Narrative Frames: A New Approach to Analysing Metaphors in AI Ethics and Policy Discourse

Metaphors actively shape AI policy debates: The paper demonstrates that metaphors aren't just decorative language—they fundamentally influence how people reason about AI governance. By systematically analyzing 685 metaphors from policy discourse, the research shows that the specific frames used to describe AI (e.g., comparing it to tools, threats, or entities) can steer public opinion and policy priorities in particular directions.

We need a shared language to identify hidden assumptions: The paper introduces "Narrative Frames," a standardized system of 49 distinct categories for analyzing metaphors in AI discussions. This common vocabulary allows researchers, journalists, and policymakers to consistently identify and compare the underlying assumptions embedded in how different groups talk about AI—revealing whose perspectives are centered and whose are marginalized.

Absent metaphors matter as much as present ones: By mapping which narrative frames appear frequently in AI discourse and which are notably missing, this approach exposes blind spots in policy conversations. For example, if certain metaphors for AI accountability are rarely used, that absence itself signals what issues may be overlooked in governance discussions.

This tool can increase transparency in AI governance: The framework enables more explicit analysis of power dynamics in policy debates by making visible the often-invisible metaphorical choices that shape how stakeholders frame AI's risks, benefits, and appropriate regulation—supporting more informed and democratic decision-making.