Three weeks from today, on October 25, the Singularity Institute for Artificial Intelligence will host their annual Singularity Summit. For all their futuristic talk and obvious intelligence, though, I'm afraid that this crowd is actually completely missing the boat—a singularity is very likely to happen right under their noses, before they even recognize it.
Not because I doubt that AGI is going to happen, mind you, or even that their timelines are too optimistic. "Within a few decades from now" doesn't seem unrealistic to me, in light of all the same trends they cite in support of their prediction: exponentially improving computational power; and of scientific knowledge, including in particular in the fields of biology and neuroscience. It does seem to me it's only a matter of time before, if not "pure" AGI, then at least either uploading, and/or the cyborg approach to super-human intelligence, will become a reality.
The problem is, AGI is not the definition of a singularity. Not, at least, according to the coiner of that term, Vernor Vinge. This much is recognized by at least one of the SIAI luminaries, Robin Hanson, whose article Economics of the Singularity, in the IEEE Spectrum Special Report: The Singularity issue of June of this year, explained that the coming singularity is not the first time such an epochal event will occur, but is rather at least the third time in human experience. The first clear precedent is the shift from a hunter-gatherer lifestyle to agrarianism, and the second is the industrial revolution. In both cases, a discontinuity appears that is so abrupt, and so dramatic, that the previous history provides essentially no basis at all for predicting beyond that transition, effectively creating a startlingly opaque prediction horizon, seemingly impenetrable by any other predictive technique than prophetic vision.
But in that same article, Hanson spends surprisingly little time entertaining any alternative hypotheses for what could produce the next singularity, other than AGI. He explains why AGI would clearly produce a singularity, and I agree completely. But is that really the only horse in the race, today?
Another article in the same IEEE special report appears to provide the answer. In Ray Kurzweil and Neil Gershenfeld: Two Paths to the Singularity, Tekla S. Perry notes two apparently oppositional visions of the future: "Gershenfeld's future in which computers collapse and simply become part of reality, and Kurzweil's future in which reality as we know it collapses and simply becomes part of computers." Gershenfeld, you may recall, discussed and demonstrated Personal Fabrication at the April 2005 Future Salon, 3.5 years ago this month. It would seem to me that Gershenfeld's horse is at a steady gallop toward the finish line, while Kurzweil's has yet to leave the starting gate! So why is the Singularity Summit so preoccupied with AGI?
Well Hanson, for one, simply dismisses the would-be contender in a solitary sentence, without so much as a footnote to back it up.
Even so extraordinary an innovation as radical nanotechnology would do no more than dramatically lower the cost of capital for manufacturing, which now makes up less than 10 percent of U.S. GDP.
Is this the same Hanson who just explained that "nobody" saw the agricultural and industrial revolutions coming? Couldn't it just as plausibly have been supposed, before the fact, that agriculture "would do no more than" incrementally increase the bounty extracted from nature, or that industrialization "would do no more than" (ahem) lower costs in industries which made up a small fraction of the economy? We now know that the change actually lead to in each case was vastly greater than virtually anyone imagined. Shouldn't we, perhaps, put a little more thought into what might eventually be the full consequences of something so novel as "radical nanotechnology", before dismissing them as virtually insignificant?
After all, it's only been eight years since we all experienced a remarkable event that I would call (with hindsight) a false singularity, a short period of time in which the conventional wisdom became that all the rules had changed and growth was going to absolutely explode, only to soon see everything stall and quickly return to a state of business-as-usual, same old same old. I'm referring, of course, to the Dot-Com Bubble. Obviously, the impact of the Internet was not yet adequate, in the year 2000, to support a sustainable, radical shift in the way the entire global economy operates. What happened then wound up amounting to only an evolution, not a revolution. If you were to try to paint a picture, though, of what the early stage of a genuine singularity might look like, the Dot-Com euphoria would be a pretty reasonable guess. And indeed, the perceived close parallels to events during a couple of the major inflection points of the previous singularity, the industrial revolution, became a large part of the cause of the "irrational exuberance" we were all witness to.
But the bubble burst, no singularity happened. Why not? What turned out to be the sticking point, the insurmountable hurdle that tripped it all up, do you remember? It was the stubborn immobility of "bricks and mortar", the failure of the "new economy" to reshape and streamline the physical world as quickly and easily as the online, virtual world. The online world actually continues to churn and evolve at a breakneck pace, despite the Dot-Com Bomb, and despite today's credit crisis, and I think can confidently be predicted to continue to do so, through Web 2.0, to Web 3.0, and far beyond. The only thing that didn't happen was escape from the drag of the physical, industrial-age foundations of our economy—the disconnect between bits and atoms.
So what happens, now, if Gershenfeld and a host of other researchers and engineers are indeed about to do for moving matter, what the Internet did for moving bits? Can we dust off all the Dot-Com prophecies, yoke them up to the new technology, and buckle our seat-belts again for the escape-velocity rocket-ship ride we were braced for back in 2000?
Well, I'm not going to assert unequivocally yes, that "digital matter" is about to launch us off on a whole new socioeconomic trajectory, a full decade or more before the development of an AGI. There are still a number of very daunting hazards for us to avoid, foremost of which in my mind is the very tricky step of completely reinventing security for an age of "knowledge-enabled mass destruction". (If this is of concern to you too, BTW, then don't miss Christine Peterson's presentation at the next Bay Area Future Salon, Friday, October 17.)
But one thing is pretty clear: if you have questions about the viability of this Drexlerian path to the next singularity, the so-called Singularity Summit is not the place to take them.
(The place to take them instead, it seems, will be Convergence08, November 15-16 at the Computer History Museum. SIAI will be there, too, but they're only one of several supporting and cooperating organizations, including Foresight Nanotech Institute, and one active topic of discussion will clearly be in what order various disruptive technology(s) are most likely to appear.)