Keith observed,
we already have a massively
paralleled computer system. It's called the Internet. The
trick is to communicate with this vast planetary intelligence...
It does already have a rudimentary ear and mouth, some call it google.
Let evolution work on this system a bit more... looks promising.
Yes & not-to-worry, I am not going to drag-out and belabor this general subject matter much longer, as it is only moderately related to future energy, but this week in May was particularly active with important announcements, and it seems that many vortexians are mildly interested (or more) in where it is all leading.
Keith is correct about the larger issue of the importance of evolution, acelerated evoltion, on several fronts - including the internet itself - which is like a "global mind". William Gibson was right-on. It is also important and maybe even strikingly important that recently it was announced that Cyc, the closest thing to real AI on the planet now, will be "plugged-into" to the internet... yes, as in doing its own surfing (augmented by a large staff of humans). It will be programmed to "learn" from the net, but not yet to "contribute"... but how long will it take for Cyc, operating 24/7 to "learn it all" (even the BS)? You may have seen that TV commercial where the frustrated web-surfer finally gets to the "end of the internet."
Sony, IBM and Toshiba have now followed MS and released data about the so-called "Cell chip" that will be able to carry out 16 trillion calculations per second (teraflops). Wow, a power of four increase in cheap-power, and in just a few days. The chip will be made of 9 different processing cores which work on tasks together (parallel processors, but not massive). The three firms have been working on the chip since 2001, and fuller details will be released in February 2006 at the International Solid State Circuits Conference, prior to its release a few months later.
It has been mentioned that because these chips are dedicated to raw video speed, like creating the giga-polygons/sec necessary for seamless movie quality video, that their raw numbers are not very meaningful as to "real" human-like intelligence... and there is (some) truth to that, although less than on the surface. As it turns out, raster processing, which is a large part of video, is very similar to both speech recognition algorithms and parsing algorithms. Not the same but very similar in being a 'reversed raster.'
IBM has been involved speech recognition since day one. Not surprising since IBM has always been "market-driven" as opposed to technology-driven, and CEOs (who are mostly all very tight with big blue) have always wanted to be able to hire secretaries for something other than their transcription speed - the joke being that the prefered "job requirement" does sound a lot like dictation.
It would not surprise anyone, that in addition to the video capabilities, IBM also will have the speech recognition problem nearly tackled by 2006. But caveat - many futurologists have been overly optimistic about that subject (seamless speech recognition) before. The next year will be particularly "telling" in regard to that all-important stepping stone for "strong AI."
The three firms claim that the "Cell chip" (why don't they just go ahead and call it a neo-ron ?) will be10 times more powerful than Pentiums and capable of handling 16 teraflops. My feeling is that 100 of these chips and the supporting hardware and memory, will end up being "human equivalent" in processing power for most tasks and way,way ahead in others, like math.
Now when you think about how complex a single floating point operation (flop) is... that is, for the human brain to handle, even one per second for long periods - this cannot be done by any human. But this is not apples-to-apples, as Terry and others complain. True, but futurologists... and most of us on Vo are in that category, if not by occupation, then by avocation... and most of us should be more interested in the **trend** rather than in the exact cross-comparison.
The next few years are going to be most interesting... for any number of reasons, and if nothing else, as far as the "new energy" goes, it looks exciting. We can see that there will be no problem whatever in using the power of an XBox to control the parameters of an energy cell which may be extremely dependent on operating within extremely narrow overlapping parameters involving heat, pressure, frequency, voltage & current and in several separate linked systems, such as high pressure on one side of a membrane and near vacuum on the other side. That kind of control may require adding a few DSPs to the "neo-ron chip" array. But eventually, in order to harness LENR, it may well be necessary to have this kind of nanosecond level of control - especially if LENR is indeed a "QM enhanced-probability regime," as many suspect.
With QM, as it now seems, the generaliztion is - when you can "control time," as it were, then you can defeat any external restriction - such as the so-called second "law" (or generalization) of thermodynamics.
Jones
