Saturday, July 25, 2009

Artificial Intelligence - Why Bother?

Sometimes a complex problem has a simple answer.

The economist recently published an article on an advanced robot arm -

A quick description:

"Shadow Robot has developed a robotic hand that closely mimics the human version. It has already sold several of them to various universities and to NASA, America’s space agency. And it has taken an order from Britain’s Ministry of Defence, which wants to try the hand out on the arm of a bomb-disposal robot. . . .

The robot hand mimics the movements of a human operator who wears a special “virtual reality” glove equipped with sensors that can determine the positions of the fingers inside it."

This seems really cool, and I can think of a lot of neat applications. What doesn't make sense to me are the next steps they plan to take:

"The next stage of development, says Mr Walker, will be to add some level of intelligence. The company is involved in a European Union programme to develop technology, such as machine vision, to make robots cleverer. This would enable the hand, for example, to recognise an object like an egg and know how to pick it up without breaking it. Unless, of course, it was clever enough to know that it was making an omelette."

Why add AI to this thing? For $4 bucks an hour or less, you could have people in India or China work remotely using these amazing hands, and they already know what an egg is. Strap the arms to a torso with wheels and a webcam and you have a fully functional C3P0. They should focus on getting the manufacturing costs of the arms down. It will take decades upon decades and millions of dollars for an AI creation to match what remote workers could do today.