Scaffolding AI Literacy: Beyond the One-Shot Workshop

By /

I’ve been writing about AI literacy for years, and most of what passes for it in higher education is still a one-time session somewhere in the first-year curriculum. So when LaFlamme (2025) argues that academic librarians should design a sustained, scaffolded program, I’m with her in principle.

The paper proposes a four-tier model anchored in two existing frameworks: the ACRL Framework for Information Literacy (2016) and Hervieux and Wheatley’s AI Literacy Framework (2024). Both prioritize critical thinking and ethical engagement, which gives librarians a familiar foundation for new tools.

The argument is straightforward. Librarians already teach source evaluation, ethical research, and information critique. AI literacy needs all three. So why are we treating AI as a separate unit when it could grow out of work libraries are already doing?

scaffolding ai literacy

Scaffolding AI Literacy: A Four-Tier Model for AI Literacy Workshops

LaFlamme’s curriculum runs across four 60-minute workshops, each building on the last. Tier 1 covers foundations: definitions, basic tools like ChatGPT, and a shared vocabulary. Tier 2 shifts to use, with students developing research questions, generating keywords, and summarizing articles using AI tools. Tier 3 puts students in the evaluator’s seat through comparison exercises, misuse case studies, and a debate on AI ethics in academia. Tier 4 pushes toward advocacy, asking students to analyze controversial AI scenarios and propose strategies for responsible AI development.

LaFlamme writes that “the proposed model not only prioritizes technical competency but also integrates the essential components of information literacy, such as critical thinking, ethical evaluation, and responsible use of resources” (p. 4). I agree with the diagnosis. AI literacy deserves the same careful sequencing we apply to reading and research skills, and that’s the move the field needs.

My Take

The framing of librarians as natural AI literacy educators is overdue. Their existing skill set in source critique and ethical research gives them a head start most faculty don’t have. The argument tracks with what Kalantzis and Cope (2025) propose in their work on literacy in the time of AI, which calls for literacy practice that handles the new information environment.

The paper also names a real problem with the standard one-shot library session. Most library AI sessions I’ve seen squeeze foundational concepts, ethical questions, and tool demonstration into 50 minutes. Students leave overwhelmed and remember little. LaFlamme’s tiered structure gives the topic the breathing room it needs.

Discipline-by-discipline adaptation is another smart move on her part. AI literacy in a humanities classroom looks different from AI literacy in a STEM lab. The paper doesn’t get caught in a single template, which makes the model usable across institutions.

What’s Missing from This AI Literacy Model

The model has real limitations. It’s theoretical, and admittedly so. LaFlamme acknowledges that “formal assessment of its effectiveness is ongoing” (p. 3). I respect that, but a year after this paper landed, the field has moved on. Chee, Ahn, and Lee (2025) have proposed empirically grounded AI literacy competency frameworks, and UNESCO (2024) has issued an AI competency framework for students. Any AI literacy work in 2026 needs to engage with these, not stop at the older Hervieux and Wheatley model.

The bigger issue is scope. Four 60-minute workshops cannot carry AI literacy on their own. AI is now embedded in everything students do, from writing to research to coding. AI literacy belongs across the curriculum, taught and reinforced by faculty in every discipline.

The library should anchor the work, but it can’t be where the work ends. Roe, Furze, and Perkins (2025), in their digital plastic metaphor, argue that AI literacy is a way of seeing, not a workshop topic. I read LaFlamme’s model as a useful entry point that needs a second floor.

The faculty collaboration question is also undertreated. LaFlamme notes that some faculty “may be hesitant due to concerns about the time required for collaboration or skepticism about the relevance of AI in their subject areas” (p. 5). She’s right about that. The paper doesn’t say what happens when faculty refuse to engage at all, which describes several institutions I know. The model assumes a collaborative ecosystem that often doesn’t exist.

I’d also raise a critical thinking concern. The Tier 3 evaluation activity asks students to compare a well-written AI response against a flawed one. That’s a useful starting exercise. The harder case, where AI produces confident-sounding text with subtle problems, isn’t covered. Mishra, Warr, and Islam (2023) made this point years ago in their TPACK work: spotting good AI from bad AI requires deep subject knowledge, not generic critical literacy.

LaFlamme’s model is a strong addition to the AI literacy conversation, and I’d recommend it to any library team serious about this work. The four-tier structure is teachable, modular, and grounded in solid pedagogical traditions. It’s a starting framework, not a finishing one.

AI literacy needs to live everywhere students learn. It belongs in writing courses, research methods seminars, and faculty conversations about assessment. The library is one of the best places to anchor that work, but it can’t be the only place. If we treat AI literacy as a four-workshop checkbox, we’ll graduate students who recognize ChatGPT but can’t engage critically with the AI systems shaping their professions.

References

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top