Melanie Mitchell's Abstract

Title: The Barrier of Meaning

Abstract: In 1986, the mathematician and philosopher Gian-Carlo Rota wrote, "I wonder

whether or when artificial intelligence will ever crash the barrier of meaning." Here, the

phrase "barrier of meaning" refers to a belief about humans versus machines: humans

are able to "actually understand" the situations they encounter, whereas AI systems (at

least current ones) do not possess such understanding. The internal representations

learned by (or programmed into) AI systems do not capture the rich "meanings" that

humans bring to bear in perception, language, and reasoning. In this talk I will assess

the state of the art of artificial intelligence in several domains, and describe some of the current

limitations and vulnerabilities, which can be accounted for by a lack of true

understanding of the domains these systems work in. I will explore the following

questions: (1) To be reliable in human domains, what do AI systems actually need to

"understand"? (2) Which domains require human-like understanding? And (3) What

does such understanding entail?