The blogger Jim Nielsen compared the sense of mission at artificial intelligence companies like Anthropic to that of the young RAND Corporation analysts who planned for nuclear war in the 1950s, in a post published on Thursday.
In the post, Mr. Nielsen draws on The Doomsday Machine: Confessions of a Nuclear War Planner, Daniel Ellsberg’s memoir. Mr. Ellsberg, best known for leaking the Pentagon Papers, writes that he and a colleague, Alain Enthoven, both turned down RAND’s retirement plan. Neither expected to live long enough to collect.
“From the Air Force intelligence estimates I was newly privy to, and the dark view of the Soviets, which my colleagues shared with the whole national security community, I couldn’t believe that the world would long escape nuclear holocaust,” Mr. Ellsberg wrote. He called the plan “extremely generous” and said that he and Mr. Enthoven were the youngest members of the department. “Neither of us believed, in our late twenties, we had a chance of collecting on it.”
Mr. Nielsen wrote that many of those descriptions “sound strikingly similar to the rhetoric coming out of places like Anthropic.”
He pointed first to the religious tone. “Much, I imagined, like members of a religious order would, I shared with my colleagues a sense of brotherhood, living and working with others for a transcendent cause,” Mr. Ellsberg wrote of RAND, where he spent 10 years. He described “the burden of believing we knew more about the dangers ahead” than the generals in the Pentagon, Congress, the public “or even the president.” He called it “an enlivening burden.” His colleagues, he wrote, believed “we were in the most literal sense working to save the world.”
Mr. Ellsberg arrived at RAND during the missile gap years, when Washington believed the United States was falling behind the Soviet Union in nuclear weapons. He was 27 when he passed on the retirement plan, according to a 2017 review of the book in Slate. The fear didn’t hold up. A later intelligence estimate showed that Nikita Khrushchev had been bluffing, Mr. Ellsberg writes, a finding that undercut years of RAND’s studies on the vulnerability of the Strategic Air Command, according to excerpts from the book.
Mr. Nielsen tied the passage to a more recent claim: that AI researchers have taken up smoking to show they believe AI will cure cancer before cigarettes can kill them.
That claim came from Shyam Sankar, the chief technology officer of Palantir, who said on CBS’s “Face the Nation” on Oct. 11 that “so many AI researchers have started smoking as a way of signaling to other researchers how much they believe in AI.” He described their thinking this way: “I believe so much in AI that it’s going to cure cancer. So I’m showing you my chips are all in.” Mr. Sankar named no researchers, cited no survey and did not say how he knew, Decrypt reported. He offered the example as a criticism, and in the same interview called effective altruism a “millenarian doomsday cult.”
Business Insider reported in September that a small group of tech founders had started smoking, some of them for the networking. “When you’re smoking, you’re networking by accident,” Stephen Campbell, a co-founder of Revamp AI, told the outlet. Smoking causes about a third of cancer deaths in the United States, according to the Centers for Disease Control and Prevention.
Anthropic’s own chief executive has been more guarded about the cancer promise. Dario Amodei said in August that saying AI will cure cancer had become “more a cliche than it is inspiring,” and that only an actual cure would win public trust, Decrypt reported.
Mr. Nielsen ended his post with a joke. He wrote that he wouldn’t be surprised to hear a lab like Anthropic announce it was dropping employer-sponsored retirement benefits. He did not say any lab was considering it.
“It’d be a small sacrifice on behalf of the employees to demonstrate their faith that universal high income is just around the corner,” he wrote. “The corporate cost savings would just be a nice bonus.”