Last week, when he announced that The Free Press would be the new home of his writing, Ross Douthat said he hopes to help readers “understand our strange-and-getting-stranger century.” In his debut column today, he’s confronting perhaps the most rapidly changing strange thing in the world today. Surrounded by scary stories about how artificial intelligence is going to kill us all—or at least steal our jobs—it’s tempting to bury one’s head in the sand, especially if you’re not an especially technologically minded person. When the engineers at the frontier labs say the product they’re building could wipe us out in 10 years, and the president says all the worry is a “HOAX”—what’s a sensible American, especially one who never studied computer science, supposed to think? Read on to find out. And to read Ross in The Free Press each week, become a paying subscriber today. —The Editors
There are two paths to worrying about an artificial intelligence takeover. One runs through the worldviews that have shaped the AI industry’s mix of breakneck competition and existential dread: the quasi-religious dream of the singularity, the eccentric subcultures of rationalism and effective altruism, the special focus on existential risk.
The other runs through storytelling and myth. Never mind the precise p(doom) calculations: We should worry about the invention of a powerful nonhuman form of intelligence because we’re familiar with the legends of genies and golems, with the works of Mary Shelley and Samuel Butler and Isaac Asimov, with The Matrix or Terminator 2 or even The Mitchells vs. The Machines. The effective altruists have only rediscovered territory that the poets and artists claimed long ago.
My own worries about the robot apocalypse belong naturally to this more literary and cinematic path. As a scientific amateur, I am ill-equipped to assess competing predictions about recursive self-improvement. As a busy journalist and father, nothing can induce me to wade through the entire intellectual corpus of the doomer prophet Eliezer Yudkowsky. As a religious believer, I will never share the materialist philosophical premises that inform much (though not quite all) of AI-world thinking.
But I’m happy to hear a gripping yarn about the kind of futures that AI doomers fear. Narrative is still a place where human art exceeds machine technique, true prophecy is still a mystic gift that can’t be generated through raw compute alone, and if the doomers have glimpsed a realistic danger through the veil, by their stories we will know it.
So in the last few weeks I’ve been reading and rereading them: the famous ones, drawn from the “AI 2027” project and If Anyone Builds It, Everyone Dies, the book that Yudkowsky recently co-authored with Nate Soares, and a number of other online exercises in narrating the apocalypse.