System Instructions and the ACRL Framework
These are my notes on how creating system instructions, in this case, the Metaphor, Anthropomorphism and Explanation Audit prompt aligns with the Framework for Information Literacy for Higher Education.
The system instructions treat AI discourse as an information environment worth reading closely. The analysis aims to trace how authority is manufactured, how explanations smuggle in agency, how value and liability get distributed, and how language steers searching, interpretation, and decision-making across contexts.
When I think about what students actually do when they build these prompts (researching theoretical frameworks, translating concepts into operational logic, testing outputs against expectations, revising based on diagnostic feedback) it maps surprisingly well onto the ACRL Framework's vision of information literacy as "reflective discovery" and "participation in communities of learning."
Quick Reference​
| ACRL Frame | How the Prompt Operationalizes It |
|---|---|
| Authority Is Constructed and Contextual | Treats authority as rhetorical effect; requires naming human actors; makes expertise visible as social relationship |
| Information Creation as a Process | Shifts attention from "what text says" to "how it was made to say it"; treats writing choices as production techniques |
| Information Has Value | Makes value recognizable as economic/political/ethical; connects language to stakes (who profits, who carries liability) |
| Research as Inquiry | Treats reading as iterative investigation; keeps uncertainty productive; frames "why" questions as contested |
| Scholarship as Conversation | Positions AI discourse as a live debate; student enters conversation through citation, comparison, reframing |
| Searching as Strategic Exploration | Research phase requires searching for theories, actors, contexts; makes information ecosystem itself searchable |
Authority Is Constructed and Contextual​
The prompt treats "authoritativeness" as an effect of rhetoric, genre, institutional position, and technical register. A confident voice, a scientific tone, or a brand name, all can can function as an authority signal even when underlying mechanisms remain opaque.
This matters for AI discourse specifically: claims about what systems "know" or "understand" often derive authority from how they are framed rather than from transparent evidence. The prompt makes authority contestable by asking for evidence inside the text itself and for context outside the text, for example, who is speaking, for whom, with what incentives, under what constraints.
The repeated insistence on naming human actors aligns with authority as something conferred and maintained within communities. When text says "the model learned," the prompt asks: Who trained it? On what data? With what objectives? This keeps "expertise" visible as a constructed social relationship.
Related knowledge practices and dispositions:
- Evaluating credibility with skepticism and self-awareness
- Recognizing that authority varies by community and need
- Questioning traditional notions of granting authority and recognizing the value of diverse perspectives
Information Creation as a Process​
The prompt shifts attention from "the text says" to "how the text was made to say it." The analysis is process-forward: it treats writing choices, metaphors, passive voice, and explanation style as production techniques with real consequences.
This is especially relevant for AI discourse, which often comes from contexts with specific constraints: proprietary systems, selective disclosure, marketing pressures, and audience calibration. The prompt encourages treating these as part of what is being evaluated.
Students also learn that "format" is not destiny. A white paper, blog post, vendor page, academic article, or policy memo can each carry authority cues while being produced under very different conditions of incentive and power. The prompt asks: What kind of document is this? Who created it? What were they trying to accomplish?
Related knowledge practices and dispositions:
- Looking for signals of underlying creation processes
- Accepting ambiguity about emergent information products and new genres
- Matching information needs to products by understanding how they were produced and disseminated
Information Has Value​
The prompt centers value as economic, political, and ethical. The "who profits" line of questioning turns discourse into a site where commodification, liability management, and reputation intersect and are negotiated.
This connects language to stakes: how anthropomorphic framing can inflate perceived capability, justify investment, deflect accountability, or reassign risk onto users. When text says "AI decided" rather than "engineers at Company X designed a system that," the redistribution of liability is itself an operation on value.
The prompt also treats attribution and responsibility as part of that value. "Naming the actor" is a literacy move that clarifies who holds power, who carries consequence, and what alternatives were available. This extends the ACRL frame's traditional focus on citation practices into accountability practices.
Related knowledge practices and dispositions:
- Recognizing how socioeconomic interests shape information production and dissemination
- Noticing marginalization and whose voices are minimized by dominant framings
- Making deliberate choices about the downstream uses of information
Research as Inquiry​
The prompt treats reading as iterative investigation. The analysis is built around repeated cycles: identify patterns, test interpretations against quoted evidence, refine categories, and surface new questions.
This is where the prompt journal becomes pedagogically important. Students are producing a final prompt but along the way they document their iterations, their failed attempts, their diagnostic reasoning. The process makes visible what the ACRL Framework calls "formulating questions for research based on information gaps or on reexamination of existing, possibly conflicting, information."
The prompt also frames "why" questions as contested. The audit of explanation types encourages learners to ask what kind of "why" is being offered, and what is being bypassed when purposive, intention-based language substitutes for mechanistic description. What counts as an explanation? The question itself is part of the inquiry.
Uncertainty stays productive throughout. Claims can be rhetorically persuasive yet mechanistically underspecified, technically precise yet agency-slippery, ethically concerned yet accountability-diffusing. The prompt makes room for this friction rather than resolving it prematurely.
Related knowledge practices and dispositions:
- Formulating questions from gaps and contradictions
- Breaking complex problems into tractable sub-questions
- Persisting through ambiguity and valuing intellectual humility
Scholarship as Conversation​
The prompt positions AI discourse as a live debate with competing vocabularies, incentives, and explanatory norms. It invites readers to locate a text within a larger argumentative ecology rather than treating it as a self-contained authority.
Students enter the conversation in multiple ways. The prompt requires careful citation of passages (quotation as evidence), explicit comparison of interpretive alternatives, and reframing as a form of contribution. When a student rewrites "the model understands context" as "the system processes token sequences based on attention weights trained on web-scale corpora," they are making a scholarly intervention by proposing an alternative vocabulary with different implications.
The prompt also highlights barriers to participation created by technical discourse, institutional prestige, and proprietary opacity. These barriers become analyzable features rather than operating as some neutral background. Who gets to speak authoritatively about AI? What credentials are required? What happens to voices that use different vocabularies?
The prompt journal and critical reflection essay are themselves forms of scholarly contribution appropriate to the student's level, as the ACRL Framework suggests. Students aren't simply consuming scholarship that is “out there” in the world, they're producing analytical instruments that could be shared, critiqued, and refined by others.
Related knowledge practices and dispositions:
- Seeing scholarship as ongoing negotiation of meaning
- Critically evaluating contributions in participatory information environments
- Recognizing how systems privilege certain voices and what that means for their own participation
Searching as Strategic Exploration​
This frame traditionally addresses source discovery processes for things like scholarly databases, search strategies, and controlled vocabulary. But when operationalized as a learning experience, building a prompt that operationalizes theory requires genuine research. Students have to find and connect theoretical frameworks, trace their intellectual histories, and identify which concepts can be made operational.
The prompt design process converts "search" from a mechanical step into a strategic practice. Students search for theories (Lakoff's conceptual metaphor theory, Brown's typology of explanation), then search within those theories for operational primitives. They search outward for actors, contexts, and incentives that might explain a text's framing choices. They search for alternative vocabularies that could reframe anthropomorphic language.
The information ecosystem itself becomes searchable. Instead of "find sources about AI," the work becomes "find how the discourse makes AI seem like a source, an agent, a knower, a decision-maker." This requires flexible movement between divergent and convergent thinking: brainstorming candidate framings, then selecting and defending interpretations with textual support.
Related knowledge practices and dispositions:
- Designing and refining strategies based on what emerges
- Using multiple kinds of searching language and multiple pathways to access context
- Persisting through search challenges and seeking guidance from experts when needed
Closing Thought​
I didn't design the prompt with the ACRL Framework in mind. But when I step back and look at what students would actually do when they engage with this material (researching theory, evaluating authority claims, tracing value flows, entering scholarly conversations, searching strategically for contexts and alternatives), well, it fits.
Maybe that's not surprising. The ACRL Framework describes what it looks like to engage critically with information. The prompt is a tool for engaging critically with a specific kind of information: discourse about AI that shapes how we understand, trust, and regulate these systems. It is a recognition that critical information literacy and critical discourse analysis share some deep roots.
Discourse Depot © 2025 by TD is licensed under CC BY-NC-SA 4.0