But are computers like people? I have already suggested that they embody many of our assumptions, and now I have been drawing an analogy between human-computer interactions and person-to-person friendships. Does this mean I buy into the first view stated at the outset the view that it is natural for us to respond emotionally and intellectually to intelligent machines as if they were persons?
Not at all. If we cannot accept the ideal of machine invisibility in any absolute sense, neither can we accept the ideal of machine personality. The problem with both ideals, at root, comes from the same source: a failure to reckon adequately with the computer as a human expression. The two ideals simply err from opposite and complementary sides: a striving for invisibility encourages dangerous neglect of the tendentious expressive content we have vested in the machine; on the other hand, trying to make the machine itself into a person mistakes the machine-as-an-expression for the humans who have done the expressing.
I am convinced that rising above these complementary errors would strikingly transform the discipline of artificial intelligence, not to mention the entire character of a machine-based society.
The world is full of human expressions that are, in part, manifestations of intelligence. The intelligence is really there, objectively, in our artifacts in the sound waves uttered from our larynxes, in the pages of text we write, in the structure and operation of a loom, automobile, or computer. It is impossible to doubt the objectivity, given that anyone who attends to these artifacts can to one degree or another decipher the intelligence that was first spoken into them. We do this all the time when we read a book. Something is there in the physical pages of the book giving me access to an elaborate world of inner, intellectual experience. That's just the nature of the world through and through: it is receptive to, and a bearer of, the intelligence we imprint upon it.
But, as I pointed out in the last chapter, it is nonsense to mistake the artifact for the artificer, or the intelligence spoken into the world as product for the speaking as productive power. The endemic preoccupation with the question whether computers are capable of human-like intelligence is one manifestation of this confusion. But if we are willing to step back from this preoccupation and look at the computer in its full human context, then we can gain a much more profound appreciation of its intelligence. At the same time, such a contextual approach can guide us toward a more balanced view of the human-machine relationship.