I am pretty sure since its my field of research.
I am working on evolutionary robotics, that is
using genetic algorithms to evolve neural networks controlling autonomous robots. I also
use with simulators, but you are never sure
how detailed the simulator should be. If you
leave something out, the robot behaviour is
different in the simulated and the real world.
You can't simulate every atom in the universe,
agree on that? Chances are the you will miss
something vital, or make the simlation too
precise (articial noise level). For this reason Rodney Brooks came up with the notion "The world is its own best model". The Matrix is a nice movie but the argument doesn't hold. You can't simulate everything, sorry. You are stuck in Kansas!
Speed is in fact important. Human level intelligence is not a symbolic mathematical
problem just can just simulate. Instead you
have to "do it", that is replicate it in a
real environment.
A brain without a body makes no sense (even though I liked that Steve Matin movie). During evolution the brain and body of the human co-evolved and
they remain totally dependant on each other.
Unfortunately those clasically symbolic manipulators, among who you can count Dr. Richard
Wallace, who calls their research Artificial
"Intelligence" are still stealing headlines. What
they do might be interesting from a matematical
point of view but has no interest if you want to
understand human level intelligence. The brain
doesn't work by manipulating rules to infer an
answer. Wheater you do that on a pocket calculator
or a supercomputer makes no difference. You won't
get it right. The brain needs a body which interacts with a physical environment in order to
show intelligence. A physical environment follow
the laws of physics which is why speed _is important_. You can't speed up the physical laws.
In 1986 Rodney Brooks published a paper called
A Robust Layered Control System for a Mobile Robot in order to get AI research on the right track. He was only partly successful, but at least at MIT where he is now the leader of the AI lab, people are doing interesting research now. Another good pointer is his resent book Flesh and Machines: How Robots Will Change Us. More information
on embodied intelligence - a brain
needs a body - can be found in the very good
by Rolf Pfeifer (leader of the AI Lab in Zurich)
called Understanding Intelligence".
Please stop making a fuzz about talkbots and stuff
like that, it's really not that interesting.
Slashdot Mirror
I am pretty sure since its my field of research. I am working on evolutionary robotics, that is using genetic algorithms to evolve neural networks controlling autonomous robots. I also use with simulators, but you are never sure how detailed the simulator should be. If you leave something out, the robot behaviour is different in the simulated and the real world. You can't simulate every atom in the universe, agree on that? Chances are the you will miss something vital, or make the simlation too precise (articial noise level). For this reason Rodney Brooks came up with the notion "The world is its own best model". The Matrix is a nice movie but the argument doesn't hold. You can't simulate everything, sorry. You are stuck in Kansas!
Speed is in fact important. Human level intelligence is not a symbolic mathematical problem just can just simulate. Instead you have to "do it", that is replicate it in a real environment. A brain without a body makes no sense (even though I liked that Steve Matin movie). During evolution the brain and body of the human co-evolved and they remain totally dependant on each other. Unfortunately those clasically symbolic manipulators, among who you can count Dr. Richard Wallace, who calls their research Artificial "Intelligence" are still stealing headlines. What they do might be interesting from a matematical point of view but has no interest if you want to understand human level intelligence. The brain doesn't work by manipulating rules to infer an answer. Wheater you do that on a pocket calculator or a supercomputer makes no difference. You won't get it right. The brain needs a body which interacts with a physical environment in order to show intelligence. A physical environment follow the laws of physics which is why speed _is important_. You can't speed up the physical laws. In 1986 Rodney Brooks published a paper called A Robust Layered Control System for a Mobile Robot in order to get AI research on the right track. He was only partly successful, but at least at MIT where he is now the leader of the AI lab, people are doing interesting research now. Another good pointer is his resent book Flesh and Machines: How Robots Will Change Us. More information on embodied intelligence - a brain needs a body - can be found in the very good by Rolf Pfeifer (leader of the AI Lab in Zurich) called Understanding Intelligence". Please stop making a fuzz about talkbots and stuff like that, it's really not that interesting.