Thaths <[EMAIL PROTECTED]> writes: > On Wed, Jun 18, 2008 at 10:55 AM, Perry E. Metzger <[EMAIL PROTECTED]> wrote: >> What if, for example, aging is "cured" in the next 20 years and the >> population decline halts because of that? What if by 2050 we have >> robots as capable as people capable of managing most of society's >> needs? What if we have robots that are more capable than people and >> humans end up as their treasured pets? What if half the population >> uploads into machines and the concept of population becomes utterly >> incomprehensible? > > What is the current thinking among Singularitarians about how the > world will be powered post-Singularity?
I'm not a "singularitarian" in that it isn't a pseudo-religion for me. For some people, the idea is more or less equivalent to the Christian idea of the rapture, and I don't buy in to that -- I have no particular expectation that the changes will be good or bad from my personal perspective, and I don't spend all my time dreaming of how to immanentize said eschaton. (There are people who do -- google, for example, for Eliezer Yudkowsky.) None the less, I think the sort of events Vernor Vinge described in his famous essay (see http://www-rohan.sdsu.edu/faculty/vinge/misc/singularity.html ) are pretty much inevitable. However, back to answering your point: pretty much by definition we can't say much about what the world would look like after a real strong-AI-singularity, for reasons I'll describe below. We know that the only really practical sources of energy out there in the big sense are fusion (either artificial or collecting the output of stars) and fission, so we can expect those to be in wide use because chemical bonds just don't yield very high amounts of energy. Beyond that, I don't think we can predict much in the way of detail about future power production and distribution systems. So, why do I say "we really can't know"? Consider the following simple gedankenexperiment: Eric Drexler describes in some of his writings a gedankendesign (not a design anyone would ever really use, but a proof of concept) for something essentially equivalent in computation to a human brain, only about 1e6 times faster, approximately 1cc in size. (It would have quite high power and cooling needs, but that's not germane to the current discussion, especially since a real design doubtless won't look like Drexler's proof of concept.) Say we were dealing with AIs or uploads that had only human equivalent intelligence but which, operating on such platforms, run a million times faster . One such a creature could do as much thinking-work in 20 days as a person could in 50,000 years. Ponder that last sentence for a few moments before going on. Now, such a creature also has a feature that humans (at least current ones) don't have, which is that it need not be raised and educated over a course of decades but can simply be copied. Given a couple of prototypes, a large group of such beings could be built (or more likely could build each other) and might make astonishingly rapid engineering progress in a period that normal people would barely notice. One of the engineering tasks they could address would be improving their own design -- a process that might seem painstaking and long to them, but which could happen in mere minutes per iteration from the point of view of a biological human. So, I don't think you can reasonably ask "what sort of energy sources would be in use" after ten years of such engineers were at work. Indeed, I don't think you can say anything much about what the world would look like after ten years, except to say that it is supremely unlikely that the laws of physics could be changed so they will still pose a constraint. Other than that, I suspect nearly anything else that is not inconsistent with the laws of physics might happen. Now, all of this is conditional on having either human equivalent AIs, or uploaded humans capable of operating on such substrates. Even given the hardware, it may be the case that solving the software problem will require more than 20 or 30 years to accomplish. Maybe it will take 40 years or 60 years or what have you. It really doesn't matter much in principle if it takes 150 years. The result is effectively the same. Regardless, looking at a complete revolution in civilization as we know it, one far greater than the appearance of Homo sapiens, and at said revolution happening a blink of an eye from now if you have a "historical" perspective and not a short term perspective. We've been around for 50 or 100 millennia -- a century or two out is nothing on such a time scale. Much more likely it will be possible within a few decades, but as I said, it matters little if I'm off by a factor of 5 or 10. > Considering recent perturbations in oil prices and effect it has > been having on other things, I am increasingly curious about the > amount of energy that can safely and sustainably harvested from this > planet and how efficiently this energy can be used. Well, locally, you've got at least the output of the sun on the earth to work with, and that's something like 10,000 times the current energy usage on earth. However, if you moved into space (probably the smart thing to do), you've got as much of the sun's output to work with as you like provided you can build enough collectors, and that's a truly vast amount of energy. Perhaps the more efficient solution is "lifting" most of the hydrogen out of the sun and running it through artificial fusion reactors in order to husband the resource more efficiently rather than leaving a "natural" star running indefinitely -- hard to know, but presumably our mind children will figure it out. Perry -- Perry E. Metzger [EMAIL PROTECTED]
