Just finished Hans Moravec, Robot: Mere Machine to Transcendental Mind. The book starts with a short history of robots and AI. There is a chapter on the computing and the mind where he predicts that we need something like 100 million MIPs (and a comparable amount of memory) to match the human mind’s processing. He answers critics to the effect that AI was oversold by arguing that AI has been stagnant partly because they were not using fasther computers. He then extrapolates from advances in robotics to science fiction claims about transcendental minds. I have to admit I started skimming when he went off the deep end – he may know his science but he isn’t that good on the sci fi. The chapter reviewing Turing’s responses to claims that artificial minds would not be possible is a good review of the arguments, but the rest of the book is a poor version of Kurzweil’s Age of the Spiritual Machine. Read on for quotes…
“I consider the development of intelligent machines a near-term inevitability.” (p. 13)
“by performing better and cheaper, the robots will displace us from existence. I’m not as alarmed as many by the latter possibility, since I consider these futures machines our progeny…” (p. 13)
“It is the ‘wild’ intelligences, however, those beyond our constraints, to whom the future belongs.” (p. 13)
“I see robotics progress as roughly recapitulating the evolution of biological minds…” (p. 24)
“Animals learn individually, but robot learning can be copied from one machine to another.” (p. 55)
“Kasparov reported signs of mind in the machine.” (p. 67) The machine was Deep Blue and it is interesting that Kasparov began to think he was playing a mind. An indication of mind as emergent property? As Moravec points out, engineers will think of these as computers, but from the outside they will appear to be intelligent, in an alien way.
On page 77 Moravec claims computers will emerge with our values and be decent. Why? Why wouldn’t they be vicious – trained for war? Later he says they will only initially share our values.
“The later robot generations will have language ability, but there seems little point in programming them to reason by talking to themselves.” (p. 115) The point is that they won’t need to conduct internal dialogues (soliloquies or meditations). But these internal dialogues are how we model or imagine conversations with others – why would AIs limit themselves to modelling only non linguistic reasoning. Interesting point.
“Given fully intelligent robots, culture becomes completely independent of biology.” (p. 126) This would allow accelerated evolution without training. Culture has allowed us to evolve without having to change our DNA, but still having to train. Without needing to train or genetically change robots could evolve at the speed of thought.
On page 133 he predicts the end of capitalism since robot industry would not waste money paying owners. Neat point.
Hans Moravec Robot: Mere Machine to Transcendental Mind Oxford: Oxford University Press, 1999.
Comments are closed.