AI has been an endlessly rich topic of conversation in basically every professional space I’m part of. As a practitioner, researcher, and teacher in the learning design field, I think about it a lot. One question that I’ve kept turning in my brain: Why is it so tempting to use AI to maximize output, when the process is where real learning happens? Is it possible to shift the needle toward more process-oriented uses of AI?
I’ve been wondering about that second question often this semester while teaching a constructivist learning theories course. One of the through lines of the class is applying theory to practice, and through guided discussion, students have come to understand the real misalignment between the values that drive a lot of business environments — scale, speed, volume of output — and the conditions that actually support deep learning. Given that, it makes sense why generative AI tools are so attractive to students and practitioners. However, inserting tools that make it faster and easier to produce only increases the gap between the experiences we design and the effectiveness of our designs.
I’ve noticed this come out in the design projects students complete for the class. They are supposed to design a lightweight “blueprint” learning experience grounded in a specific theory. When I evaulate the work, I’m looking for harmony across the stated objectives, learning activities, and assessments against the backdrop of the theory of focus. The “paint job” — the polish on the final deliverable — can look convincing even when the underlying design thinking hasn’t really been done. What I rely on to understand the context of what’s actually happening in these designs is the reflections students write about their process and their design judgments, like the tradeoffs for specific choices. That’s where I see whether a student is genuinely engaging with the theory and the design process, or just producing something that looks like they did.
What the Research Says
A recent publication offered helpful vocabulary and scholarly rationale for what I was observing and trying to combat in my classes. Wang & Zhang (2026) look at how learners engage with generative AI in higher education, and their conclusions pull in two directions that I think are worth holding together. The first probably seems like a no brainer: the quality of engagement with AI relies on what the user brings to the interaction, not what they’re able to produce with it. Without the foundational knowledge to critically evaluate what AI generates, users risk being led astray and the outputs tend to show it. In my class, this resulted in design blueprints that were shallow, generic, and not strongly tied to any underlying theory or context. More pizzazz does not hide the lack of attention to the design process.
The second conclusion is the one I keep coming back to as a designer: what drives transformative learning isn’t how much you delegate to AI, but whether you bring critical evaluation to what it gives you. This intrigues my because I think this applies in two directions at once. Not only is it true for students learning with AI tools; it’s equally true for learning designers doing our work with these tools. The same conditions that support deep learning — genuine engagement with the theory, critical evaluation, reflective practice — are the conditions we should be building into our own design processes when AI is in the loop. This is why I’ve struggled to blanket-ban AI tools in my class and my practice. Instead, I think this is a compelling argument for knowing your problem, your theory, and your context well enough to evaluate what AI hands back to you.
Wang, S., Zhang, H. Pedagogical partnerships with generative AI in higher education: how dual cognitive pathways paradoxically enable transformative learning. Int J Educ Technol High Educ 23, 11 (2026). https://doi.org/10.1186/s41239-026-00585-x
The Conditions Don’t Change
What Wang & Zhang’s work ultimately reinforced for me is the lesson we in the edtech field have seen play out time and time again. The conditions that support transformative learning don’t change just because the tool does. Scale and speed are seductive, but they’re not design values. Knowing your problem deeply enough to evaluate what AI gives back to you is the key skill that we human learning designers bring to our work as our practice evolves to incorporate emerging AI tools.
In a follow-up post, I’ll get into what that actually looks like in practice, including how I use AI as a thinking partner in my own design process and the tools I’ve developed to help make that approach more intentional.