Figurative discourse about AI: A Combined Input Hypothesis analysis

Abstract

Public discourse about Artificial Intelligence (AI) relies on figurative language as a semiotic resource for conceptualising the technology. While previous research has examined metaphor and metonymy in such discourse, less attention has been paid to how these processes interact in shaping figurative construals of AI. Drawing on the Combined Input Hypothesis, this study analyses naturally occurring examples from published blogs and articles as instantiations of two broad types of figurative framings: those positioning the technology in relation to human activity and interaction, and those emerging from debates about its capabilities, risks, and limitations. The findings show that metonymy and metaphor perform distinct but complementary functions. Metonymy foregrounds contextually relevant aspects of AI, while metaphor organises these selected aspects into coherent interpretive configurations, shaping how the technology is construed in terms of its capabilities, relation to human agency, epistemic accessibility, and risks. This interaction gives rise to evaluations ranging from trust and collaboration to scepticism, awe, and fear. The study demonstrates the explanatory value of the Combined Input Hypothesis for understanding how metonymic profiling and metaphorical projection jointly structure figurative discourse about AI.