Monday, June 23, 2008

No "Singularity" for you.

I wasn't going to even mention the idea of a Singularity, but then the IEEE made a point of dedicating an issue of their magazine to the concept. For those who don't know, the term "Singularity" originates with sci-fi author Vernor Vinge, who has written some compelling novels. Proponents of the concept believe that we live in an era of exponentially accelerating technological change, and that at some point (the Singularity) there will be a complete break in the nature of our species and societies, ushering in what some call a transhumanist future. The technologies typically associated with this idea are (1) Drexlerian molecular nanotechnology, so that we can eliminate scarcity by building anything we want anytime we want via (self-reproducing) nanomachines; (2) immortality via nanotechnological or biochemical control over biological processes that lead to senescence; and (3) strong AI, often including the concept of people uploading their minds to constructed hardware. The thing that continues to surprise me about this idea is that so many people seem to take it so seriously.

Hey, I'm all for optimism, and I'm generally bullish on the future of the species despite current scariness and some scientific arguments, but asserting that we will have a transhumanist utopia in twenty or thirty years is a wee bit of a reach, to put it mildly.


Anonymous said...

(1) and (2) may be overly optimistic but (3) is not for the next 40 years.

At the rate microprocessors and storage are advancing, it's only a matter of time. Furthermore, AI that can pass the Turing test may already be possible on today's supercomputer hardware, theoretically, but no one has tried or devoted the proper resources to it, as far as I'm aware.

Moreover (1) and (2) could be a consequence or byproduct of (3) after a few decades.

thm said...

First observation: every single one in the "who's who" of the singularity, and all but one of the "tech luminaries" asked to comment, is a white male.

There might well be a singularity in our future, but I think these techno-pundits have the sign wrong. James Howard Kunstler calls it "The Long Emergency," and the signs that both Global Warming and Peak Oil are becoming reality are hard to ignore. There's a sort of built-in assumption to all the singularity speculation that we'll continue to be able to get as much energy as we want, anytime we want it, pretty much forever. I'm not bullish on that idea.

Fortunately (in a self-serving way), the work of physicists should be central to finding a technological path to a post-fossil-fuel future, if such a path exists.

Doug Natelson said...

In my post I didn't get into why precisely I think the Singularity idea is bunk (unless you consider the possible global collapse of civilization to be a kind of singularity, though i don't think that'll happen either). Molecular assemblers of the Drexlerian variety are, as far as I know, not possible. You simply can't just grab carbon atoms and build diamonds, for example. There's no physical mechanism that could work to do this arbitrarily. Sure, biological systems are good at building complex structures even out of inorganic substances. I'll be prepared to reconsider my view on this as soon as you can show me genetically tailored bio-goo that can build a simple ratchet and spring by design. Sure, sure, a hyperintelligent AI may be able to figure something out that, at present, we can't fathom, but that's a dodge.

Tweaking biochemistry to drastically alter the human lifespan may be possible. I'll give you that one. I'll also point out that giving quasi-immortality to some subset of the human race is unlikely to bring about a utopia. If anything, I would think that it would be horribly destabilizing to society.

As for AI, it's not at all clear to me that raw processor and storage power automatically hold the key to consciousness. I don't mean anything mystical here - just that really duplicating the learning capabilities of biological systems is a very very hard problem, whether it's handled in hardware or software. Even if you postulate that we can create AIs that past the Turing test, it's not at all clear that this would necessarily usher in a new age. I'm all in favor of working on this area, since it's tremendously interesting. I just think that there's a huge leap from where we are to strong AI, and an even bigger leap from strong AI to strong AI solving major human societal problems.

Anonymous said...


I wonder if there were indivdiauls that thought like you in the 1930's. I wonder how they would have reacted to the suggestion that the sound barrier would be being broken, nuclear energy, relatively routine space travel, little devices you can carry around with you to communicate with others around the globe, etc.

Doug Natelson said...

Physicists in the 1930s would not have been at all surprised by the breaking of the sound barrier (since bullets and whips do it all the time), the existence of atomic energy (the potential of which was realized immediately upon the discovery of radioactivity), space travel (Goddard flew sounding rockets in the 1920s), or cell phones (radios had been around for decades).

I'm hardly a Luddite. I have tremendous faith in our ability to develop advanced technologies. Still, some problems are much harder than others. Molecular nanotechnology, immortality, and strong AI are very very very hard problems. Hey, I'd be happy to be proven wrong by even one of these being done in the next thirty years, but I'm not holding my breath.

Matthew said...

I just have to point this out to anonymous, as it's one of my favorite technology factoids -- not only would scientists of the 1930s be entirely unsurprised by cell phones or the technology that underlies their use, neither would actors or musicians of the day. (Okay, so Hedy Lamarr and George Antheil received the patent on their frequency-hopping device in 1941. Bit of a fudge there. Still, so cool!)

chad said...

Every time I hear about the Singularity I'm reminded how well flying cars and nuclear fusion have worked out.