PBS Paints Rosy—Maybe a Little Too Rosy—Picture of Technology's Future
Documentaries on robots and big data avoid some big issues.
Nova: Rise of the Robots. PBS. Wednesday, February 24, 9 p.m.
- The Human Face of Big Data. PBS. Wednesday, February 24, 10 p.m.
Being a little kid in the early 1960s was an ulcer-inducing experience. Aside from traditional anxieties like parents, teachers and girls (cooties were not officially stamped out until sometime around 1972), we had to worry about whether envious and possibly cannibalistically inclined neighbors would break into our fallout shelters when the sirens finally went off (or, if our dads were heedless slackers, whether we'd be able to break into the shelters of others). Or the epidemic of prison escapes by homicidal maniacs with hooks for hands.
And worst of all, the robots, insensate killing machines with gruesome mechanical claws and death-ray-firing cathode tubes for faces. When I was in the fourth grade, the comic book Magnus, Robot Fighter, in which a manic human karate warrior battled a government of totalitarian robots, was stealthily passed around like a samizdat resistance manual.
So imagine my disbelief while watching a new episode of PBS' Nova documentary series declare that "for generations science fiction has portrayed robots as our loyal servants." All I can say, buddy, is you haven't been watching the same movies I have.
The Nova episode Rise of the Robots is half of double shot of PBS don't-fear-the-tech documentaries, along with The Human Face of Big Data. Both are crisply written updates on the tech revolution that, while somewhat airy on the dark underbellies of their subjects (and I don't mean just the complete dismissal of the possibility that a Martian robot could kidnap Santa Claus).
If Rise of the Robots is to be believed, then I underwent all that fourth-grade psychological scarring for nothing. Even 50 years later, robots are a pack of cloddish spastics who haven't even mastered the arts of climbing a ladder or stepping over a stray cinder block, much less the enslavement of the human race.
While robot devices in carefully controlled factory environments are already building cars, packing boxes and sorting jelly donuts—the Japanese have even fulfilled the ancient human dream of a robot who can free us of the dread task of making pancakes—they haven't even begun to cope with the wild and wooly exigencies of the world outside.
"You don't realize how hard something like walking is until you try to reproduce it in a machine," says an abashed engineer in Rise of the Robots. One of the featured robots in the show required two million lines of written code, even though it can't really do much more than walk around fall down, the latter activity so frequent that it's kept on a safety tether to avoid broken arms and legs.
Much of Rise of the Robots revolves around a million-dollar contest among engineers to see whose robot can most quickly negotiate an obstacle course including eight simple tasks. The robots being the hopeless blockheads they are, contest rules permit them to be directed by human operators rather than artificial intelligence. And the imbecile lack of discernment of on the part of robots can scarcely be overstated. "We let the human tell the difference between a cat and dog, a valve and a door," one researcher explains apologetically.
The supposed point of the contest is the development of disaster-rescue robots who could drag people out of burning buildings or control a meltdown at a nuclear power plant by flipping switches and closing valves—tasks that proved insurmountably more difficult than flipping pancakes for the Japanese robots who were fruitlessly dispatched to the 2011 Fukushima.
But Rise of the Robots takes little notice of the identity of the contest's sponsor, the Defense Advanced Research Projects Agency (DARPA), which over the years has spent a lot more time and money on developing Predator drones and Arclight missiles than on rescuing kittens from trees. It may well be true at, for the moment, the only targets really threatened by robots are the jobs of $15-an-hour "living wage" burger-flippers at fast-food restaurants—a Big Mac is not significantly more challenging than a pancake—but count on DARPA to extend their range as quickly as possible.
Artificial intelligence, the flip side of data collection (data isn't much good unless you have a way to sort through it) also has grand cinematic tradition of malicious dysfunction, from the homicidally self-aware HAL of 2001 to the priapic Proteus, who rapes and impregnates Julie Christie in Demon Seed.
Yet rogue AI or pretty much any other downside is absent from The Human Face of Big Data, aside from a brief and de rigueur invocation of the National Security Agency (NSA). Instead, the documentary spends a big chunk of its time making wowie-zowie noises about the amount of data being scooped up these days.
"In the near future," observes the narrator in Big Data, "every object on earth will be generating data, including our homes, our cars, even our bodies. Almost everything we do today leaves a trail of digital exhaust."
By the year 2020, we'll already have on hand 40 zettabytes of collected data. Which, if you're wondering, is the same number you get if you count up every single grain of sand on the planet and then multiply by 75 … the most significant implication of which, I should think, is that somebody is actively conducting a census of Earth's grain of sand, which rather goes to prove Big Data's point.
Most of the rest of the show is an admittedly impressive of ways in which data collection can immeasurably improve human life. It turns out that real-time tracking of Google searches about the flu is a better predictor of outbreaks of the disease than the CDC, where doctors' reports take a couple of weeks to accumulate. And minute fluctuations in the heartbeats of newborns detected by maternity-ward monitors signal the presence of infections long before traditional warning signals like high temperatures or vomiting appear.
To collect this, of course, requires a Big-Brotherish amount of monitoring. Big Data describes an MIT study aimed at sussing out the details of how children acquire language. One of the researchers wired every room of his home to record every moment of the first two years of his first-born son's life, transcribing 8 million words of speech in the process. The study's two revelations were that words are learned less through sheer bulk of repetition than by repetition in different contexts—that is, "water" spoken aloud in the presence of a kitchen sink, a toilet and a garden hose will be picked up more quickly by a baby than hearing it 20 times off a flash card—and also that "water," spoken aloud by a toddler who has just learned it, starts to sound like "kill me now" around its 500th repetition.
Yet Big Data seems not to have contemplated how this degree of observation may affect the observed. "There's a company right now in Boston that can actually predict that you're going to get depressed two days before you get depressed," marvels one researcher. Perhaps we need a study of how many times the phrase "self-fulfilling prophecy" must be repeated to an Ivy League academic before they grasp its meaning.
Even more chilling was an unassuming interview with an executive at a Silicon Valley data-collection company in which he mused: "How do you get another two billion people on the planet? You can't do it unless you start instrumenting every little thing and dialing it just right." If that doesn't suggest to you the totalitarian impulse of the social engineer's soul, you need to be rewired, quick.