On Monday 19 March 2007 17:30, Ben Goertzel wrote: ... > My own view these days is that a wild combination of agents is > probably not the right approach, in terms of building AGI. > > Novamente consists of a set of agents that have been very carefully > sculpted to work together in such a way as to (when fully implemented > and tuned) give rise to the right overall emergent structures.
There is one way you can form a coherent, working system from a congeries of random agents: put them in a marketplace. This has a fairly rigorous discipline of its own and most of them will not survive... and of course the system has to have some way of coming up with new ones that will. And before any of that happens, the agents have to be able to communicate, if not in a unified language, a language with one universal concept: money. (It doesn't matter if I don't understand your ad -- I'll just do business with whomever I do understand. But the money I pay them has to be good for them to pay the next guy, whom I don't understand.) One metaphor I've found helpful thinking about the mind is to realize that it was formed by evolution and thus probably has many of the same higher-level architectural properties as the body. The body has a relatively straightforward higher-level structure (e.g. bones and muscles in the hundreds) that gets very hairy as you get down to the biochemical level. I'm sure the brain does too. What we're all hoping here is that we can capture the essence of intelligence at the high level, like simulating the skeleton and muscles in a walking robot, while being able to substitute the hairy substrate of current-day computers for the hairy parts of the neural architecture. The body has major organs that are very special-purpose -- heart; lungs, liver -- but also ones that are as general-purpose as anything nature ever made (hands). A big chunk of the brain clearly goes to forming and maintaining a world model, Brooks and the robo-know-nothings to the contrary notwithstanding. Think of that as like legs -- special-purpose in one sense, but general in a higher one (they're mainly for locomotion, but they'll take you anywhere). There's also a "hands" part of the mind -- that which lets us grasp ideas. Most the special-purpose parts can be finessed one way or another: wheels for legs, wires for veins, what have you. But it needs hands that are like hands. Show me a robot that can do cat's cradle, make a whistle of its fists, do shadow pictures on the wall, and make an omelette from whole eggs and you've got something that's not just a toy. Same with the AGI. You can finesse most of the special-purpose parts -- just build something to do the function, which is basically what engineering is about. But universal machines are an arcane and little-studied phenomenon -- one can't even say field. There's no design methodology for a machine that can do anything. That's what hands are, and we've got something like them in our minds. Josh "Greenspun's Tenth Rule of Programming: any sufficiently complicated C or Fortran program contains an ad hoc informally-specified bug-ridden slow implementation of half of Common Lisp." - Philip Greenspun ----- This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?list_id=303
