We've all had the experience of talking in a chatroom to someone we've never met before. You type something, and seconds later a reply appears on your computer screen. The conversation can be inane or seriously intellectual, but we take on faith that somewhere in the world, a human being, just like us, is typing replies on his or her own computer.
But what if the server handling the chatroom is not relaying our messages to another human being? What if the server has its own sophisticated program capable of mimicking human responses to input? Such programs have existed for quite a while, of course, but at some point it becomes apparent from the canned nature of the reply that we are not dealing with a human. But suppose the software becomes so sophisticated -- volunteering information, initiating new topics, expressing surprise and humor, responding completely naturally to our own remarks -- that it becomes impossible for any observer to determine objectively whether we are dealing with a person or a program.
Is the program, or the server on which the program is installed, "conscious"? If, hypothetically, there is no test that I can perform at my terminal that would differentiate the mimicking program from a human being, how do we know, objectively, that the program is not conscious? Or, conversely, how do we know that the people with whom we interact daily are not robots -- robots programmed to speak and behave as they do, robots which do not share "consciousness" with us?
And, for that matter, what do we mean when we claim that we ourselves are "conscious"?
All this is a well-traveled road in science fiction, of course. Recall HAL, the computer in the Kubrick movie 2001, Space Odyssey. HAL appeared completely, humanly conscious, even to the extent of disobeying instructions, killing its (his?) wards, and expressing remorse and fear of death as it (he?)was being shut down by the crew. These questions are also familiar territory in philosophy (see solipsism) and in psychology, as well as in the speculation of 12-year-old kids lying out on the lawn staring at the stars.
What brings it to mind is a review in this week's Economist of a book by a Brandeis professor named Irene Pepperberg. (Caveat: I haven't read the book.) According to the review, Dr. Pepperberg lived for 30 years or so with a roommate named Alex, an African Grey Parrot. She taught Alex to learn the English words for about 50 objects, to have a total vocabulary of about 100 words, to count from one to six, to "perform simple addition," and to make use of categories such as "same," "different," shape, color and material. Alex also could combine words in a rational manner to make new words for new objects.
Alex made headlines last year when he died. His last words to his owner were, "You be good. I love you."
Was Alex consciously speaking in the same sense as a young child consciously speaks? Or was he just "parroting" words? How do we know? If we assume that Alex could actually perform the tasks that Dr. Pepperberg claims for him, how do we know if he was merely a computer responding to stimuli as she had programmed him to do, or if -- on the other hand -- he was actually using the speech that Dr. Pepperberg had taught him to express "himself" in the same way that a child would? If he was simply a robot, a black box that spoke back automatically to input stimuli, then are our dogs and cats also robots when they express themselves non-verbally? When my cat approaches softly and puts his cheek against mine, is he simply reacting to expected warmth? Or is his behavior simply a learned response to his past experience that I tend to scratch his itchy head when he does so?
If we grant dogs and cats some level of conscious behavior, and parrots some ability to use human speech as an expression of consciousness, where does that leave the computer program? If a program can be designed to mimic a dog's behavior -- a far simpler matter than my original hypothetical of one imitating a human being -- does the computer have a dog's consciousness? Is the fact that a real dog's "program" is encoded in organic neural synapses rather than in semi-conductors a critical distinction? Why and how?
Hey, I don't have any solutions! And as far as I can tell, neither do scientists. The nature of consciousness, as viewed as a scientific question rather than as a matter best left to philosophy or theology, appears totally up in the air at present. Maybe, insofar as "consciousness" seems to be a purely subjective phenomenon, it is not even a subject with which science can deal.
But I expect we will hear much more about it from the scientific community in the coming years.
-------------------------Irene M. Pepperberg, Alex & Me: How a Scientist and a Parrot Uncovered a Hidden World of Animal Intelligence -- And Formed a Deep Bond in the Process (2008). The author published a less anecdotal study earlier, entitled The Alex Studies: Cognitive and Communicative Abilities of Grey Parrots (2000).
No comments:
Post a Comment