How to analyze the frames and metaphors that hide power in AI
A guide to interrogating rhetorical power in sociotechnical systems
Overview
We make a problem through frames, narratives, and rhetorical power. The frames we select — ”fraud,” “empowerment,” etc. — make certain elements of a narrative salient, and hide or downplay others. Frames aren’t neutral — indeed, they’re often intentionally created and marketed. This is why they’re extremely powerful. Ya know how we say “climate change” rather than “global warming?” That was a strategic choice. In tacitly adopting a frame, we’re aligning ourselves with a set of interests, values, and politics, often without knowing it.
Frames and narratives are especially powerful tools amidst technological change and uncertainty -- when our expectations aren’t anchored to a predetermined future. It’s in these moments -- e.g. the rise of ‘AI -- wherein, according to Beckert and Bronk in Uncertain Futures, “some stabilisation of expectations is required for investment to be made in new products or processes.” “If innovative ideas are to corner the resources needed to make them a reality,” they continue, “they require promissory stories that help coordinate investment.” In short, our sociotechnical future is shaped by market, political, and rhetorical power to offer the stability -- and, in turn, close off plausible futures -- required for investment.
E.G.
Take the frame ‘hallucination’ in the context of generative AI. ‘Hallucination,’ anthropomorphizes the technology, and it implies an aberration, a mistake of some kind, as if it isn’t supposed to make things up. But that’s actually exactly what generative models do — given a bunch of words, the model probabilistically makes up the next word in that sequence. Presuming that AI models are making a mistake when they’re actually doing what they’re supposed to do has profound implications for how we think about accountability for harm in this context.
Or take ‘decentralization’ in crypto. In ‘Crypto is not decentralized,’ I argued that while crypto proponents contend that technical decentralization decentralizes power, it in fact hides power. I drew on the work of Angela Walch to illustrate that power concentrates in intermediaries, voting and governance, token ownership, coordinated decision-making, computing power, and in informal authority. Though technically decentralized, certain nodes can still hold more voting power.
Resources
Approaches
Interrogate the Narrative & Dominant Frame
What is the narrative you are telling yourself about the problem at hand?
What is the dominant frame of your problem? What metaphors are often used? List them.
Where do the narratives, frames, and metaphors come from? When were they popularized and by whom?
What does the narrative, frame, and attendant metaphors make more salient concerning your problem?
What do the narrative, frames, and metaphors downplay, neutralize, or hide altogether?
How might the narrative, frames, and metaphors hide power dynamics?
What is being presented as certain or stable that might actually be uncertain?
What values are implicit in the narrative, frames, and metaphors?
Whose interests might they promote?
Stakeholder & Power Mapping - Determine who is making & sustaining your problem.
Framers & Narrators: Who are the key people/organizations who proactively ‘make’ your problem by framing it, and telling its story?
Actors: Who are the key people/organizations who must act if the problem is going to be solved?
Stakeholders: Who are the key people/organizations who want to resolve the problem
Beneficiaries: Who are the key people/organizations who benefit from the status quo/the problem existing?
Zoom Out - If the frame or narrative is individualizing the problem, zoom out and consider the other parts of the system contributing to the problem.
What are the policies shaping the problem?
What are the institutional practices and incentives shaping the problem?
What are the social dynamics shaping the problem?
What are the cultural norms shaping the problem?
What common perceptions and patterns of behavior are shaping the problem?
What are the power dynamics shaping the problem?
Okay, that’s it for now.
Charley