Artificial Intelligence

John and Ken begin by trying to define intelligence in systems, and decide that anything with intelligence must be able to receive data from the external world, and then change the way it operates based on that experience. John thinks that anything capable of such acts has some sort of intelligence, but Ken wonders: why is it considered artificial? It may be synthesized rather than organic, but what makes it artificial rather than genuine? John notes that this definition of intelligence allows for many systems which display intelligence but do so in very odd and truly inhuman ways. Ken compares intelligence to flight, noting that many things fly: birds, insects, planes, helicopters, and blimps, but each in its own different and unique way. John and Ken go on to describe artificial intelligence as a tool for cognitive science and as a method of computer engineering, and how these two distinct approaches differ.

In order to delve deeper into artificial intelligence, Ken introduces Marvin Minsky, Professor Emeritus of Media Arts and Sciences at MIT and major contributor to the fields of artificial intelligence, robotics, and cognitive psychology. John begins by asking Marvin what the original vision for artificial intelligence was at the beginning of the field in the 1950’s and 1960’s and how it has changed with new technologies and the successes and failures of computer science. Marvin explains that artificial intelligence developed at the same time computers were first being developed, and that the field originally thought they could replicate most human actions on computers. Initially this seemed promising, since within the first few years they had created a computer that could do calculus as fast as an MIT student, but they soon found that computers could do difficult things like calculus easily, but had immense trouble doing easy things like having a conversation. What does this mean? How come calculus and theorem-proving are so easy for computers but so hard for us? Does this strange disconnect reveal something about the nature of the human mind? Marvin discusses his views and Ken puts forth his own theory about what these differences reveal.

Marvin Minsky argues that “intelligence” is a social relation which involves necessary interpersonal interaction, whereas what these computers and programs do is really about “resourcefulness,” or using excellent strategies to respond to data they receive. John and Ken wonder about the different strategies for chess-playing, a realm in which computers have slowly come to dominate human opponents. Marvin explains the differing strategies that human beings and computers use to play chess, and how computers use raw power to exhaustively search through moves whereas humans use common sense to eliminate many possibilities. Ken remarks on this theme contrasting raw power and common sense. What is common sense? Is it possible to emulate it somehow in computers? Should we bother? What is the point of creating computers that think like us when they are so successful at thinking in different ways? John, Ken, and Marvin discuss these issues and take calls from listeners interested in the details of artificial intelligence, the realities of many science fiction robots, and the future of human-robot interaction.

• Roving Philosophical Reporter (Seek to 4:38): Zoe Corneli travels to a Stanford Artificial Intelligence lab to learn about robotics and how researchers attempt to build robots that bridge the gap between industrial assembly lines and normal human environments.
   
• Sixty Second Philosopher (Seek to 49:56): Ian Shoales burns through the modern history of robotics and artificial intelligence, including some fantastical visions of human society deeply integrated with robotics and AI in the not-so-distant future.