Today, we find many different opinions regarding what constitutes human intelligence. There is no one widely accepted answer. Here are two definitions that have found some acceptance among the scientific community.
- “A very general mental capability that, among other things, involves the ability to reason, plan, solve problems, think abstractly, comprehend complex ideas, learn quickly, and learn from experience. It is not merely book learning, a narrow academic skill, or test-taking smarts. Rather, it reflects a broader and deeper capability for comprehending our surroundings—‘catching on,’ ‘making sense of things,” or ‘figuring out’ what to do” (“Mainstream Science on Intelligence,” an editorial statement by fifty-two researchers, The Wall Street Journal, December 13, 1994).
- “Individuals differ from one another in their ability to understand complex ideas, to adapt effectively to the environment, to learn from experience, to engage in various forms of reasoning, to overcome obstacles by thinking. Although these individual differences can be substantial, they are never entirely consistent: a given person’s intellectual performance will vary on different occasions, in different domains, as judged by different criteria. Concepts of ‘intelligence’ attempt to clarify and organize this complex set of phenomena. Although considerable clarity has been achieved in some areas, no such conceptualization has yet answered all the important questions, and none commands universal assent. Indeed, when two dozen prominent theorists were recently asked to define intelligence, they gave two dozen, somewhat different, definitions.” (“Intelligence: Knowns and Unknowns,” a report published by the Board of Scientific Affairs of the American Psychological Association, 1995).
Now that we have some basis for defining human intelligence, let us attempt to define a test that we could use to assert that artificial intelligence emulates human intelligence.
Alan Turing is widely considered the father of theoretical computer science and artificial intelligence. He became prominent for his pivotal role in developing a computer that cracked the daily settings for the Enigma machine, Germany’s technology for coding messages during World War II. This breakthrough allowed the Allies to defeat the Nazis in many crucial engagements. Some credit Turing’s work, as a cryptanalyst, for shortening the war in Europe by as many as two to four years. After World War II, in 1950, Alan Turing turned his attention to artificial intelligence and proposed the now-famous Turing test. The Turing test is a methodology to test the intelligence of a computer. The Turing test requires a human “judge” to engage both a human and a computer with strong AI in a natural-language conversation. None of the participants, however, can see each other. If the judge cannot distinguish between the human and strong AI computer, the computer passes the Turing test and is equivalent to human intelligence. This test does not require that the answers be correct, just indistinguishable. Passing the Turing test requires almost all the major capabilities associated with strong AI to be equivalent to those of a human brain. It is a challenging test, and to date, no intelligent agent has passed it. However, over the years, there have been numerous attempts to pass the Turing Test, with associated claims of success. Here is a summary of major attempts to pass the Turing Test:
- In 1966, Joseph Weizenbaum created the ELIZA program, which examined a user’s typed comments for keywords. If the program found a keyword, its algorithm used a rule to return a reply. Although Weizenbaum and others claim success, their claim is highly contentious. In effect, this is the same type of algorithm (i.e., a set of rules followed in problem-solving operations by a computer) early search engines used to provide search returns before Google’s use of “link popularity” (i.e., the number of links that point to a website using an imbedded keyword) to improve search return relevance.
- In 1972, Kenneth Colby created PARRY, which was characterized as “ELIZA with attitude.” The PARRY program took the ELIZA algorithm and additionally modeled the behavior of a paranoid schizophrenic. Once again, the results were disappointing. It was not able to consistently convince professional psychiatrists that it was a real patient.
- In 2015, the developers of a program called Eugene made a claim it passed the Turing Test. However, their claim turned out to be bogus. Eugene was able to convince 10 of 30 judges from the Royal Society that it was human. Although augmentative, there is a strong consensus based on the test conditions and results that Eugene did not pass the Turing test.
Although other tests claim to go beyond the Turing Test, no new test has gained wide support in the scientific community. Therefore, even today, the Turing Test remains the gold standard concerning an AI machine emulating human intelligence. Despite recent claims to the contrary, no AI machine has been able to pass the Turing Test.
