Has anyone ever seen that movie "The Miracle Worker" about Helen Keller? She is blind, deaf, and unable to speak, but she still learns to communicate. Initially, she just learns words (by using sign language, but feeling the shape of the hand) but then one day she is getting water and it hits her... she understands that the word corresponds to the thing. I think understanding that moment is a major key to AI.

Which leads me to wonder, is it useless to try to write a program that is intelligent? Will it have to be a robot that has some sort of inputs from the real world like sight or sound?

Rob

Hmmm, not really an answer. But suppose a robot *does* get a grip of sensible thinking and acting: That's the situation lots of people (who obviously don't know anything about A.I. btw.) are afraid of: A computer that starts thinking and acting of it's own (and then takes over the world!)

Robert Brooks believes this very thing. And it is experience built upon experience built upon experience that allows a robot to develop intelligent behaviour. And not only that, it uses considerably less memory and processing power to achieve it.

His views seem to make a lot of sense. How easy is it to understand a physical concept without any experience of it? Learning is considerably sped up when ideas are "put into practice." How do you learn about water without sensing any of its properties? A human would simply touch it or observe its action to have at least a basic understanding of its physical properties.

Also, its interesting to note how people with different senses inhibited have a different overall view of the world around them. Deaf people, for instance, have a different picture of objects to those that can see because their viewing is done by touch. So if a deaf person drew a 3D picture of a bottle, they would do it by drawing a flat bottle shape with a small flat circle on top of it and a large flat circle underneath it.

Linden

Isn't it Rodney Brooks? If not, then Rodney believes the same as Robert!

This concept is a key issue of robotics, and is known as grounding. Basically, whatever you use as an internal representation for reasoning needs to be physically based on something in the real world.

For example, my sonar sensor reveals the side of a big rectangular box, which could be grounded as a workbench...

It's one of the major challenges of robotics and AI in general!

Bit of a stinker that one. Sorry Mr Brookes.

I think Helen Keller has the best sense there is, very well developed by nearly all people. Don't they like to be touched.

Feeling is the basics of all other senses. Feeling is close, hearing and seeing is for far away. But a so close sense is giant good. I see no single problem she can learn and being intelligent like we.

I've seen people with only that very extended feeling, and they had a fun together, touching each other hands. No they live and love in a normal way, and for me not surprisingly.

Question, do you think mono-cells can hear and see? Don't they live, aren't they our ancestors?