Alternative Intelligence: Anthropomorphism, Agency and Affordance

The present paper argues that anthropomorphism, used to describe and design Artificial Intelligence, AI, distorts the relationship between computational and human agency. Applying Giddens’ structuration theory to a network of peer agents, the present paper proposes that computational agents inhabit a society distinct from, though interacting with, human society. The approach is developed through affordance, functional fixedness, and user responsibility. AI systems designed along anthropomorphic lines, to replicate human intelligence, are redundant: the apparent replication is illusory, and the illusion is what misleads the user as to what the system actually affords. The value of computational agency lies instead in different perspectives. It further argues that anthropomorphism obscures the affordances offered by AI systems and how those systems are used, and proposes that alternative intelligence is a better foundation for AI design and governance.