I’m amazed at the progress of technology over the past 100 years. Modes of transportation have gone from wheeled carts pulled by horses to vehicles that are measured in thousands of horsepower. There have been major advances in medicine—doctors can transplant most organs of the human body, and there has even been partial success in transplanting the head of a monkey to a different body. Or was it the body to a different head? We’ve been to the moon and Mars. We’ve split the atom. We can talk instantly with someone on the opposite side of the planet, and computers that would have filled entire rooms years ago, can now fit on your wrist or in your pocket.
So… I pose this question. When do you think we will be able to talk to our computers?
We do this somewhat now, in limited capacities, so let me rephrase the question. When do you think conversations with our computers will be as real and indistinguishable as conversations with other humans? I’m not talking about sentience, that’s an entirely different conversation, but when will computers be able to simulate human conversation?
ELIZA was one of the first programs to attempt a conversation between man and machine, and was created by Joseph Weizenbaum in 1966. Dr. Weizenbaum was a computer science professor at MIT, and created ELIZA to simulate an empathic psychologist. It would take statements made by patients and rephrase them as questions. For example, a response to “My back hurts” might be “Why do you say your back hurts?” It’s a simple little trick that seems to mimic intelligence, but is easily dismissed as it becomes apparent that there is no reasoning behind the responses. There are many other conversational programs available today—Jabberwacky, ALICE, PARRY, ELLA, and HAL to name a few, but they all fall short when it comes to true natural language processing (NLP).
This idea is not new to the field of artificial intelligence. Alan Turing first proposed a test of intelligence in the 1950 edition of Computing Machinery and Intelligence. The test goes like this: a human judge has a text-only conversation with a computer program and another human. If he is unable to distinguish between the computer and the other human, then that program passes the test—the Turning Test.
In 1990 Hugh Loebner brought this test to life by offering $100,000 and a gold medal to the first computer program whose responses were indistinguishable from a human’s. This grand prize is still unclaimed, and programmers still compete annually for a bronze medal and a $2000 prize.
You might think that computers are smart, but the ability to do something quickly and efficiently doesn’t indicate intelligence. For the most part, computers just do what you tell them. They don’t think for themselves. They just follow instructions. There has been significant progress in the field of artificial intelligence (AI), but the smartest machines of today still only have the intelligence of an insect.
In 2005 Ray Kurzweil wrote a book called The Singularity is Near that suggests sentience will happen not too far in the future. Ray has been hailed by many as a modern day scientific prophet. Bill Gates said, “Ray Kurzweil is the best person I know at predicting the future of artificial intelligence.” His book suggests that technology is growing exponentially—by 2020 machine intelligence will equal human intelligence, and by 2040 machines will surpass the intelligence of all humanity combined.
Is sentient life possible? Ehh… I don’t know, we’ll have to wait and see, but I do think it will be possible for a computer to emulate a human—first in speech, and then in action.