That human cognition is inherently metaphorical is as much a fact of ordinary thinking as it is of science and engineering, where metaphors loom large in the conception, articulation, and communication of ideas, models, and theories. Artificial Intelligence, as a set of scientific and engineering practices, also thrives on metaphors, putting them to good use in a generative manner. However, there are interesting differences between this use and how metaphors have been traditionally used in other areas of science and engineering. In this talk, I explore the differences on the three dimensions of modeling, implementation, and evaluation. These have to do with the relationship, respectively, between models and mathematics, abstraction and materiality, and behavior and mechanism. I discuss the broader implications of this exploration for our understanding of the economy of metaphors as social practice.
Conference: “Beyond AI: Artificial Golem Intelligence” (BAI2013)
Speaker: Hamid Ekbia
Title: AI, Its Metaphors, and Their Mutation