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 : Idea Of The Day -- Ignore unavailable to you. Want to Upgrade?

To: IQBAL LATIF who wrote (39986)	6/20/2001 1:45:29 PM
From: IQBAL LATIF	Respond to of 50167

And that leads to this <Our specific cognitive architecture and development plan forms our basis for answering questions such as "Will transhumans be friendly to humanity?" and "When will the Singularity occur?" At the Singularity Institute, we believe that the answer to the first question is "Yes" with respect to our proposed AI design - if we didn't believe that, the Singularity Institute would not exist. Our best guess for the timescale is that our final-stage AI will reach transhumanity sometime between 2005 and 2020, probably around 2008 or 2010. As always with basic research, this is only a guess, and heavily contingent on funding levels. The impact of even a single transhuman AI would be tremendous. The basic dynamic of the Singularity is positive feedback - smarter minds are better at inventing even smarter minds. Intelligence creates technology which enhances intelligence. An AI which can optimize its source code - rewrite the underlying computer program - will be able to think faster, perhaps fast enough to spot new avenues for optimization; an artificial mind can learn new methods for learning. Even in the very early stages of AI development, such self-improvement is likely to be crucial - this is why the Singularity Institute's Bylaws cite "Artificial Intelligence capable of self-modification, self-understanding, or self-enhancement" as a specific objective. While the process of self-enhancement must eventually reach a limit for any given piece of hardware, this bottleneck only exists until the transhuman AI can assist in the development of new hardware. Dr. Eric Drexler has published a thorough proof-of-concept (Nanosystems, 1992) for molecular manufacturing - material science capable of synthesizing complex objects from individual atoms, or of manipulating materials to within atomic (tenth-of-a-nanometer) precision. Although the nanotechnology described in Nanosystems is extremely primitive - Dr. Drexler was interested in feasibility, not optimal designs - it nevertheless suffices to build a small box, weighing about a kilogram, which can produce another small box in less than an hour. Or produce any other one-kilogram object, from diamond to hamburger, given a complete specification and a feedstock containing the needed atoms. Nanosystems also describes a "rod logic" nanocomputer which uses moving diamondoid rods instead of electronics; even operating at the speed of sound, a one-kilogram nanocomputer could perform 10^21 operations per second. (By comparison, the usual estimate for the human brain is on the order of 10^17 operations per second.) And such a nanocomputer is far from optimal, cited by Drexler only to prove feasibility and because it was easy to model; a real nanocomputer would use electronics, probably operating somewhere around 10^25 or 10^26 operations per second. This is more than enough to run an AI, even a transhuman AI, at a millionfold or greater subjective speedup. Nanotechnology is a field that has matured considerably in the past decade. Developments include devices that can pick up and drop individual atoms, positioning systems with nanometer-scale precision, self-assembling structures, artificial proteins, fullerene "tweezers", and advanced modeling systems for molecular dynamics. >>