Computations at 59

Two years ago, turning 57, I noted that number’s a “Grothendieck prime,” based on a reported error by Alexander Grothendieck, mathematical genius and later hermit, in citing it as a prime, even though it’s 19 times three. Now, turning 59, I go into a year of a real prime, also a twin prime (in that 59 and 61 are just two apart; whether there’s a highest twin prime, or they go on forever, is an unsolved problem) as well as a super-prime, or higher-order prime, because 59 is the 17th prime number, and 17 is a prime. As 17 is the seventh prime number, the sequence 7, 17, 59, 277, has what number next? (See answer.)

ChatGPT gave me wrong answers to that question, first something based on the spacing of the numbers, then after I added the context about primes, a number that’s plausible but off by 10. The program was better at responding to my statement, “You gyre and gimble me,” noting that’s a reference to Lewis Carroll’s Jabberwocky—which I’d not realized it was. Some 40 years ago, I read a book called The Universe Within, by science writer Morton Hunt, who dismissively mentioned a 1960s faux-therapy bot: “It understands nothing; if you spoke nonsense to it, you’d get nonsense back. Were you to tell it, ‘You gyre and gimble me,’ it would ask, ‘What makes you think I gyre and gimble you?’” Today’s chatbots convey vast information, true or false, along with an appearance—however illusory—of understanding.

I’ve been reading a lot about AI lately, much of it carrying a message of skepticism about expansive claims of recent years; a backlash is taking shape. An essay in The New Yorker, “Why A.I. Isn’t Going to Make Art,” by sci-fi writer Ted Chiang, makes a valuable case for the importance of personal experience in art and writing, precisely what’s lacking in generative AI. He points out that artists and writers are constantly making choices about what to include in their work; that assuming an AI can provide “inspiration” for human creativity pits undue weight on the initial choices in conceiving a project, when numerous smaller-scale choices in implementation are no less important.

The Atomic Human: What Makes Us Unique in the Age of AI, an impressive book by computer scientist Neil D. Lawrence, outlines differences between humans and computers, with wide-ranging topics including the author’s experiences developing Amazon’ delivery system, and various decisions made in World War II by figures well-known or obscure. Human intelligence, Lawrence emphasizes, is shaped by our bodies and their limitations. In a striking analogy, he writes that we’re all, in a sense, “locked in,” different only in degree from Jean Dominique-Bauby, a paralyzed man who wrote a memoir, The Diving Bell and the Butterfly, by winking to indicate letters on an alphabet grid.

Humans’ narrow bandwidth—the volume of information we can transmit and receive—means our communication must consider interlocutors and context in ways an AI doesn’t, Lawrence emphasizes. If one walked into a retail shop, for example, and told the owner that it would be a shame if someone threw a rock through the window, this likely would be seen as a threat related to a protection racket. However, a “customerbot” and a “shopkeeperbot” would regard this exchange as mundane, “a small part of the myriad facts and figures the two machines are capable of exchanging in milliseconds.”

Lawrence is skeptical that AI is leading to a “technological singularity” or “super-intelligence,” with domination by smarter machines or machine-human hybrids, ideas popularized by thinkers such as technologist Ray Kurzweil and philosopher Nick Bostrom (about whom I recently wrote regarding his concept of a world-governing “singleton”). That’s all “hooey,” according to Lawrence, who argues such notions “misrepresent intelligence as a unidimensional quality.” Rather, he contends, intelligence has a “competitive intransitivity,” similar to that of children’s games such as Rock, Paper, Scissors, or the card game Top Trumps, in which there’s no one strategy or asset that enables a player to win all the time. Different types of intelligence are suited for different things, and not amenable to across-the-board rankings.

Turning 59, I hope to be around a few more decades. I’ve no expectation that anyone in my lifetime, and probably not for centuries or millennia to come, if ever, is going to live indefinitely by uploading their mind or through some other tech fix. I doubt that computers or cyborgs are going to run the world, though there may be some moments when the failures of ordinary humans make that an appealing fantasy. My guess, as well as my hope, is that humanity will go on for a long time.

—Follow Kenneth Silber on X: @kennethsilber