Main | Introduction | The Turing Test | The Loebner Prize | Strong AI | Sources | About
Inseparability of Mind and Body | The Frame Problem | A Holistic Approach to AI
A Holistic Approach to AI
Strong AI is a term used to describe a certain mindset of artificial intelligence development. Strong AI's goal is to develop artificial intelligence to the point where the machine's intellectual capability is functionally equal to a human's. There are several fundamental differences between the chatbots of the Loebner Prize and a true example of Strong AI. Chatbots are built for the sole purpose of conversation, using increasingly complex algorithms and response databases to determine an appropriate reply to a query or statement. They are an example of Applied AI, where systems are built specifically for one usage. The ideal Strong AI machine, however, would be built in the form of a man, have the same sensory perception as a human, and go through the same education and learning processes as a human child. (Copeland) Essentially, the machine would be "born" as a child and eventually develop to an adult in a way analogous to human development.
This approach presents a solution to the problems of symbolic attempts to create human intelligence in computers. Instead of trying to give the computer adult-like knowledge from the outset, the computer would only have to be given the ability to interact with the environment and the ability to learn from those interactions. As time passed it would gain common sense and language on its own. This paradigm seeks to combine the mind and the body, whereas the common trend in symbolic programming (i.e. CYC) has been to disregard the body to the detriment of the computer's intellect.
At present Strong AI development is overshadowed by the immediate applications of Applied AI, as that field yields more tangible and quantifiable results. In addition, development in embodied systems has not progressed beyond the most basic constructs without the overall intelligence of even a cockroach. However, it can be argued that the single largest stumbling block to Strong AI is the lack of definition of intelligence. (Copeland) Strong AI's ultimate goal is to make an intelligent computer that can think and understand, but those terms remain ambiguous and undefinable; hence, there is no general measure of "success" in the field of Strong AI. The concept of intelligence varies from person to person, and as a result, a standardized test like the Turing Test becomes less important.
The Turing Test has been interpreted as a way to define intelligence, that is, if one cannot tell a computer from a human being then it is intelligent. However, Turing himself countered that an intelligent entity can fail the test because it cannot successfuly imitate a human being. (Copeland) His response sums up the reason for the Turing Test's increasing irrelevance in AI development. Turing offered his test to question the preconcieved notion that computers cannot be intelligent by providing a scenario in which computers are indistinguishable from humans. However, by urging that a computer's mind be judged without regard to the body, he inadvertently appears to ally himself with an Extropian mindset that has carried over into artificial intelligence development. What can be interpreted as a proposed direction for development was simply a setting for testing; ironically, it appears the best way to create a Turing-capable machine is to develop the mind and body together. Even so, the goal of passing the test becomes more and more unimportant as the perception of intelligence becomes more complex; in the case of the Loebner Prize, the goal of passing the test even seems to inspire pure trickery rather than encourage earnest progress toward intelligent systems. By ignoring the Turing Test, researchers are free to work based on their own judgments about intelligence and how to achieve human-like AI.
Machine intelligence, even at the lowest levels, remains elusive, and the definition of intelligence is as ambiguous as ever. However, the holistic paradigm of Strong AI represents a break from conventional ideas of disembodied AI systems, and may hold the key to finally realizing a kind of intelligent AI. Just as Turing questioned the notion that computers cannot think, Strong AI questions that idea that the mind should be isolated from the body. The Turing Test has accomplished its goal by bringing into general acceptance the idea of machine intelligence; as an actual test, it is too simplistic to be sufficient as an indicator of intelligence. Flawed from the outset and distorted by time, the Turing Test has become constraining and ultimately irrelevant to AI development, but it remains an important and historical thought experiment bold enough to imagine the achievement of humanoid intelligence in the early years of artificial intelligence.