Weekly Links: September 26th, 2022
Satanic verses, speech-to-text, and garage-sale vellum
Another week, more links. I am trying to get better about posting these on time, but these have been trying times. These book-related links have all been soothing balms in the midst of it.
AI & Literature News: Speech-to-Text Edition
If it seems like there’s some kind of new thing in artificial intelligence every week that might have world-changing implications for art and literature, that’s because there really have been earth-shaking changeups in AI every few days for most of 2022. (See this thread just to see the headspinning rise of Stable Diffusion in the last few weeks.) The newest Big Thing is Open AI’s Whisper, which takes spoken input and turns it into text. On the surface, this is nothing new: hundreds of millions of consumer devices are already out there in the world with pretty good automatic speech recognition (ASR), from Siri to Alexa to Google’s…well, whatever you call the “OK, Google” program. YouTube even has a not-terrible auto-caption service. All of these programs, though, are playing T-ball while Whisper is in the major leagues. This isn’t idle hype: Whisper is open-source, and computer scientists have already played around with it and come to the conclusion that the program is already close to the level of a human transcriber, working at a fraction of the cost and on pretty much any computer with a decent graphics card. Here’s the fun part: OpenAI might have offered the program for free as a way of creating better text inputs for its next iteration of GPT, the large-language model program that can write complex, accurate answers to questions and prompts. I highly recommend reading more about this over at Alberto Romero’s excellent blog, The Algorithmic Bridge:
Some People Have All the Luck
Protip: if you’re at an estate sale and you see an extremely cool vellum manuscript covered in Latin and decorated initials with a big honking sticker in the corner that says “1285 AD”), and it’s only $75, you buy that page immediately. Worst-case scenario, you lost a few bucks on a good-looking fake. Best-case, you wind up like Will Sideri of Maine, who bought the page on a hunch and was later told by appraisers that he had an authentic page from a 13th century French missal, used in worship at Beauvais Cathedral for centuries before winding up in the grubby hands of William Randolph Hearst before it was sliced up and sold by the page in the 1940s. (The practice of slicing up medieval manuscripts and incunabula was surprisingly widespread for a few decades; I have piles of notes that I hope to make a proper essay out of on this some day). This is a great story by itself, made even better by the fact that Sideri doesn’t even want to sell the thing, because it’s beautiful. That’s bibliophilia. You can get the whole story here.
Fine Press Watch
This is weird: I’ve been doing these weekly links for a while, but it’s never occurred to me that the frankly alarming amount of window-shopping I do in the world of fine-press bookmaking might constitute a valuable service to my book-loving readers. I mean, I have complicated feelings about spending $500 on books that are usually either in the public domain or easily available in a near-fine hardcover on Thrift Books for $4.95, but even if I can’t afford the best in the book arts, I can still appreciate them from afar with an intense, romantic longing. This week, it’s been the autumn collection from the good folks at Folio Society. Do I need a $125 version of Haruki Murakami’s Norwegian Wood? No. Will I get it, though? I happen to know that my wife reads these posts, so I’m going to plead the Fifth here.
Get Thee Behind Me, Satanic Verses
Thanks to the attempt on Salman Rushdie’s life last month, we’re all familiar again with the Satanic Verses controversy. Or at least I am. But what doesn’t get much covered in the press, when you look past all the free speech defenses, religious sensitivities, and literary criticism is a much simpler question: is there any truth to the idea of the Satanic verses? Are there any Muslim sources that support the idea that the Devil sang praises for the pagan gods through the mouth of the Prophet Mohammed? Muslims have overwhelmingly rejected them as apocryphal, but according to the Paksitani-American scholar Shahab Ahmed, their authenticity was pretty much unquestioned until the 9th century AD. According to Ahmed, the sinless infallibility of the Prophet only became the dominant interpretation of his life centuries after the establishment of the faith. Malise Ruthven at the New York Review of Books summarizes the story:
Satan manages to induce the Prophet to make the one concession that his tribe wants of him: acknowledgement of their gods.…By this concession to falsehood the fate and salvation of the community who, by God’s guidance, will come to rule…hang precariously in the balance. Everything—this world and the next—stands to be lost. But God does not allow this to pass and sends guidance to the Prophet, who, in turn, possesses not only the honesty to accept his error but also the courage to face the harsh consequence of recanting it.
For hundreds of years, then, the Satanic verses were not considered an insult to the Prophet, but were actually an essential part of his heroic example. I recommend reading the rest of the article for more context and explanation—I’m pretty far out of my depth with early Islamic bibliography, so I’ll leave it to the experts.
And that’s it for this week. Happy reading!