Thursday, August 29, 2013


Robot Futures
By Illah Reza Nourbakhsh
MIT Press

Forget the scary scenarios and dire special effects of our recently concluded Apocalypse Summer at the cinema.  Don't even bother anticipating the next climate crisis report.  This book is really frightening about a future that's coming on fast, and will be here in a decade or so. And we're really not ready for it, at all.

First of all, what is a robot?  Machine intelligence in a human-like shape, hostile or friendly?  An impressive tangle of arms and pistons making cars in the modern factory?  These images and the issues they raise just scratch the surface.  Nourbakhsh, who is Professor of Robotics at Carnegie Mellon University, breaks down the functions to perception, action and cognition (which is "the ability to reason, to make decisions about what to do next.")  Robots have these in unequal measure, but together they exist or will exist with effects we can only imagine (and Nourbakhsh does.)  We aren't necessarily on a direct path to Robbie the Robot or even Data the android, "but rather on the road to a strange stable of mechanical creatures that have both subhuman and superhuman qualities all jumbled together, and this near future is for us, not just for our descendants."

Some of the most powerful robots aren't even physical in any familiar sense.  They operate mostly or entirely in cyberspace. They gather and analyze information and make decisions based on it.  Some are already doing so, helping marketers to not only learn consumer preferences but manipulate choices, and even set different prices for the same product for individual consumers.  Add inputs like cameras, and robots can predict what a consumer will like based on the make of the car pulling into the driveway.

Physical robots will also be different from our preconceptions--they will be larger and much smaller than people, able to see in the dark, snake through wreckage to sense survivors, and probably leap tall buildings in a single bound.  As microprocessors and sensors get smaller and more energy efficient, all kinds of robots for amusement as well as mischief become possible, and as costs drop and designs are standardized, they will become ubiquitous.  With 3-D printing, maybe even uncontrollable. "We will not be able to distinguish potential Borg from homebrew."

More complex robots will also be possible because all the information doesn't have to be stored within it--the robot's brain will link to the immensity of the Internet.  Some of the traditional issues will arise, though: when robots look like people or even like dogs, in what sense are they alive?  Is cruelty to robots even possible?  How do you act when you can't tell if the voice on the phone belongs to a robot or a human being?

Nourbakhsh communicates a lot of information in this small book, about what's possible now, what the limitations are and how soon they're likely to be overcome, as well as what's on the drawing board or could be.  He also produces future dialogues and scenarios that do what stories can do best--show us the possible effects of these technologies in the real world.  These are perhaps the most effective--and scariest--parts of the book.  Especially the stories based on something that's already happened.

He suggests that the most important effects may be the unintended consequences of decisions made by individuals, companies and other entities without sufficient regard for the public good.  He offers some ideas for turning potential horrors around before they happen, and suggests "we should become more deliberate and considered as we imagine and design technologies that carry us forward."  Knowing what we might face is the motivating first step, and this book helps us take it.  It should be widely read by all who care about the future they or anyone they care about will live in.  Meanwhile, there's a website: robotfutures.org.

No comments: