Marvin Minsky, co-founder of the MIT Artificial Intelligence Laboratory, seems to think so. He lashes out at the focus on robots and laments the lack of vision and grand directive towards true intelligence, sacrificied in favor of smaller bits of working knowledge and robotics.
But it seems to me that this criticism is likely to be misplaced. Correction, it's probably a great criticism to have been made, as it keeps people focused and prevents drifting too far in one direction or the other, and I suspect that if the research trends had been solely in the direction of processing at the expense of robotics, Minsky probably would be complaining that we have all this intelligent capacity but no way for it to interact with the world.
But two things in the article stand out to me. The lesser of the two is at the end of the article when a fellow AI researcher points out that once we solve a particiular problem, it ceases to be seen as intelligence and gets relegated to just part of the standard repertoire of stuff that computers do. The whole definition of what is intelligent behavior by a computer tends to shift, often unintentionally.
The other thing was more important, or at least more substantial. Neuroscience has shown our minds are highly dependent on feedback from our physical bodies. I'm not just talking about the conscious senses, but also the unconscious senses. Our brains are constantly monitoring the state of our physiology and our minds are constituted, in part, from the information provided by this constant feedback.
Imagine if you were to be transplanted into someone else's body. If you close your eyes and relax, you can bring your sense of awareness about the rest of your body to the conscious forefront. But even when you aren't aware of, say, the weight of a shirt on your shoulders or the pressure from an elastic waistband, your brain is nevertheless monitoring it and sending reports to your mind. When something happens that it outside of some learned or conditioned range of 'normal,' our brains then prioritize this information and our minds bring it to the front, and we respon accordingly.
What this means is that building robots may not only be useful for creating artificial intelligence but that it may actually be necessary. I don't know if AI can be formally programmed or only prepped via programming, but either way, its going to need a physiological matrix of self to define its entity-ness from which to percieve and interact with the outside world.
So I guess I'm saying I don't think that the future of AI is as bleak as Minsky fears (but I pray to god its better than movie of the same initials).
(Editor's note: Astute readers may notice that the last two entries today both ended with references to shitty movies. We apologize for this inconvenience; It was solely a coincidence and we assure you we are doing everything we can to avoid future expressions such as this. This was not an editorial decision for the future tenor of this webspace. Respectively, ---The Editor.)
Posted by Nutrimentia at May 22, 2003 01:05 AM | TrackBack