Beyond the Hype: How Embedded AI Can Help Higher Education Move from Experimentation to Instructional Impact
Go beyond the poster session with D2L and Digital Promise’s whitepaper.
topics
AI tools are becoming more common across higher education. But making a tool available doesn’t mean educators will use it—or see a reason to.
For faculty, instructional designers and LMS administrators, meaningful adoption depends on more than a tool’s features. It depends on whether the technology fits into existing workflows, supports professional judgment and gives educators a safe place to learn by trying.
To explore what helps educators adopt an AI tool in meaningful ways, researchers formed a research-practice partnership (RPP). The work involved Yoon Ha Choi of Digital Promise, Alice Wu Swift of the University of Hawaiʻi System and UH Online Innovation Center, Ashley Morris Biddle of Leeward Community College, and 26 practitioners from seven higher education institutions. With funding from D2L, the partnership examined how educators used D2L Lumi, D2L’s AI tool embedded in D2L Brightspace, for instructional support.
The findings offer useful lessons for institutions considering how to introduce AI in ways that reflect educators’ needs.
The partnership also brought practitioner perspectives into a conversation that can otherwise focus mainly on what AI tools can do. As Ashley Biddle, associate professor of psychology at Leeward Community College, explains, “There are a lot of things about being an instructor that you can’t know unless you do it. There are also things that are no big deal when you’re teaching one course, but a huge deal when you’re teaching six courses.”
This point speaks to the value of understanding how AI works in different teaching contexts—not just what it can do in theory.
Start with the work, not the technology
Workflow integration was important to participants.
Participants valued D2L Lumi because it supported work they were already doing. One faculty participant specifically appreciated being able to batch questions for review while staying within the LMS. The tool was fast enough to be useful and prompted the educator to check that the questions were accurate and aligned.
When educators have to leave familiar systems, open another platform and learn a new process, even useful technology can feel like one more thing to manage. When AI support like D2L Lumi is built into existing workflows, educators can more easily decide whether it’s right for a particular task.
Keep educators in control
The research also showed that educators have different ideas about which parts of their work AI should support.
One faculty participant explained that creating assignments was one of the parts of teaching they valued most. Understanding their students, working with course material and finding creative ways to engage learners were not tasks they wanted to hand over to AI.
The findings suggest that educators want the freedom to decide where AI can help and where human expertise should remain at the center. Meaningful adoption doesn’t require educators to use AI for everything. Trust also depends on oversight. One participant valued D2L Lumi’s features that prompted them to review and confirm that AI-generated questions were accurate and aligned before using them.
As a psychology instructor, I ground my teaching in Deci and Ryan’s Self-Determination Theory, which says that all people need to feel a sense of autonomy, competence and relatedness in order to be motivated. Most instructors teach because they enjoy the challenge of imparting knowledge to students and live for the moments when students’ eyes light up with understanding. I don’t want AI to get these moments—I want AI to help me with menial tasks, such as writing catchy announcements.
-Ashley Biddle, associate professor of psychology at Leeward Community College
Make experimentation feel manageable
Participants described D2L Lumi as a low-stakes way for instructors who were new to AI to begin experimenting. Ashley Biddle began the RPP with some skepticism about D2L Lumi. “It really took my favorite instructional designer remaking a part of one of my courses—using it for practice questions—to convince me it was worth a try,” she says. “Like all new things to adopt, I think we’re hesitant at first until there’s firsthand experience or we hear of someone we respect using it.”
A familiar environment, a focused instructional task and opportunities to review or revise the output helped make that experimentation more manageable.
Participants said the RPP motivated them to try D2L Lumi more intentionally and recommend it to others.
Build opportunities to learn together
The partnership created another condition that supported adoption: connection with peers.
The RPP provided a unique opportunity for faculty, staff and researchers to come together and share various perspectives and use cases. Being able to hear from others in different roles and contexts really allowed me to expand my thinking, both from a teaching perspective and from an instructional design and faculty support perspective.
-Alice Wu Swift, hospitality lecturer at University of Hawaiʻi System at Mānoa and instructional desinger, UH Online Innovation Center
Participants took part in institution-specific focus groups and interviews, along with three cross-institution meetings over five months. These conversations brought together faculty, staff and instructional designers from different institutions and disciplines.
Participants said the partnership encouraged them to talk about D2L Lumi with colleagues, including instructors who were new to AI. Those conversations helped move AI adoption beyond individual experimentation and into a shared process of learning.
These findings suggest that institutions could create similar opportunities through communities of practice, cross-functional working groups and regular time to share examples and questions. Educators need to learn not only how a tool works, but when and why it makes sense to use it.
The takeaway is clear: meaningful AI strategies start with educator needs. They fit naturally into existing workflows, respect professional judgment and make room for careful experimentation. The researchers will discuss their findings during the poster session AI Use for Instructional Support: Insights and Implications from a Research-Practice Partnership at the 2026 EDUCAUSE Annual Conference on September 30, from 4:30–5:30pm MT in the EDUCAUSE Commons poster area.
The RPP was all about collaboration and learning. We want to continue these types of conversations with EDUCAUSE attendees and invite you to stop by our poster.
-Yoon Ha Choi, postsecondary senior researcher, Digital Promise
AI adoption isn’t about asking educators to hand over their work. It’s about giving them the support and space to decide where AI belongs—and where it doesn’t.
Written by:
