in the British Museum. Lewis Carroll (one of the other refuters) observes in the second part of his extraordinary dream novel Sylvie and Bruno–in the year 1893-that as the number of words in any language is limited, so too is the number of their possible combinations or of their books. “Soon,” he says, “literary men will not ask themselves, ‘What book shall I write?’ but ‘Which book?’ “
Another connection here is John Searle’s famous Chinese room thought experiment, which in my view definitively refutes the idea that running a program employing predictive algorithms can by itself produce meaningful texts, although he does not phrase his conclusion precisely like that. (See also on this point Steven Knapp’s and Walter Benn Michael’s well-known, in literary theory circles anyway, essay “Against Theory,” which makes the same point from a somewhat different dirrection).
Consider this a thread to discuss all matters related to the uses and abuses of AI.