In her famous description of intellectual awakening, Helen Keller first felt water running over one hand, then recognized a pattern of shapes in the other. After merely associating the two, she grasped the shapes and association in fresh light: the shapes signified something, and she understood water through a proper concept. “Suddenly I felt a misty consciousness as of something forgotten—a thrill of returning thought; and somehow the mystery of language was revealed to me.”
Keller’s full account is a most remarkable description of someone activating the power of conceptual awareness, with accompanying intellectual wonder. Other animals experience sensation, recognize patterns, even associate one with another. But only the human animal experiences one sensible object as a symbol, a vehicle of meaning, and through it grasps an intellectual object that lies beyond sensation.
Can any of that be replicated in silicon, circuits and code?
Some people seem to expect that “artificial” intelligence will be a genuine kind of intelligence, like synthetic diamonds are a kind of diamond. My thesis is that artificial intelligence cannot be a kind of intelligence: it must be to real intelligence only as artificial flowers are to real flowers.
Can we gain insight into flowers by mimicking them in silk? Yes. Here I agree with Stokes. He clarifies that the project of technologically imitating behavior associated with intelligence will force us to think about what we mean by intelligence. We might become more attuned to the workings of our minds, develop vocabulary for new analogies, signify new concepts reflecting greater understanding.
But Stokes and I also seem to agree that the AI project neither presumes nor seeks to capture the nature of human intelligence, any more than artificial flowers capture the organic essence of living flowers.
That was the thesis of my initial post: that the project of imitating (some functions of) human intelligence cannot be confused with the activity of understanding the nature of human intelligence.
As far as I can tell most people agree on this. We know the AI chatbots that seem “insightful” and “articulate” aren’t actually thinking or communicating. (One of their benefits has been to expose how mindless and mechanical—how literally thoughtless—much human-generated text is!) Those few who are tricked aren’t really taken seriously. (Months ago a Google employee believed he was conversing with a conscious chatbot; widely mocked, he was quickly fired.) So yes, we can gain some insight into the nature of human intelligence through the AI project—because that project at best only simulates intelligent human behavior, it doesn’t generate or recreate it.
What is the difference? The very asking of the question calls forth a mode of inquiry beyond the project of AI: a more speculative discipline, seeking not means to a practical end but insight into the natures of things. Far from blurring the line between man and machine, the wondrous AI project makes us more aware of what sets apart the animal that wonders, who seeks the delight of mysteries revealed.