NEW YORK – Members of the general public are smart not to get too worked up when celebrities such as the late Stephen Hawking or Elon Musk predict that robots endowed with artificial intelligence will kill or conquer their human creators.
Musk put it in particularly colorful terms in an interview last fall, predicting that AI will force us into marginal habitats, or zoo cages, as we’ve done to our fellow primates.
It’s not that the concern is completely on the fringe. Various academics take it seriously, their concern boosted in 2014 with the publication of the book “Superintelligence” by philosopher Nick Bostrom. There’s even a group co-founded by an MIT physicist, the Future of Life Institute, devoted largely to this consideration.
The problem is that it’s hard to get worried about potential misdeeds of artificial intelligence when humans are creating so much trouble for ourselves — not just with wars and nuclear arms races but with mass deception, propaganda and con games so effective they’ve led some to question if truth itself is obsolete. Maybe AI would do a better job of running things than we have.
The Musk scenario — reminiscent of the movie “Planet of the Apes” except that all the primates would end up in cages, not just the humans — can’t happen until artificial intelligence becomes a lot smarter, achieving what the experts call general artificial intelligence. Current technology is pretty smart, but only at achieving set tasks.
But in a subsequent part of Musk’s interview, he hit on a path of destruction possible with today’s technology: “incredibly effective propaganda” to influence society and elections.
This is already catalogued by scientists. In his research, Indiana University computer scientist Filippo Menczer has found that automated, or “bot,” accounts on social media are sabotaging public discourse by amplifying messages from “low-credibility sources.” These bots aren’t all that smart, but they can fool a lot of the people a lot of the time.
In 2017, Menczer created a system for distinguishing bots from real accounts, and last December, he and colleagues analyzed 14 million tweets to show how much power automated accounts are wielding. His results were published in Nature Communications.
Bots make it look like something has been shared by thousands of people, he told me. And people tend to share what they think others are sharing. “Humans are retweeting bots as often as they’re retweeting humans,” he said. The number of bot accounts being created, he said, is an order of magnitude greater than the real accounts being created.
Something amplified by bots will rise in the ranks of the Facebook and Twitter algorithms, so even more people see it. While newspaper editors are supposed to elevate stories because of perceived importance, algorithms are designed to pull users in by amplifying whatever is trending. Human news judgment was always fallible, but at least there was an effort, while social media algorithms weren’t designed to inform the public.
I also checked in with virtual reality pioneer Jaron Lanier, author of the recent book “Ten Arguments for Deleting Your Social Media Accounts Right Now.” He makes a good case that social media is geared to manipulate our behavior in the service of the humans who profit from it. (I didn’t take his advice and delete my accounts, it’s true, but I spend less than 15 minutes a week on social media, so I doubt it applies to me.)
Lanier believes that people who use social media are losing their autonomy, he said, just as if we’d signed up for a massive cult. “Cults are scams in that people are tricked into becoming subservient to someone else’s agenda,” he said. And like cult members, he said, social media addicts insist they are completely free.
Lanier said Elon Musk’s robotic takeover prediction is an expression of a sort of Silicon Valley religion in which a great computer will rise to the status of superior being and become God. Musk’s interview responses are like a religious allegory: When homo sapiens became much smarter than other primates, “it pushed all the others into very small habitat.”
Other primates are threatened, to be sure. But is this really what happens when one species becomes smarter than another? In the same way we are presumably smarter than the chimpanzee, aren’t we also smarter than the rat, and the pigeon? And yet these animals thrive. When the ancestors of marine mammals entered the oceans, they were smarter than the fish, but they didn’t cause fish habitat to dwindle. There must be a smartest species of whale, and yet it hasn’t subjugated all other whales.
Maybe the problem isn’t superior intelligence, but a combination of intelligence with aggression and mob mentality. In the near future, people may be able to blame machines for the harm they are causing. But we would be wise to keep our attention focused, for the time being, on our fellow humans.
Faye Flam is a Bloomberg Opinion columnist. She has written for the Economist, the New York Times, the Washington Post, Psychology Today, Science and other publications. She has a degree in geophysics from the California Institute of Technology.
In a time of both misinformation and too much information, quality journalism is more crucial than ever.
By subscribing, you can help us get the story right.