by Benjamin Thernstrom
Artificial intelligence may be the greatest technological revolution of our generation, and so I’m always interested in learning more about it and how it could change our current society. Kevin Roose’s book “Futureproof: 9 Rules for Humans in the Age of Automation” is one look at how artificial intelligence could affect the society including the economy, and what steps humans should take to avoid allowing AI to make countless jobs obsolete.
A fear of AI taking over is not new: that fear goes back over a century, with the 1920 play R.U.R. coining the term “robot” in a story where robots lead a revolution against humans. Thousands of AI stories since then have followed the same idea, with our own creations fundamentally changing society and at worst demolishing it.
Roose (Image courtesy of NPR)One of the first science fiction writers to take a different approach was Isaac Asimov, who disliked how all the robot stories he read presented artificial intelligences as either identical to humans or destructive monsters. Asimov’s robot stories introduced the idea of robots as being typical machines that could be used for good or ill.
Asimov’s stories generally avoided depicting robots dominating humanity. But by the time of “The Evitable Conflict,” one of his final robot stories, computers do manage every aspect of the Earth’s economy, with human governmental positions being largely symbolic — still, it presents a world where AI and humans can exist in some kind of harmony.
This ambivalence about artificial intelligence is similar to Roose’s views. Roose’s central argument is that no matter what some science fiction has told us, humans have the ability to control their own creations and choose how to use them. Knowing what we need to do to prepare for the coming AI revolution is important for us as individuals and as a society.
Roose does look at some of the problems AI is causing now and in the near future. His book warns that many current AI designers seem to be “trying to boost their app’s engagement metrics, or wring 30 percent more efficiency out of the accounting department” instead of benefiting humankind. He also sees a need to plan ahead with “big net” programs like a Universal Basic Income for every adult or retraining programs for workers “to help people who are knocked off-balance by technological change.”
These preparations are needed because the simple truth is that for many tasks, AI can accomplish more than humans can at a faster speed. Indeed, Roose informs us that “after reaching a certain performance threshold, AI tends to outperform not only humans, but human-AI teams,” and could accomplish the things that we think of as necessarily “human” jobs including interpreting human emotions.
Aside from Asimov, a great many science fiction stories offer scenarios where AI takes over, such as Jack Williamson’s “With Folded Hands,” where AI is developed to make life easier for humanity, but the AI decides that can only be accomplished by becoming humanity’s domineering masters. And some of the examples Roose uses do seem dystopian, such as AI software that is already performing supervising work to track underperforming employees, most notably in the Uber algorithm that determines pay in “opaque, inscrutable” ways that have made drivers feel “frustrated and dehumanized.”
Roose also echoes science fiction writers in considering weaknesses of AI in solving some problems. Roose stresses that humans have the advantage when adapting to surprises and making inferences. AI is usually programmed to do one thing, and anomalies that enter the repetitive, closed systems they deal with can cause problems. “WarGames,” a Cold War sci-fi film that holds up well today, provides an especially dramatic version of this, with a computer system unable to differentiate a simulated nuclear attack from the real ones it’s been programmed to defend against.
While we may not be dealing with killer robots anytime soon, Roose tells us that we are dealing with ones that struggle with doing things we don’t directly tell them to. It’s no surprise that Asimov’s story “Lenny” treats the arrival of a creative, free-thinking robot as a revolutionary event capable of rejuvenating the entire fading industry, because versatility is precisely what AI lacks in comparison with humans.
According to Roose, AIs are “pretty far away from being able to navigate them [new situations] with ease,” which “bodes well for people whose jobs involve constant change.” The final message of his book is that workers cannot outwork AIs at their own games, but by learning to play more and changing games, we’re likely to avoid economic obsolesence.
Benjamin Thernstrom, a senior at Washington-Liberty High School in Arlington, Virginia, is an intern with PBS NewsHour Classroom. You can read more movie reviews by Thernstrom on LetterBoxd.