Re: Robot Evolution
- From: "feedbackdroid" <feedbackdroid@xxxxxxxxx>
- Date: Sun, 19 Nov 2006 17:59:38 -0500 (EST)
RAGLANDMYCOOL@xxxxxxx wrote:
Note: As I read this, assuming what Hans Moravec says is accurate, I
wondered how and if we can genetically keep up with the robot
revolution. Certainly a fourth generation and beyond robot would exceed
us at our current genetic "level". There is talk of the doubling of
power of computers "Moore's Law and the increase in complexity of
robotics but currently nowhere is there a genetic equivalent.
Moravec has been saying pretty much the same thing for 20 years, or
more ....
Mind Children: the future of robot and human intelligence, 1988
http://www.frc.ri.cmu.edu/~hpm/hpm.pubs.html
He seems to think the main thing holding back robotics and AI is the
amount of computing power available, and that the way to produce
human-level machine intelligence is to continue as we are, only do it
with more powerful computers [unless he's changed his viewpoint
recently .... seems not] ....
http://www.frc.ri.cmu.edu/~hpm/project.archive/robot.papers/2004/Predictions.html
He seems to always talk about MIPS and MOPS, etc, rather than concepts
and algorithms.
However, after several years of monitoring various AI forums, I don't
think anyone has a real idea of how to produce "general AI". It's not
clear that the concepts of today will simply "scale up" to human
intelligence levels, simply by using more powerful processors, as they
are mainly domain-specific solutions, and not general AI solutions. And
it's not clear that anyone else is on the right track either, at least
given comments on the forums. Mostly people repeat past arguments
interminably, state opinions [like I'm doing here, of course], or
voicing sameold platitudes - [open to revision, it someone actually
steps forward with a critical answer]. It's really kind of a shame.
If you've read about the idea of "The Singularity" by Vernor Vinge, Ray
Kurzweil, or Damien Broderick, you know they talk about a largely
unpredictable future, exactly for the reason stated in the first
sentence by the original poster, namely the problem of "keeping up"
with the rate of acceleration of technology. To wit, advances in
biotechnology, AI, and nanotechnology will make a future much different
from what we see today, since it's so difficult to predict where the
combination of the 3 will take us, especially given their supposedly
"exponential" growth.
However, for my money, AI is lagging behind the others, especially
biotechnology. My opinion is that more processing power is not going to
solve the problem of "general AI". Also, I'm not much worried about
humans genetically keeping up with the rate of advancement of robots,
since - at least today - biotechnology and molecular biology are
advancing at a rate that seems to me to far surpass the genesis of new
ideas in AI. This will continue, at least until people break out of the
ruts of endessly revisiting the old ideas, instead of finding new ones.
So, Michael, by time Moravec's robots ever get here, your grandchildren
or great-grandchildren will probably be manipulating their personal DNA
using at-home kits sold by Edmund Scientific. [I say this only
half-jokingly]
.
- Follow-Ups:
- Re: Robot Evolution
- From: Tim Tyler
- Re: Robot Evolution
- Prev by Date: Re: What Use is Religion
- Next by Date: Re: Darwinian Language (Long Article)
- Previous by thread: Re: What Use is Religion
- Next by thread: Re: Robot Evolution
- Index(es):
Relevant Pages
|
|
