the child is mostly concerned with filling pails and things like that. Suppose, though, by some accident, a child got interested in how that pail-filling activity itself improved over time, and how the mind's inner dispositions affected that improvement. If only once a child became involved (even unconsciously) in how to learn better, then that could lead to exponential learning growth.

     Each better way to learn to learn would lead to better ways to build more skills ­ until that little difference had magnified itself into an awesome, qualitative change. In this view, first-rank "creativity" could be just the consequence of childhood accidents in which a person's learning gets to be a little more "self-applied" than usual. (6) If this image is correct, then we might see creativity happen in machines, once we begin to travel down the road of making machines that learn ­ and learn to learn better.

     Then why is genius so rare? Well, first of all, the question might be inessential, because the "tail" of every distribution must be small by definition. But in the case of self-directed human thought-improvement, it may well be that all of us are already "close to some edge" of safety in some sociobiological sense. Perhaps it's really relatively easy for certain genes to change our brains to make them focus even more on learning better ways to learn. But quantity is not the same as quality ­ and, possibly, no culture could survive in which each different person finds some wildly different, better way to think! It might be true, and rather sad, if there were genes for genius that weren't hard at all for Evolution to come upon ­ but needed (instead of nurturing) a frequent, thorough weeding out, to help us keep our balance on some larger social scale.

Can Computers Choose Their Own Problems?

     Some people even ask "How could computers make mistakes?" as though, somehow, ability to err itself might be some precious gift. There s nothing wrong with seeking for some precious quality, but only some form of quiet desperation would lead one to seek for it in error and mistake. It seems to stem from the misconception that creativity is rooted in some chance or random element that can't be found in any well-defined machine. This is silly, first because machines can simulate random behavior as well as one can want, and, second because it doesn't explain the consistency and coherency with which creative people produce.

Another often-heard speculation: "I can see how a machine could solve very difficult problems that are given to it by someone. But isn't the very hardest and most important problem, really, to figure out what problem to solve? Perhaps the thing machines can't do is to invent their own problems?" This is wonderfully profound and silly at the same

time. Really, it's usually much easier to think of good problems than to solve them ­ though sometimes it is profoundly hard to find exactly the right question to ask. In any case, a culture frames its history of ideas so that the rewards are largest for opening new areas. But the problems inside each subject can be just as hard.

     The reason this speculation is wrong is that, in order to solve any really hard problem (by definition of "hard"), one has to find a way to break it down into other problems that one can solve. Therefore, the ability to invent and formulate new problems must already be a part of being reasonably intelligent. It only obscures the point to argue that those are "only sub-problems." The ability to compose good questions is a requisite of intelligence, not a special sine qua non for creativity.

     Besides, some people, more than others prefer to look outside a present context and ask larger questions like "Am I working on the right problem?" But everyone can do this to some degree ­ and can be worse off by doing it excessively. I see nothing especially mysterious about that inclination to "take a larger view."(7) The interesting problem is less in what generates the originality, and more in how we build control mechanisms that appropriately exploit and suppress it.

     The rest of this essay explains the weaknesses of several other common theories of how machines must differ fundamentally from minds. Those theories are unproved today ­ not because of anything about machines, but just because we know too little about how human minds really work. We are simply not prepared to search for things that we can do but machines cannot. Because of this, we'll focus on a more constructive kind of question: why people are so very bad at making theories of what they can or cannot do!

Can Computers Think Only Logically?

     Our culture is addicted to images of minds divided into two parts. Usually, one mind-half is seen as calculating. logical, and pretty brittle; the other half seems sort of soft and vague. There are so many variants of this, and all so ill-defined, that it's impossible to tell them apart: let's call them Soft~Hard Dumbbell theories:

Logic - Intuition
Spatial - Verbal
Quantitative - Qualitative
Local - Global
Reason - Emotion
Thinking - Feeling
Literal - Metaphorical, etc.

     There's nothing wrong with starting with two-part theories ­ if you use them as steps toward better theories. But

Introduction - Table of Contents

PREVIOUS

NEXT