Complexity Explorer Santa Few Institute

Dr. Melanie Mitchell introduces the expectations and limitations of Artificial Intelligence (AI) in two lectures from 2019-2020. AI is the process of engineering intelligent machines using knowledge from science and philosophy about information processing in various contexts - though as you will hear, there is not complete agreement on a universal definition of AI.

AI is generally used for very concrete applications, such as self-driving cars as Dr. Mitchell explores in her lectures. AI is also related to understanding how humans think: some approaches to AI are inspired by the brain's structure and cognitive processes, but AI need not be limited to biological templates or human measures of intelligence.

Within the domain of AI, "intelligence" refers to the computational process by which a machine accomplishes a goal (assigned by a human) without being an algorithm or equation for exactly how to do so. AIs must be trained by a human user, but the "intelligence" they develop is not specified: the AI learns its own way of accomplishing the assigned goal. Dr. Mitchell discusses the challenges of defining "intelligence" for a machine and why sometimes the AIs learn things the human user did not intend.