Reading :: The Age of Spiritual Machines

The Age of Spiritual Machines
By Ray Kurzweil

In one part of his popular futurist book The Age of Spiritual Machines, Ray Kurzweil contemplates spiritual experiences, which are currently the provenance of human beings but which, he says, will eventually occur in the intelligent machines we will create. "Spritual experiences are not all of the same sort," he points out, "but appear to encompass a broad range of mental phenomena. The ecstatic dancing of a Baptist revival appears to be a different phenomenon than the quiet transcendence of a Buddhist monk" (p.151).

A dancing Baptist? Apparently Kurzweil is not that familiar with Baptists!

I bring up this point not to be an ankle-biter, but to illustrate what I think is a recurring problem with the book. Although Kurzweil is intelligent and widely read, I don't think he spends much time with the details. In person, I imagine him to be the kind of guy who excitedly finishes your sentences, and doesn't notice that he's finishing them in ways that you never would. The details, of which there are many, serve to propel the overall narrative rather than to constrain it.

And what's the narrative? We get a sense in the first chapters, in which Kurzweil postulates teleological laws of the universe. He notes that major milestones in the development of the universe seem to be spreading out: it took trillionths of a second after the Big Bang for gravity and subparticles to emerge, about a minute for atomic nuclei to form, and 300,000 years for those nuclei to capture electrons, etc. On the other hand, evolution's rate increases exponentially, with evolutionary milestones happening at shorter and shorter intervals. On these two curves, he maps Moore's Law, the exponential increase in computing power, as well as the fast pace of milestones reached by a developing human fetus. From all of this, he extrapolates a universal law that guarantees and compels technoevolutionary change, the Law of Accelerating Returns:

Moore's Law came along in 1958 just when it was needed and will have done its sixty years of service by 2018, a rather long period of time for a paradigm nowadays. Unlike Moore's Law, however, the Law of Accelerating Returns is not a temporary methodology. It is a basic attribute of the nature of time and chaos -- a sublaw of the Law of Time and Chaos -- and describes a wide range of apparently divergent phenomena and trends. In accordance with the Law of Accelerating Returns, another computational technology will pick up where Moore's Law will have left off, without missing a beat. (p.33)

Got that? Kurzweil has selected arbitrary milestones out of a much larger set of possible milestones, noted an apparent pattern, named that pattern a Law, and used it to make confident predictions about technological development. This is the same sort of reasoning that underpins numerology and some conspiracy theories. Like numerologists and conspiracy theorists, Kurzweil is very intelligent and very good at finding patterns, but not so good at recognizing when those patterns emerge from overly selective filtering of data points.

Treating the Law of Accelerating Returns as a proven law or destiny, Kurzweil then goes on to demonstrate that this Law will inevitably lead to intelligent machines -- machines that are each more intelligent than all human brains combined by 2060 -- and to argue that this change is the natural result of human evolution.

Kurzweil also makes the assumption that cognition goes on entirely inside the skull, an assumption that drives his predictions about machine intelligence, since in that case all one has to do is to reproduce the computational power of the brain. To be fair, many cognitivists rely on this same assumption. But sociocognitivist and postcognitivist perspectives (such as distributed and situated cognition) present some sharp critiques of this assumption.

The result is entertaining science fiction, but not convincing futurism.