Artificial Intelligence

Roles: model, training-data, parameters, inference, prompt, output, alignment, autonomy, capability

Systems that perform tasks typically requiring human cognition — pattern recognition, language generation, decision-making, planning. As a target domain, artificial intelligence is structured almost entirely by metaphor: we call its outputs “hallucinations,” its preparation “training,” its boundaries “alignment,” its text “reasoning.” The frame is a battleground of competing source domains (tool, agent, oracle, partner, threat), each importing different assumptions about autonomy, reliability, and moral status. What makes AI unusually productive as a target is that no literal vocabulary exists for what these systems actually do, so every description is already a metaphorical choice.

Applied To This Frame (43)