When the machines revolt, will you be prepared for it?

San Francisco Chronicle
Nov. 28, 2005

How to Survive a Robot Uprising

Tips on Defending Yourself

Against the Coming Rebellion

The unnerving thing about "How to Survive a Robot Uprising," this self-defense handbook with its splashy red neo-retro graphics showing tiny people fleeing huge machines, is that, now and then, you can't tell whether or not it's a joke. You suspect that it is, because -- well, because you're laughing. And because of its deliberate deadpan prose style that echoes those pamphlets you find in government offices and hospital waiting rooms.

After dryly lobbing Doomsday our way -- explaining that "someday mankind must face and destroy the robot menace" -- Wilson goes all educational (he is, after all, a Ph.D. candidate in robotics at Carnegie Mellon University), explaining that robot is Czech for "laborer" and was coined in a 1920 play about mechanical men wiping out humankind. He warns that the androids-gone-wild scenarios he is about to showcase are quite possible; and if your mental image of what Wilson calls "the robot race" ranges whimsically from "Lost in Space" to "Star Wars" to "The Terminator" to "Futurama," the danger in his warning swells and hums as he delineates just how many artificially intelligent mechanized life-forms are already in development and production. Robot toys, robot servants, robot explorers, robot consolers, robot appliances in robotic "smart houses," robot rescuers, robot soldiers and robot spies -- they're all "quietly insinuating themselves into our lives" right here, right now.

"Fellow humans," Wilson soothes -- and that greeting is such science-fiction camp that by now you know it's a joke -- "we will live to fight another day."

Yet these are shivery titters for sure.

What a thin bit of impulse-table fluff this book would be if its author weren't an actual scientist. But Wilson's presence in the field and his interviews with top researchers make this a rich AI crash-course, laid out in language as light and plain as titanium and rendered painless by its "the-robots-are-coming" gimmickry. "For now, the human is the puppet master," Wilson asserts. But as robots learn to "jump farther, run faster, and lift heavier loads -- the threat to our species will be revealed, especially when a humanoid robot steals your deer-hunting rifle and drives away in your pickup truck, firing wildly into the night."

Ingenious devices are now patrolling harbors, locating land mines, driving themselves unmanned down highways, winning chess games, scaling walls and replacing light bulbs on the outsides of orbiting spacecraft. Looking like boats and planes and monsters and animals and us, these devices can recognize faces, see through clothing and skin, hear sounds underwater, predict human behavior, and replicate human voices. "Designed for danger," they can withstand bullets, shoot lasers, repair themselves, change shape and communicate with their own kind instantaneously. They're huge; they're tiny. One airborne AI device "looks like a flying albino whale," while housefly-sized micro air vehicles (MAVs) gather information via microscopic electronics while hovering undetected; UC Berkeley researchers have built a solar-powered robotic fly complete with carbon-fiber wings that beat 150 times a second. While all these expensive wonders were created to solve our problems, it's conceivable -- just barely -- that someday, through nefarious programming or just plain mistakes, the wonders might turn on us.

If you can overlook Wilson's over-the-mark musings about "evil robot logic" and "mad metallic fiends" -- both evil and madness, after all, require the presence of a soul -- lots of laughs rest on his nightmare vision of robot autonomy and strategies for outlasting it. To fool speech- and facial-recognition systems gone sinister, he suggests mumbling, adopting a fake accent, using body language inconsistent with your words, and calling the robot's mother a slut. To detect an android posing as a human, get close and sniff: "Does your friend smell like a brand-new soccer ball?" To evade robotic pursuers, run a zig-zag course over tussocky ground while throwing obstacles in your wake.

It's a brilliant way of explaining to the rank and file how robots actually work, how they move, what they're made of and how they think. Robots' limitations are almost as intriguing as their amazing capabilities -- and comforting, because in this day and age we cannot help but make constant comparisons between flesh and machine. Although, as Wilson points out bluntly, "the human species is almost singularly pathetic," at least we can feel. And they can't. Yet.

But nowadays, any book about lethal machines is going to make you think of other uprisings, other invasions, other outbreaks, other explosions, other us-against-thems. Wilson's first-aid lesson for laser wounds -- "a reality on today's battlefield" -- is as grimly pragmatic ("Do not try to remove any clothing melted to the wound") as it is hilarious ("Stay alert and watch the ground around you for dropping limbs"). Science fiction is getting less and less fictional every day. "Knowing when something is wrong -- even a split second before an attack -- can save your precious human life," Wilson warns. The trouble is it's true.