How many students feel pride in the work they are made to do for school? My guess is very few. Pride if anything comes from the final grade not the work itself. The work too often is just meant to be seen by the teacher. I have been thinking about this idea of pride in the context of the advent of GenAI and the whole issue of plagiarism that has consumed so much of the conversation in education over the past few years. If students take pride in what they produce, the temptation to hand the whole thing over to a chatbot loses much of its pull. Not because detection has gotten better or because policies have gotten stricter, but because outsourcing work you care about feels wrong in a way that doesn’t require a rule to enforce.

That idea, and several others I’ve been developing on this blog, found their way into a recent scholarly dialogue with Aras Bozkurt, published in Open Praxis. The paper is called TPACK in the Age of Context and Artificial Intelligence: Punya Mishra on the “Wicked Problem” of AI and Why Algorithms Can’t Replace Context. It covers a lot of ground, but at its heart it’s a conversation about what teachers need to know in an era when GenAI can simulate much of what we used to think of as teaching.

It has been a while since I’ve written about TPACK on this blog, though the framework keeps evolving in ways that feel increasingly urgent. The paper addresses several threads I’ve been pulling on. One is the argument that content knowledge and expertise has become more important in the GenAI era, not less. This sounds counterintuitive, but the stochastic, hallucinatory nature of these technologies means they go off track in subtle ways that only someone with deep domain expertise can catch. I’ve called GenAI a “smart drunk intern” (I’ve written about this before), and the metaphor keeps proving its worth: incredibly capable, occasionally brilliant, and fundamentally unreliable.

Another thread is contextual knowledge, the XK addition to the TPACK framework. The original Venn diagram had a dotted line around the three circles labeled “contexts,” which always bothered me semantically. Contexts aren’t a backdrop; they’re a form of knowledge that teachers possess and deploy, knowing which students have tablets and which have limited computer lab time, what state standards apply, whether cell phones are banned, what the sociology of this particular classroom looks like. That knowledge is precisely what algorithms lack, and the paper argues it’s where human teachers hold their most significant advantage. (I’ve written about the XK upgrade before as well.)

The conversation also focused on what I’ve been calling “curriculum-shaped objects”: the lesson plans, syllabi, and assessment rubrics that GenAI produces with all the right headings and all the right vocabulary, and that look, on the surface, like perfectly competent curriculum. The problem is what’s missing. These outputs lack the practical wisdom that comes from having taught this unit to actual ninth-graders, from knowing where students get stuck, from understanding why a particular analogy works in one classroom and falls flat in another. When the form is right but the judgment is absent, the teacher’s role shifts toward curation and critical evaluation, asking not just “is this correct?” but “is this any good?”

And then there’s the section I suspect readers will find most surprising. Aras asked me about my vision for the future of teacher expertise, and I was candid: I’m pessimistic. Maybe that’s my own prejudice showing, a bias shaped by watching hype cycles in ed tech for three decades. But the patterns worry me. The teacher shortages, the equity implications of replacing human mentorship with automated tutoring for those who can’t afford better, the inevitability rhetoric that makes resistance feel futile… I’ll let you read that part for yourself.

Mishra, P., & Bozkurt, A. (2026). TPACK in the Age of Context and Artificial Intelligence: Punya Mishra on the “Wicked Problem” of AI and Why Algorithms Can’t Replace Context. Open Praxis. (18) 2. DOI: 10.55982/openpraxis.18.2.1093 [Link to PDF]

Abstract: This scholarly dialogue engages Professor Punya Mishra, co-developer of the Technological Pedagogical Content Knowledge (TPACK) framework, to examine the shifting landscape of teacher expertise in the era of Generative Artificial Intelligence (GenAI). As AI agents increasingly simulate pedagogical roles and generate educational materials, the conversation probes the resilience and necessary evolution of the TPACK model. Mishra argues that rather than diminishing the need for human expertise, the rise of AI makes deep Content Knowledge (CK) more vital than ever to verify and curate the output of algorithmic tools, which he characterizes as “smart drunk interns”; highly capable but fundamentally unreliable. The dialogue highlights the critical importance of Contextual Knowledge (XK) within the framework, emphasizing that the human understanding of specific students and local settings remains a domain where teachers hold a distinct advantage over de-contextualized algorithms. Furthermore, Mishra introduces the concept of “pride in work” as a crucial pedagogical strategy to foster student agency and creativity, serving as a counterbalance to the risks of automated mediocrity and plagiarism. Looking toward the future, the interview offers a critical vision, warning against the potential devaluation of “practical wisdom” and the widening of equity gaps. In all, Mishra suggests that the “Future-PACK” for educators must extend beyond technical skills to include a quasi-political understanding of the complex societal and ethical systems that shape the educational landscape.