The Filing Cabinet
Reclaiming Online Pedagogy from the Canvas Repository
For most of my career, teaching online graduate courses at Portland State University, my Canvas pages looked like most faculty pages: a list of links. Blue hyperlinks on a white background, organized by week, pointing students toward PDFs and discussion boards. Functional. Legible. It was, I now realize, the digital equivalent of handing someone a filing cabinet and calling it a classroom.
I never thought much about my course design. Most of us don’t. We’re hired for our disciplinary expertise, not our fluency with HTML or our eye for visual hierarchy. The LMS becomes a container we fill with content, where the default template—plain text and the occasional bolded heading—ends up defining the student experience.
That changed when I started playing around with generative AI. What began as curiosity about syllabus formatting turned into a months-long overhaul of five graduate courses, spanning education policy, social foundations, and curriculum theory.
The Repository Trap
Online higher education has a dirty secret: most of us have never been taught how to design an online course. We’ve been told to “put it on Canvas,” and we comply. The result is a repository, not a learning environment. Students navigate modules that look like file directories, with readings listed but never contextualized.
Take a typical module from my old “filing cabinet” days. A student would see a bare list of links: Arafeh_2015.pdf or Dawson-Amoah_2024.pdf. Now, that same module is a structured experience. It starts with a clear overview of the policy cycle, moves into specific learning objectives, and provides a “This Week’s Readings” section where each source is properly cited and accompanied by its corresponding lecture slides. There is a visual language now that suggests how much thought went into curating the material.
Stoesz and Niknam (2022) found that how a course looks—its visual design—directly correlates with how much students engage, even when the content stays the same. Consistency, typography, and layout are proven predictors of whether a student can use a site effectively (Ghai & Tandon, 2022). Visually deliberate design increases understanding and motivation, yet no one trains faculty to consider any of it.
Graduate students in programs serving working adults are logging into Canvas at ten o’clock at night after a full workday, after getting their kids to bed, after answering the last email from their supervisor. The interface they encounter either invites them in or tells them what follows is just another obligation to check off. A wall of unstyled links does the latter.
Ask the students. Scroll through any university subreddit, and you’ll find the same complaints: courses that feel self-taught, professors who haven’t posted a single announcement since week three, modules that go up late with no explanation, and grades that don’t appear until the term is nearly over. One student asked why their online courses amounted to self-study. Another said they’d never even seen their professor’s face. A third said it was strange that no instructor had assessed their work by week seven. The most common defense from other students was that this is just how online classes work everywhere—as if that made it acceptable. What none of them described was a course that seemed built for them. What they described was abandonment with a syllabus.
Designing for the Midnight Learner
Working with AI, I developed a consistent visual identity for each course using institutional green branding: custom HTML module pages with gradient banners, icon-based navigation cards, and color-coded sections for readings, discussions, and assignments. Each weekly module included a header graphic identifying the theme, organized content sections with visual cues, callout boxes for key dates and reminders, and consistent typography throughout.
I created custom announcement banners: nine distinct graphics for recurring announcement types such as “My Commitment to You,” “Writing Guidelines,” and “How to Structure Your Final Paper.” Instead of plain text blocks that students scrolled past, announcements became visually distinct and identifiable.
Every assigned reading came with a PowerPoint presentation and speaker notes. These were not meant to replace the readings but to scaffold them, so that a student encountering a dense argument about theory in adult education or a policy fact sheet on school funding had an entry point before turning to the text itself.
Each course homepage became a landing page: a welcome space with the instructor’s photo, course navigation tiles, a quick-start guide, and links to self-care resources. Not because wellness content is fashionable, but because adult learners managing work, family, and school need to know their instructor sees them as whole people.
Bridging the Technical Divide
I did not learn CSS. I did not take a course in instructional design. I did not suddenly develop graphic design skills this late in my career. I acted as a creative director, translating the vision in my head into prompts until the AI produced the right code.
Other faculty are discovering this, too. Fang and Broussard (2024) describe how generative AI can help faculty who know their subject thoroughly but have no training in course design. AI tools can generate module structures, draft rubrics, recommend layout improvements, and transform basic outlines into visual presentations. AI serves as a “springboard,” but the intellectual judgment about what to keep, revise, or discard remains with the instructor. Conklin, Dorgan, and Barreto (2024) documented a similar process in which faculty used ChatGPT to develop a full online course, including HTML and CSS code for the LMS, cutting total development time to roughly 55 hours—compared to the six to nine months typically required.
Some colleagues find this uncomfortable. If you didn’t hand-code it yourself, if you didn’t learn the technical skill from scratch, the result doesn’t count. But that mischaracterizes the work. The design decisions were mine. The content structure was mine. The choice to create visual scaffolding for complex readings, to brand announcements so students would read them, to build a homepage that told students they were expected and welcome—those were teaching decisions. They came from years of working with adult learners who log in exhausted and still want to be challenged.
Without AI, I wouldn’t have attempted it. The vision was there. The technical barrier was beyond what most faculty could manage on their own, and, where they exist, instructional design offices are understaffed and overcommitted. Figure it out or accept the default—that’s the message. Most of us accept the default.
The Invisible Labor of Care
Redesigning these courses forced me to think about what happens when a student opens a module. Was I asking them to navigate a filing system, or was I inviting them into an intellectual space? Was I communicating that I had spent time preparing this material for them, or was I signaling that the LMS was a box to check?
The answer, for years, was the second. I suspect it’s the same for most faculty teaching online. The tools required skills we were never trained in, and nobody showed us what was possible.
This work stays invisible. No tenure file includes screenshots of a well-designed Canvas module. No annual review rewards the faculty member who spent forty hours building custom HTML pages so that working parents could navigate course materials more intuitively at midnight. Forty hours is forty hours. None of it shows up in a personnel file. The impact on student experience, however, is measurable.
As college becomes increasingly unaffordable and students expect more for their investment, we ought to ask what we mean by quality in online education. If the answer is a list of links, we can automate that. If it’s a learning environment that communicates seriousness, attentiveness, and respect for the adult learner’s time, then we need to invest in the conditions that make that possible. Or at least stop penalizing the faculty who do it on their own.
My courses look different now. They feel different. To me and, I believe, to my students. I’m not willing to act like that doesn’t count. It shouldn’t have taken artificial intelligence to make it happen. But it did, and I’m not going to pretend otherwise.
References
Conklin, S., Dorgan, T., & Barreto, D. (2024). Is AI the new course creator? Discover Education, 3, 285. https://doi.org/10.1007/s44217-024-00386-2
Fang, B., & Broussard, K. (2024, August 7). Augmented course design: Using AI to boost efficiency and expand capacity. EDUCAUSE Review. https://er.educause.edu/articles/2024/8/augmented-course-design-using-ai-to-boost-efficiency-and-expand-capacity
Ghai, A., & Tandon, U. (2022). Analyzing the impact of aesthetic visual design on usability of e-learning: An emerging economy perspective. Higher Learning Research Communications, 12(2), 1–22. https://doi.org/10.18870/hlrc.v12i2.1325
Mallary, K. J., Moore, E. J., & McClain, A. L. (2025). Artificial intelligence and universal design for learning: Transforming teaching and learning in adult and continuing higher education. New Directions for Adult and Continuing Education, 2025(188), 39–47. https://doi.org/10.1002/ace.70016
Stoesz, B. M., & Niknam, M. (2022). Student perceptions of the visual design of learning management systems. Canadian Journal of Learning and Technology, 48(3). https://doi.org/10.21432