Most helpful critical review
19 of 20 people found the following review helpful
A warning, but one to examine closely
on May 26, 2014
Del Monte examines the current state of Artificial Intelligence (AI), and extrapolates to the day when machines have thought capacity - not just computing ability, but consciousness - at or beyond humans'. The outlook seems grim. Why, after all, would the automated factories, mining and power generation, and machine intelligences bother to keep us around? Especially when we might present a threat to their existence. We might expect no better treatment at their hands than the smallpox virus received at ours.
It's the job of futurists, technology forecasters, and even SF writers to ask the "what if" questions and to come up with answers, however speculative or tentative. It's also the job of informed readers to examine the reasoning provided and its basis in current fact - and that's where I think other lines of reasoning than Del Monte's might apply. First, there's the notion of a human brain's computing capacity. Reasonable estimates put current supercomputers' capacity within an order of magnitude of a brain's, so raw processing ability might not be the issue it has been. But biological, neural computing differs in many ways from silicon's. Neurons have hundreds or thousands of interconnections but limited processing ability per individual cell. Current supercomputers offer billions of arithmetic operations per second in each of their thousands of processors, but typically have only a dozen or fewer connections between them. Yes, programming can let any unit contact any other, but communication costs dominate actual computation quickly as the mismatch between physical and logical computing structures widens. And, depending on how you define terms, a few cubic inches of muscle tissue has huge computing power in its biochemical and DNA-based regulatory cycles - but don't expect a muscle tissue culture to beat you at chess.
Del Monte frequently cites Ray Kurzweil as a leading thinker about AI, and presents Kurzweil's notion of "The Singularity," that indefinable event horizon beyond which AI expands without limit. His description of Kurzweil notes that he's "Inventor of ...the first music synthesizer" (p.84), a claim that Robert Moog's heirs would surely find open to debate. And, citing Kurzweil's own count, ascribes to him a 94% accuracy rate in technological predictions. For such a remarkable claim, I would hope for independent accounting - self-reporting tends toward the unreliable in every context I've seen it used. Del Monte notes that, as of 2010, "software to emulate human thinking ... does not exist." This report, coming so soon on the heels of many historical, over-zealous reports he mentions, leaves me thinking that AI really is the wave of the future - and could be for a very long time to come.
Although I agree with many of Del Monte's assertions and projections in a broad sense, there's too much (a lot more than I've mentioned) that weakens the foundations of his projected timelines. And, I am not convinced that a competitive, even adversarial relationship with metal minds is the only one, or even the most likely. Despite many misgivings, though, his forebodings seem well worth consideration. It will be a very new thing to interact with minds both like and unlike our own. In the best of worlds, that conversation would help us understand more of what makes us truly human at the same time we learn what makes them truly other.