In January, the computer scientist Geoffrey Hinton gave a lecture to Are We Doomed?, a course at the University of Chicago. He spoke via Zoom about whether artificial intelligence poses an existential threat. He was cheerful and expansive and apparently certain that everything was going to go terribly wrong, and soon. “I timed my life perfectly,” Hinton, who is seventy-six, told the class. “I was born just after the end of the Second World War. I was a teen-ager before there was AIDS. And now I’m going to die before the end.”
Most of the several dozen students had not been alive for even a day of the twentieth century; they laughed. In advance of Hinton’s talk, they had read about how A.I. could simplify the engineering of synthetic bioweapons and concentrate surveillance power into the hands of the few, and how a rogue A.I. could relentlessly pursue its goals regardless of the intentions of its makers—the whole grim caboodle. Hinton—who was a leader in the development of machine learning and who, since resigning from Google, last year, has become a public authority on A.I. threats—was asked about the efficacy of safeguards on A.I. “My advice is to be seventy-six,” he said. More laughter. A student followed up with a question about what careers he saw being eliminated by A.I. “It’s the first time I’ve seen anything that makes it good to be old,” he replied. He recommended becoming a plumber. “We all think what’s special about us is our intelligence, but it might be the sort of physiology of our bodies . . . is what’s, in the end, the last thing that’s better,” he said.
I was getting a sense of how Hinton processed existential threat: like the Fool in “King Lear.” And I knew how I processed it: in a Morse code of anxiety and calm, but with less intensity than I think about my pets or about Anna’s Swedish ginger thins. But how did these young people take in, or not take in, all the chatter about A.I. menaces, dying oceans, and nuclear arsenals, in addition to the generally pretty convincing end-times mood over all? I often hear people say that the youth give them hope for the future. This obscures the question of whether young people themselves have hope, or even think in such terms.
Are We Doomed? was made up of undergraduate and graduate students, and met for about three hours on Thursday afternoons. Each week, a guest expert gave a lecture and fielded questions about a topic related to existential risk: nuclear annihilation, climate catastrophe, biothreats, misinformation, A.I. The assigned materials were varied in genre, tone, and perspective. They included a 2023 report by the Intergovernmental Panel on Climate Change; the films “Dr. Strangelove,” by Stanley Kubrick, and “Wall-E,” by Pixar; Ursula K. Le Guin’s novel “The Dispossessed”; a publication from the Bipartisan Commission on Biodefense and Max Brooks called “Germ Warfare: A Very Graphic History”; and chapters of “The Precipice: Existential Risk and the Future of Humanity,” by the philosopher Toby Ord.
Daniel Holz, an astrophysicist, and James Evans, a computational scientist and sociologist, co-taught the course. Evans looks like he’s about to give a presentation on conceptual art, and Holz like he’s about to go hiking; both wear jeans. Holz is boyish, brightly melancholy, generous, and gently intense, and Evans is spirited, fun, and intimidatingly well and widely read. Evans and Holz taught Are We Doomed? once before, online, in the spring of 2021. “As difficult as the pandemic was, my mood was better then,” Holz told me in his office, where the most prominent decoration was a framed photograph of a very tall ocean wave. He had conceived of the course after making a series of thrilling research breakthroughs on black holes, neutron stars, and gravitational waves. “I fell into a postpartum depression of sorts,” he said. “I wanted to do something that felt relevant.” In addition to heading an astrophysics research group, Holz is the founding director of the Existential Risk Laboratory (XLab), at the University of Chicago, which describes itself as “dedicated to the analysis and mitigation of risks that threaten human civilization’s long-term survival.” In college, the other path of study that tempted Holz was poetry.
Evans’s research is focussed in part on how knowledge is built, especially scientific knowledge. He is the founder and director of Knowledge Lab, also at the University of Chicago, which uses computational science and other tools to make inquiries that can’t be made by more traditional means. Evans and a co-author recently published an article in Nature which, following the analysis of tens of millions of papers and patents, suggested that the most cited and impactful work is produced by researchers working outside their disciplines—a physicist doing biology, to give one example. Evans also studies complex systems, focussing on what leads them to collapse. He likes, basically, to be surprised, and to be open to surprise. “It was important to Daniel and me that there be a sense of play in the course, that there be a level of comfort with uncertainty and ignorance and being wrong,” Evans told me. It’s hard to envision what the future will look like, he said, because “today just feels like it did yesterday. It doesn’t feel like it’s any different. But there’s the potential for really nonlinear negative outcomes.” “Nonlinear” was a word that became as familiar as toast while I was observing this class—the idea of little changes that, at some threshold, lead to tremendous, possibly catastrophic, shifts.
On the first day of class, Holz told a story that is famous among scientists, though accounts of it vary. About five years after the end of the Second World War, during a visit to Los Alamos, the physicist Enrico Fermi was walking to lunch with a few colleagues. Scientists there were trying to develop a hydrogen bomb, a weapon easily a hundred times more powerful than the atomic bombs that devastated Japan. One of the scientists brought up a New Yorker cartoon that showed aliens unloading Department of Sanitation trash cans from a spaceship. The conversation moved on to other topics. Then Fermi asked, “But where is everybody?” They all laughed; somehow everyone understood that he was talking about aliens. Surely there existed alien life that was sufficiently advanced to say hello, and yet humanity had received no such greeting. How could that be?
The “Where is everybody?” problem came to be known as the Fermi paradox. One of the more compelling responses to the paradox is to ask, Can a civilization become technologically advanced enough to contact us before blowing itself up? For Fermi and his colleagues, the prospect of nuclear annihilation required no imaginative leap.
The average age of the people who worked on the Manhattan Project at Los Alamos was twenty-five, which is not much older than the students in the class at Chicago. The energy and conviction of youth is a superpower, for better and for worse. But young people live on the highest floors of the teetering tower of our civilization, and they will be the last ones to leave the building. They have the most to lose if the stairwells start to crumble.
On a sunny February afternoon, midway through the course, I spoke with some of the students in a conference room on the fourth floor of the building that houses the department of astronomy and astrophysics. The room overlooks a polymorphous Henry Moore sculpture (from different angles it looks like a skull, an army helmet, or a mushroom cloud) and the glass-domed university library, where robots retrieve your books from stacks that run fifty feet down.
Lucy, a senior majoring in math, deadpanned that she was taking the course because it wasn’t math. “And, also, I have an unrealized prepper soul,” she said. Olivia, a senior who designed her major around the question “How do we agreeably disagree?,” had previously taken a class on the history of the bomb. She thought that her interest also had to do with family background. “When you have people in your family who have survived the Holocaust, the question of ‘Are we doomed?’ is a really serious one,” she said. Audrey and Aidan, both physics majors, were especially interested in nuclear risk. Isaiah, a sociology major, said that he valued thinking about problems over the long term, on both a personal and a societal level. Mikko, a graduate student in sociology, had two relatives who worked in the nuclear field, which made him feel close to the topic; he was also invested in how the course related to sustainability. (Later, he told me that his own work was on a very different topic: it was about “shitty food porn” and the online communities in which people post photos of unappetizing food.)
Cartoon by Jason Adam Katzenstein
The students were talkative, confident, buoyant, very much at ease, and clever. Isaiah, for example, pointed out that “doom” was a pre-modern fire-and-brimstone term, quite different from “risk,” which was tied to modern ideas of chance and probability. In various ways, the students declared the class to be a form of social therapy. Although most described themselves as “pretty pessimistic” or “not a fatalist but not an optimist,” they seemed, as a group, to intuitively inhabit, and occasionally switch, roles: the pragmatist, the persuadable, the expert. But Mikko, who had long hair and black-painted fingernails, and often wore a trenchcoat, was the designated class naysayer. He argued that the question “Are we doomed?” was unproductive, because it obscured a progressive future for climate change. He found it problematic that the A.I. conversation was driven by its makers rather than by the people most affected by the technology. “I’m a natural-born hater,” he said, acknowledging that his fellow-students sometimes looked at him as if he were wearing spurs on a shared life raft.
{Categories} _Category: Takes{/Categories}
{URL}https://www.newyorker.com/magazine/2024/06/10/are-we-doomed-heres-how-to-think-about-it{/URL}
{Author}unknown{/Author}
{Image}https://cdn.digg.com/submitted-links/664×373/1717409230-WZOlcN6BsV.jpg{/Image}
{Keywords}{/Keywords}
{Source}POV{/Source}
{Thumb}{/Thumb}