I was glad to come across a recent paper of Kalantzis and Cope, two scholars whose work has shaped tremendously shaped my understanding of multiliteracies. In this paper title “Literacy in the Time of Artificial Intelligence”, Kalantzis and Cope make this thoughtful argument : generative AI is fundamentally a writing technology whose implications run deep and therefore deserve more attention than they’re getting.
They further added previous technologies mechanized the reproduction of text. The printing press, the photocopier, the word processor. Generative AI does something different. It mechanizes the production of text. And that changes what literacy means at a foundational level.
Kalantzis and Cope frame this as a historical turning point:
For the first time in human history, a machine can design and communicate written-textual meanings (and, derivatively, image, speech, and other forms of meaning) that have never been created before but are nevertheless coherent and meaningful to humans. (p. 11).
I read that sentence several times. We’re not talking about autocomplete or spell-check. We’re talking about a machine that produces ‘original’, coherent writing. If writing has always been how humans think, argue, discover, and communicate ideas, then having a machine that can do it too forces us to rethink what we’re teaching when we teach literacy.
How AI Is Changing Literacy: Grammar vs. Statistics
Kalantzis and Cope draw a sharp line between how humans and machines produce language. Humans organize experience grammatically, through subjects, actions, relations, and purposes. We classify the world to make meaning. LLMs do something fundamentally different. They calculate statistical proximity between tokens. They predict what word comes next based on patterns across massive datasets.
As the authors put it: “The human mind works grammatically. Generative AI works statistically” (p. 15). They go further: “The LLM can never know the meaning of ‘walk.’ It treats language as a meaningless ‘bag of words'” (p. 13).
And yet the outputs look meaningful. Coherent, well-structured, often persuasive. Statistical patterns across billions of documents approximate what meaningful language looks like, even though the machine has zero understanding of what it’s saying.
For teaching, the implication is huge. If students believe AI “understands” their prompts and “thinks” through its responses, they’ll accept its outputs uncritically. Shaw and Nave (2026) documented exactly that in their cognitive surrender research: students defer to AI-generated text without evaluating it. Kalantzis and Cope help explain why. The outputs look and feel like the product of understanding, even though they’re the product of calculation. Helping students see that gap is one of the most important things we can do.
Literacy as Design, Not Rule-Following
Kalantzis and Cope have been developing their multiliteracies framework for decades (e.g., Kalantzis & Cope, 2004; 2005; 2006; 2008), and they bring it here with renewed urgency. Literacy, in their view, has never been about memorizing grammar rules or passing comprehension tests. Literacy is design. The active work of making meaning, drawing on available resources, transforming them for a purpose, and leaving a trace in the world.
“Meaning is transpositional,” they write. “Meanings must be remade as meanings-for-oneself to become meanings for others, and the meanings of others are remade as they become meanings for-oneself” (p. 10).
So here’s the productive question. If we define literacy as rule mastery, AI outperforms humans in many routine tasks. Grammar, spelling, structure, formatting, all handled faster and more consistently by a machine. But if we define literacy as design agency, as the ability to shape, question, and transform meaning with intention, then the human role doesn’t shrink. It becomes more important.
We need to move literacy pedagogy away from the mechanical and toward the creative, the critical, the intentional. Students should be using AI as a writing partner: generating drafts, critiquing them, reshaping them with their own voice and purpose. The value isn’t in the first draft. It’s in what the student does with it.
Cheng et al. (2025) showed something similar in their study on AI and writing performance. Students who asked AI direct, pragmatic questions performed better on writing tasks. The student’s agency in shaping the exchange predicted the outcome. AI didn’t replace the thinking. The thinking determined how well AI was used.
Cheating Is a Symptom, Not the Problem
Kalantzis and Cope address academic integrity directly. They cite evidence that “94% of AI submissions were undetected by the human exam readers and AI submissions outperformed real students” (p. 7 citing Scarfe et al., 2024). But they treat cheating as a surface symptom. The deeper issue is what writing represents in education and how we assess it.
If uniqueness once served as proof of authorship, they argue, that metric collapses when machines generate unique text at scale. I’ve argued the same point in my posts on AI and assessment: detection is a dead end, and the energy spent on policing would be better invested in redesigning what we ask students to do.
Kalantzis and Cope propose “cyber-social literacy learning,” where AI becomes a scaffolded writing partner and student agency is made visible and assessable. That direction aligns with everything I’ve been advocating: build pedagogy around AI, make the process transparent, and assess the thinking and the process, not just the product.
The Political Economy of AI and Writing
One of the most provocative sections in the paper concerns ownership. Kalantzis and Cope argue that LLMs privatize collective human expression. They invoke Marx’s concept of “general intellect” and describe generative AI as “an extractive industry and rent-seeker as it sells the human collective intellect back to us” (p. 8).
Strong language. And I think it’s warranted. I wrote about a related dimension in my post on AI and intellectual property, where Bharati (2025) analyzed how different countries handle AI-generated works and who owns what AI creates. Kalantzis and Cope add the literacy angle: the raw material being extracted is human writing. Billions of texts produced over centuries, now compressed into statistical models and sold back through subscription services.
AI literacy can’t stop at prompts and productivity. It has to include the political and economic structures behind these systems. Roe, Furze, and Perkins (2025) made the same argument in their Critical AI Literacy paper, defining CAIL as requiring understanding of “embedded power structures” and “broader social, environmental, and economic impacts.” Kalantzis and Cope reinforce that call from the literacy studies side.
What Literacy Needs to Become
The authors close with a provocation I find compelling: “The turn to an AI whose technological basis is written text means that literacy has a big, new job to do” (p. 8).
I agree. And that new job includes helping students understand how AI produces text, developing their capacity to evaluate and reshape AI outputs, and building the kind of design agency that machines can’t replicate. Literacy in the age of AI isn’t about competing with the machine. It’s about knowing what the machine can and can’t do, and bringing the human capacities that it lacks: judgment, voice, purpose, critical awareness.
We have a chance to make literacy education more meaningful than it’s been in a long time. Let’s take it.
Reference
- Bharati, R. K. (2025). Intellectual property rights in AI-generated creative works: Human authorship in automated production. Indian Journal of Law and Human Behavior, 11(2), 77–92. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6120486
- Cheng, Y., Fan, Y., Li, X., Chen, G., Gašević, D., & Swiecki, Z. (2025). Asking generative artificial intelligence the right questions improves writing performance. Computers and Education: Artificial Intelligence, 8, 100374. https://doi.org/10.1016/j.caeai.2025.100374
- Kalantzis, M., & Cope, B. (2004). Designs for learning. E-Learning, 1, 38–92.
- Kalantzis, M., & Cope, B. (2005). Learning by design. Melbourne, Australia: Victorian Schools Innovation Commission.
- Kalantzis, M., & Cope, B. (2006). The learning by design guide. Melbourne, Australia: Common Ground.
- Kalantzis, M., & Cope, B. (2008). New learning: Elements of a science of education. Cambridge, Victoria, Australia: Cambridge University Press.
- Kalantzis, M., & Cope, B. (2025). Literacy in the time of artificial intelligence. Reading Research Quarterly, 60, e591. https://doi.org/10.1002/rrq.591
- 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
- Scarfe, P., Watcham, K., Clarke, A. D. F., & Roesch, E. B. (2024). A realworld test of artificial intelligence infiltration of a university examinations system: A ‘Turing test’ case study. PLoS One, 19(e0305354). https://doi.org/10.1371/journal.pone.0305354
- Shaw, S. D., & Nave, G. (2026). Thinking fast, slow, and artificial: How AI is reshaping human reasoning and the rise of cognitive surrender. Working paper, The Wharton School, University of Pennsylvania. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6097646