The letters page is an underrated form. At its best it does something reviews and essays rarely manage: it locates a cultural moment in a specific reader's specific life. Ed Kelsey's letter to the Guardian does exactly that. Reading Jeanette Winterson's interview about AI and the future of humanity, he was reminded of an Isaac Asimov story he has carried for fifty years — Profession, published in Nine Tomorrows in 1959 — in which the specialisation demands of a highly automated society turn out to require, urgently, a class of people who are generalists. Winterson's point and Asimov's story are, Kelsey argues, almost uncannily aligned. The alignment is worth sitting with. Asimov was writing Profession at the same moment the first commercial computers were entering the workforce, before anyone had coined the term 'software engineer' or imagined that 'coding' would become both a credential and a cultural identity. His fictional Multivac supercomputer is essentially a thought experiment about what happens when specialisation is automated down to the installation of knowledge directly into a human brain — and what that leaves out. What it leaves out, Asimov concluded, was exactly what Winterson describes as 'well-rounded people.' The story's resolution hinges on this. Kelsey is right to flag it. Jill Lepore's observation — that we are living out the sci-fi books that tech leaders read when they were thirteen — is doing a lot of work in this conversation. It is partly a critique (the futures imagined by adolescent readers of Asimov and Heinlein were not necessarily blueprints for the good society) and partly an anthropological observation about how mythology circulates through technocracy. The tech-bro who absorbed Foundation as gospel is a recognisable type. Kelsey's gentle counter-move — 'I was one of those thirteen-year-olds, and I became a computer programmer, not a ruling tech-bro' — is the most quietly useful thing in either letter. Arthur Gould's letter operates differently. Where Kelsey grounds his optimism in a specific text and a specific argument, Gould's is more aspirational: if AI is already smarter than us, perhaps it will be better at the decisions we keep getting catastrophically wrong — climate, war, the nuclear button. It is the most earnest version of the AI-as-superior-arbiter argument, and it deserves honest engagement rather than dismissal. The appeal is real. The anxiety it rests on — that human decision-making is reliably worse than any alternative we might construct — is not irrational given the evidence of the last century. The problem with Gould's framing is not that it is naive but that it assumes the question of who designs, trains, and deploys the superior intelligence is somehow separate from the question of whether its decisions will be good. It won't see the senselessness of bombing cities if it was trained on data generated by people who don't see it either. The decisions removed from humans will not be made by some neutral superior mind; they will be made by a system built by humans with interests. This is the gap Winterson's interview, and Kelsey's response, navigate more carefully than Gould's does. Taken together, these two letters sketch the live poles of the public AI debate more cleanly than most opinion columns manage. One pole: AI is a mirror of whoever built it, and what we should worry about is who that is and what they read when they were thirteen. The other: AI might transcend its builders and become something genuinely corrective. Asimov, characteristically, would have written a story about both possibilities, made them structurally inseparable, and left the reader to work out which one they were living in.
