(These are excerpts from my book "Intelligence is not Artificial")
How NOT to Find a Breakthrough
Don't ask me what the breakthrough will be in A.I. If i knew it, i wouldn't be wasting my time writing articles like this one. But i have a hunch it has to do with recursive mechanisms for endlessly remodeling internal states: not data storage, but real "memory".
For historians a more interesting question is what conditions may foster such a breakthrough. In my opinion, it is not the abundance of a resource (such as computing power or information) that triggers a major paradigm shift but the scarcity of a resource. For example, James Watt invented the modern steam engine when and because Britain was in the middle of a fuel crisis (caused by the utter deforestation of the country). For example, Edwin Drake discovered petroleum ("oil") in Pennsylvania when and because whale oil for lamps was becoming scarce. Both innovations caused an economic and social revolution (a kind of "exponential progress") that completely changed the face of the world. The steam engine created an economic boom, reshaped the landscape, revolutionized transportation, and dramatically improved living conditions. Petroleum went on to provide much more than lighting to the point that the contemporary world is (alas) addicted to it. I doubt that either revolution would have happened in a world with infinite amounts of wood and infinite amounts of whale oil.
The very fact that computational power is becoming an infinite inexpensive resource makes me doubt that it will lead to a breakthrough in Artificial Intelligence.
Water power was widely available to Romans and Chinese, and they had the scientific know-how to create machines propelled by water; but the industrial revolution had to wait more than one thousand years. One reason (not the only one but a key one) why the Romans and the Chinese never started an industrial revolution is simple: they had plentiful cheap labor (the Romans had slaves, the Chinese emperors specialized in mobilizing masses).
Abundance of a resource is the greatest deterrent to finding an alternative to that resource. If "necessity is the mother of ingenuity", as Plato said, then abundance is the killer of ingenuity.
We live in the age of plentiful computational power. To some observers this looks like evidence that super-human machine intelligence is around the corner; to me this looks like evidence that our age doesn't even have the motivation to try.
My fear, in other words, is that the current success in "brute-force A.I." is slowing down (not accelerating) research in higher-level intelligence (the real meaning of "human intelligence"). If a robot can fix a car without knowing anything about cars, why bother to teach the robot how cars work? The success in (occasionally) recognizing cats, beating go/weiqi champions and so forth is indirectly reducing the motivation to understand how the human mind (or, for that matter, the chimp's mind, or even a worm's mind) manages to recognize so many things in split seconds and perform all sorts of actions. The success in building robots that perform this or that task with amazing dexterity is indirectly reducing the motivation to understand how the human mind can control the human body in such sophisticated manners in all sorts of situations and sometimes in completely novel ways.
Bill Joy wrote that "The future doesn't need us", but maybe it's the other way around: we will not need the future if soon the present will give us machines that can do everything we need.
Back to the Table of Contents
Purchase "Intelligence is not Artificial"