Friday, 23 October 2015

SMARTS Update: Anthropomorphizing the Machine

A friend sends me a link. I click. It's a documentary about humanoid robots (carried on the BBC and incorporating a lot of footage from Japan's NHK) which according to the link has had some 4 million views. My friend sent it because of my new book, SMARTS. It details when and why science threw out an old idea about intelligence (that only humans are smart) and replaced it with the notion that all of nature is intelligent in ways as varied as its trillions of embodiments. SMARTS tells the story of how this new idea of intelligence -- adaptive behavior that emerges from specific bodies in specific circumstances -- led to smart machines, machines that can sense reality, and adapt to it, and remember what works.

My friend thought I had probably seen the doc already, but of course I had not. There is just too much out there, a tsunami of information that no one can ride. I watched it with a certain fascination. It's about how autonomous robots are being designed to behave like us. In effect, they are being made in our image. (Does that make robotocists into gods?) Their argument for going this route seems to be that if you are going to make a smart machine, you will want it to do the kinds of things that people do, only better. So it's best to model it on humans rather than sheep or cows or dogs or plants or microbes.

In other words, though we have driven a stake through the heart of one bad idea (only humans have feelings and only humans can think) anthropocentrism (another bad one) is making a comeback.

The current generation of humanoid machines are pretty slick. Because they have arms and legs and joints like ours, they can run, jump, turn, climb, dive, pick and place, bring hot tea and pour it from a thermos into a paper cup without crushing the cup or being stymied by the thermos lid. And oh by the way, the plan is to design them so they can go places where we don't really like to go, like onto battle fields with guns a-blazing, or into burning buildings, or into the Fukushima Nuclear facility to clean up the post 2011 meltdown leavings, a mess of radioactive rubble.

While the documentary is good journalism, it also serves a marketing function, which is no doubt why the producers got such great access to the backrooms of robot makers like Honda, and the front rooms at DARPA. They present a nice story about how humanoid robots will become our new best slaves, welcoming us at the door, cleaning, cooking, fetching, guarding, and most of all going into  frightening and dangerous places on our behalf, sparing us the risk. It barely hints at the dark side of humanoid machines, that thousands upon thousands of human workers are going to be displaced by them. The corporations which will build them, sell them, and use them will no longer need factory floor workers, receptionists, secretaries, executive assistants, drivers....

Darwin was the first biologist to study the way animals and plants adapt and learn. It is fitting that about 100 years later, his evolutionary theory would be adapted by psychologists investigating intelligence who joined forces with computer scientists and engineers looking to mimic it in machines. They called themselves evolutionary roboticists. They found a way to use a simplified version of the evolutionary process to design robots that learn. The results of their early labors are already at work on American factory floors. There, a machine called Baxter (produced by the company started by Rodney Brooks, an evolutionary robotocist who built the autonomous vacuum cleaner, Roomba) can learn new tasks just by being shown how to do it by a human. Baxter is humanoid. Baxter has arms, and something approximating a head-- a square screen where a head would be, with eyes and eyebrows that move in order to suggest human emotions. And who knows: if intelligence is a function of a body struggling in a difficult environment and emotions evolved to help us make decisions, maybe Baxter actually feels something? Or will? This attempt to give Baxter the outward show of human emotions may make the humans beside him more comfortable. On the other hand, humans can get comfortable with just about anything as the pet rock craze of the late 1970s made clear.

Baxter's real virtue is that he is cheap, cheap, cheap when compared to human workers. Baxter will never insist on being paid a living wage year after year, will never get snippy about bathroom breaks, will never insist on an annual Christmas bonus, and will never, ever go out on strike. Baxter will reverse the flow of multinationals to China and Vietnam and Mexico where labor is cheap. But American and Canadian factory workers who still have jobs in manufacturing should not conclude they have anything like stable futures.

Honda is working on a humanoid called Asimo who will replace human receptionists, tea bearers, and care givers in people`s homes. Asimo can remember and recognize faces, sort out meaning when more than one person speaks at the same time, etc. Asimo is coming soon to a nursing home near you.

What will happen to the men and women who cannot find manufacturing, caregiver, receptionist, secretarial, and other service jobs about fifteen years from now? When will our political masters start asking these questions and make rules to guide us through what promises to be some very hard times in future? Instead of thought, we get announcements. About ten days ago the Province of Ontario, where I live, said it would become a test site for driverless cars next year, pushing itself to the forefront of the robotics revolution. There was nothing said about what driverless cars will mean to those who drive cars for a living.

One of the interviewees in the documentary tries to make the robotics revolution seem not in the least threatening, just another expression of human evolution, so how bad can that be? He reminds us that at the turn of the previous century we had an industrial revolution  (actually, it started about two hundred and fifty years ago, but I quibble). Then, at the turn of the last century we had the computer revolution (actually, it started in the middle of the last century, but I quibble again). In this century, he says, we are in the midst of the robotics revolution. This suggests that revolutions are wonderful, happiness to follow.

Yet any historian will point out that revolutions generally involve vast political upheavals, mass migrations, mass starvation, civil and regional and even world wars. Revolutions, in other words, are what you get after a dam burst of new ideas. Dam bursts leave a lot of dead bodies in their wake.

When will our political masters take note?

No comments:

Post a Comment