The most alarming news of the month, if you ask me, is that Robo Brain, a “super-intelligent” robotic brain, is about to be turned loose on the world. Shades of the classic sci-fi film 2001: A Space Odyssey and Hal, the malevolent computer!
But why not, you say? Given the follies our own brains lead us into, who are we to say artificial intelligence is a bad thing? How often are we forced to admit we’ve made a big mistake?
Yet Robo Brain may be the Mother of All Mistakes. Why? Because this robotic brain is going to teach itself what to do and how to behave by consulting … um … the Internet. I am not making this up, people.
Robo Brain is designed to search the Internet for information and to “build up its own understanding from the information it gets from the Internet rather than being programmed by humans,” BBC News recently explained. So, strictly speaking, it seems Robo Brain would not be entirely dependent on its human creators for actionable intelligence.
Say what? A robot that will program itself from information it finds on the Internet? Is any sentient being in charge here? Hello?
Stanford, Brown and Cornell universities and “Do No Evil” Google and Microsoft are among the institutions that designed Robo Brain to acquire vast amounts of publicly available information from such reliable sources as YouTube (not a misprint). Not to be outdone, the Europeans also are developing their own Robo Brain.
Isn’t this precisely what the country has been crying out for: an army of robots that have educated — or mis-educated — themselves with offerings found on the Internet?
Even as we speak, the American Robo Brain is digesting a billion images, including 100 million appliance manuals and 120,000 YouTube videos. Surely Facebook, that remarkable compendium of human intelligence, will not long be overlooked as another primary source of information to shape the behavior of robots? Will they have their own Twitter accounts?
Robo Brain already is said to have learned to recognize chairs and to understand how microwaves and umbrellas are used. It can recognize a coffee mug and understand what it is used for and how it is carried. It can tell when someone is watching TV and knows not to get in the way, the BBC reported.
Robo Brain is expected to be available within 10 years in our very own homes and at such venues as elder care institutions — elder care! It even is said to be capable of recognizing complex concepts such as human language and behavior. And then what?
Smarter than us?
The inspired idea was to create a database that robots can use to perform tasks around the house or at work, according to the brain’s creators. “If a robot encounters a situation it hasn’t seen before, it can query Robo Brain in the cloud,” researcher Ashutosh Saxena of Cornell soothingly explained.
Robots already are widely and successfully used in manufacturing, but mistakes do seem to happen, even without Internet instruction for the robots. The New York Times examined eight deaths, one in a plant involving molten aluminum (don’t ask) as well as robot-caused fatalities in meatpacking, plastics, and auto factories.
Most of these unforeseen robot-human interactions appear to suggest the fatalities arose from human error. This means it won’t be only Robo Brains that will have to be trained how to safely work around humans. We’ll have to be trained to watch our step around them. They will be much stronger than we are. And will they also one day be smarter?
The point of artificial intelligence, after all, is to create a brain that makes no mistakes. So why would anyone send a robot to school on the Internet, of all places?
The Internet is a mirror of our civilization in all its glory, gore and goofiness. It’s a vast compendium of human smarts and misjudgments. Will Robo Brain be able to tell the difference? Or to stick to the script? Robo Brain’s creators obviously believe that Robo Brain will only learn good and proper things from the Internet.
Frankly, this search for the holy grail of error-free artificial intelligence strikes me as an admission of defeat by us as a species, of loss of faith in the perfectability of our own brains.
True, the offerings of the daily news cycle argue against the likelihood of perfecting the human brain any time soon, though we do tend to forget that our brains have made stunning progress over time. But how our still-imperfect brains can cobble together an artificial one that makes no mistakes remains a profound conundrum.
I’m betting on the triumph of the eternal human paradox here: we’re just clever enough — or dumb enough — as a species to eventually create an artificial brain that can outwit us.
The road to Hell, as we so often are shown but rarely recall, is paved with the most excellent of intentions.
I’m just saying …
Solveig Torvik lives in Winthrop.