Most of the AI literacy frameworks I’ve covered focus on what students should know. Dilinika’s (2026) article in College & Research Libraries News takes a different angle. It focuses on what teachers should do. The four practical approaches she proposes are concrete, classroom-ready, and grounded in two theoretical traditions that don’t usually meet: constructivist learning theory and data justice. The combination works.
Dilinika positions critical AI literacy as an extension of information literacy, not a separate domain. The argument is that traditional digital literacy doesn’t address what AI tools actually require. Students need to question outputs, recognize hallucinations, understand inherited biases, and engage with the broader socio-ethical implications of generative AI. Technical proficiency on its own isn’t enough.
The data justice framing is the more striking part. The framing draws on Linnet Taylor’s (2017) definition. Dilinika (2026) argues that “when applied to AI literacy, data justice principles help students recognize that AI systems are not neutral; they reflect the values, biases, and limitations embedded in their training data and design processes” (p. 2). AI literacy that doesn’t engage with the politics of training data isn’t critical AI literacy.
Why Constructivism + Data Justice
The pairing matters. Constructivism gives the active-learning method: productive struggle, guided discovery, knowledge built through participation, not passive consumption. Data justice gives the ethical lens: whose data, whose values, whose harms. Together they shape instruction that builds real critical engagement with AI.
This pairing distinguishes Dilinika’s framework from the more technical or competency-focused frameworks in the field. Hervieux and Wheatley’s (2024) six-frame AI literacy framework, which I’ve covered before, builds a hierarchical structure based on Bloom’s taxonomy. Dilinika’s framework instead asks teachers to consider both how students learn and what they’re learning to question. Both approaches have merit. The data justice grounding gives Dilinika’s framework an ethical anchor the others lack.
Critical AI Literacy: The Four Approaches
The four practical approaches are the heart of the article.
The first is design-based learning. Students engage in open-ended projects where they identify problems, empathize with users, and prototype solutions. Dilinika names three design traditions: Value-Sensitive Design, Speculative Design, and User-Centered Design. Speculative design gets particular attention, with students imagining future scenarios (an AI determining social benefits via predictive algorithm) and engaging through digital storytelling to question fairness.
The second is metacognitive activities and ethical reflection. Dilinika (2026) emphasizes that “critical AI literacy instruction needs to be flexible, encouraging learners to be innovative, creative, and open to making mistakes and learning from them” (p. 3).
The concrete classroom moves are the strongest part: guided reflection on AI-generated texts, metacognitive journals, think-aloud exercises, source evaluation checklists, peer discussions. Her metacognitive prompts are reusable: “How might this content influence decisions or perceptions if shared without verification?” and “What information might be missing or overlooked in this AI-generated text?”
The third is maker activities. Students learn AI literacy by creating artifacts that express critical standpoints. Her example draws on Lee, Gobir, Gurn, and Soep’s work where students built multimedia prototypes investigating recommender systems and facial recognition surveillance. The maker movement’s roots in constructivism make this fit naturally.
The fourth is dialogical activities, grounded in Vygotsky and social constructivist theory. Specific formats include role play, peer discussions, data storytelling, data mapping, and online or in-person forums.
What This Means for Teachers
The four approaches transfer cleanly to non-library contexts. Teachers in K-12 or higher ed can use any of the four in their own classes without the librarian framing.
The metacognitive approach is the lowest-friction starting point. Three or four reflection prompts on any AI-using assignment can change how students engage with outputs. Roe, Furze, and Perkins’s (2025) digital plastic metaphor for AI writing, which I’ve covered before, makes a related point: students need scaffolds to evaluate what AI produces, not just permission to use it.
The design-based approach is the most ambitious. A unit built around speculative design (“imagine a future where AI does X”) asks students to think through implications they wouldn’t otherwise consider. The investment pays off in critical thinking gains that go past AI itself.
The maker approach is the most underused in school contexts. Students rarely build artifacts about AI; they consume AI-generated artifacts. The reversal is high-leverage.
Limitations
The article has limits worth naming. It’s brief (about four pages), practitioner-focused, and conceptual not empirical. Dilinika doesn’t test any of the four approaches against student outcomes. The references support the conceptual moves but the argument runs faster than the evidence base can carry.
The data justice grounding is the most original contribution but also the least developed. Dilinika cites Taylor’s (2017) definition and a few studies applying data justice to literacy education, but the article doesn’t show what data justice looks like in a specific lesson plan. A follow-up piece with worked examples would strengthen the framework substantially.
The author concludes that “as society is increasingly shaped by emerging technologies, critical AI literacy has become a civic imperative. Students need support to move beyond technical skills and critically engage with AI by analyzing outputs, recognizing biases, and reflecting on ethical and social implications” (p. 3). I agree with that argument. The four approaches give teachers concrete starting points. The civic framing names what’s at stake.
Pick one approach. Build it into the next AI-using assignment. The metacognitive prompts are the lowest-friction place to start. Critical AI literacy is taught one assignment at a time.
References
- Dilinika, J. M. S. (2026). Teaching students to think critically about AI: Practical approaches for academic librarians in designing literacy instruction. College & Research Libraries News, 87(5).
- Hervieux, S., & Wheatley, A. (2024). Building an AI literacy framework: Perspectives from instruction librarians and current information literacy tools [Choice White Paper]. Choice / Association of College and Research Libraries.
- Linnet Taylor, “What Is Data Justice? The Case for Connecting Digital Rights and Freedoms Globally,” Big Data & Society 4, no. 2 (2017), doi:10.1177/2053951717736335.
- Roe, J., Furze, L., & Perkins, M. (2025). Digital plastic: A metaphorical framework for Critical AI Literacy in the multiliteracies era. Pedagogies: An International Journal. Advance online publication. https://doi.org/10.1080/1554480X.2025.2557491