Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?

To: i-node who wrote (485366)	6/5/2009 2:32:43 PM
From: combjelly	Read Replies (2) \| Respond to of 1576811

"AI is difficult but highly limited by processor power and memory." At this point in time, it isn't merely difficult, it is impossible. We can do the equivalent of an ant, i.e. follow rules, but we can't do a cat equivalent. Merely navigating a room is an extremely difficult task. I follow the Robotics Lab at A&M, and they are cutting edge. Nothing suggests the capabilities you are claiming are even close to existence. "If you have sufficient processor power and memory then you can start to do things that look like intelligence" No they don't. Nothing can come even close to passing a Turing test. "Expert Systems are totally adequate for today's needs." For some things. But an expert system is just a set of rules. If something doesn't fit a rule, it can't be evaluated. One of the most important factor in warfare is things don't follow rules. "There is little doubt that within 20 years we will see significant autonomy." There is considerable doubt. Now, it could happen. But it probably won't. To solve the problem you are talking about is what they call "strong AI". That is a very long term goal currently. My intuition is the problem can only be solved with a biologically modeling approach. Rodney Brook's ideas on subsumption architecture makes a lot of sense to me. Actually, bentway's mention of fusion is very apt. An ongoing assumption for decades has been we understand the fundamentals and it is all engineering now. That assumption has been proven false with every machine we build. The fact that ITER was recently scaled back indicates even the faithful are losing hope. It is the same way in AI. Despite decades of it being just around the corner, all we have to show is expert systems and neural nets. While neural nets show promise for running sensors, they show no promise on the cognition side. But, the bulk of the work is doing what we have been doing for a half a century, just bigger and better. There is a growing consensus that the traditional approaches are fundamentally flawed and a direction is needed. Given that Rodney Brooks and others working with biologically inspired agents have managed to exceed what the others have done in a very short period of time argues for a different approach.