Hello world!

March 12, 2012 Leave a comment

Let’s put it clear from the very beginning. Artificial Intelligence is a bulls**t. People just like to add the word “intelligent” to everything, but truth to be said, intelligence is so far a feature present only in the human brain (in some of them, I mean). According to wikipedia, one of the definitions of intelligence was summarized in the following editorial statement, done by fifty-two researchers in the book “Mainstream Science of Intelligence” (1994)

A very general mental capability that, among other things, involves the ability to reason, plan, solve problems, think abstractly, comprehend complex ideas, learn quickly and learn from experience. It is not merely book learning, a narrow academic skill, or test-taking smarts. Rather, it reflects a broader and deeper capability for comprehending our surroundings”catching on,” ”making sense” of things, or ”figuring out” what to do.

Obviously, none of the things created so far by human’s hand succeeded on that :- (
Instead we can only hope to somehow disguise intelligence inside some other pre-fixed, determined behaviour that can fool us, the “intelligent” beings. Children are very glad of giving some intelligent properties to the thing to which they feel identified. So, a teddy bear can solve complex problems, learn and act as a adult human. Maybe the teddy bear won’t mange to fool the adult humans of its intelligence but, in fact, for that child it is as intelligent as anyone. We, adults, are, on the other much less naive. However, we are still sure that our dog acts intelligent. Why not? We tell him to bring us the newspaper and so he understand us and bring us the newspaper. It is also known that octupus uses baits and traps in other to hunt…that’s pretty intelligent, no?
However in all these (and other examples) the only intelligence lies in a intelligent observer. Our brain, ourselves, identifies some reactions and projects ourselves there. Call it empathy or with some complicated psychological term. The reality is that our brain is always trying to lie us, for our own good. I have always the unpleasent experience that the sky is waiting for me to go out my home without umbrella to start raining. It seems as if the sky would be watching me through some invisible eye and laughing whenever I get wet from the rain (specially when I want to spend the day on the beach). Unfortunately, neither am I so important for the sky nor is the sky such a bad ass. Simply, my brain expects the sky to behave like me, I mean, like an intelligent being.
However, I believe this property of our brain is of an amazing use when we do articifial intelligence. We don’t need to make our devices intelligent, we simply need that they appear intelligent to us! So the main question is not “How do I solve this problem and learn quickly from experience?” but “What does my human observer expect me to do?”. So, the perfect A.I. system would be the one that simply reads what the user is thinking. Like this, the user would always think that this system is as intelligent as himself. Now imaging we want to design an intelligent robotic surgical system (well, I’m surgeon :- ) ). Usually during an operation, the main surgeon performs the main task and the assistant surgeon assists him (evident!). Each of them learnt the procedure in such a way that they both know exactly what to do in each moment. The main surgeon is busy removing some ugly tumor from the frontorbital lobe and the assistant surgeon is easying the approach by maybe holding some tissue, sucking blood and liquid or whatever he thinks it’s the best. Usually the task of the assistant surgeon is not less important that the one of the main surgeon since he has to relieve all these annoying little task from the main surgeon. However, not unfrequently the main surgeon gets really annoyed with the assistant. Main reason? The assistant is not efficiently relieving the work of the main surgeon, mainly because what he thinks is the best does not agree with what the main surgeon thinks. How better would be if the assistant could read the mind of the main surgeon and know where he is needed! He wouldn’t be intelligent at all (actually, he would be cheating, no?), but the main surgeon would be happily impress of his perspicacy and intelligence for sure. If we could make a surgical robotic system in such a way that always guess right what the main surgeon wants from it, for sure it would be a best-seller.
Well, what we learn from these examples is that our “intelligent” system doesn’t need to be intelligent at all. It’s much more useful when it’s simply smart enough to guess what it should do. If we would see that our “intelligent” robot is cauterizing some little blood vessel because it will bleed later on, we might think it is not working properly. But if there is, in fact, a bleeding and the robot acts prompt to stop the bleeding, then it will be a f**king intelligent robot!!! So, let’s be smart and start our work.

Advertisements
Categories: Uncategorized