<br><div><span class="gmail_quote">On 2/14/07, <b class="gmail_sendername">Ben Goertzel</b> <<a href="mailto:email@example.com">firstname.lastname@example.org</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
It is worth distinguishing as a different category, AI systems that aim at general intelligence (in roughly the sense of the g-factor from psychology) rather than achievement in highly specialized domains. I have come to the conclusion that pursuit of general intelligence and pursuit of specialized intelligence are quite different sorts of science/engineering tasks.
</blockquote><div><br>I agree. And given the 3 problems I pointed out earlier I have little hope that a good human level or even better than human level AGI, even a self-improving AGI, would be as good at solving those problems as dedicated hardware and software solutions should be. Given the difference that good solutions to those problems would make I'd rather see the emphasis placed on them than on QC or AGI.
<br><br>One interesting question, IMO, would be what *fraction* of the neurons (or power consumed) by the brain is actually dedicated to intelligent thought vs. what fraction is dedicated to storing and retreiving memory, sensory processing, motor control, maintenance of internal state, etc.? Given the way memory trends seem to be going we are going to cross over the memory storage metrics (W/bit or bits/sec) much sooner than we will cross over the computational metrics (instructions/W or instructions/sec).