Case study

ALIGN

AI Levels for Instructional Guidance and Navigation

A college asked for AI-inclusive curriculum guidelines. What the team built instead of a policy was a shared vocabulary, set at the level of the assignment.

My part
Led the cross-functional faculty and staff team that created it, and co-authored the white paper, the faculty guides and the assignment-level checklists.
Where
Montgomery College, through the Digital Learning Center.
Status
Published as draft v3.3b, in its feedback stage.

The framework, running

This is the live site, not a picture of it. It loads as you reach it and unloads as you leave; on a phone it arrives at phone size with one tap between the page and the app, and the link opens the real thing.

Skip the embedded ALIGN site

An open question

Montgomery College's Academic Master Plan asked for AI-inclusive curriculum guidelines, and a goal set at that altitude does not tell you what to build. When I picked this work up, the question in front of the college was open. What do you say about generative AI in coursework when faculty disagree about it, and when that disagreement is pedagogically sound? A student can sit in two classrooms on the same day, one that bans AI outright and one that requires it, and then read a job posting that evening listing AI fluency as a requirement. Faculty disagreement was never the thing I wanted to fix, since a composition instructor and a nursing instructor have good reasons to draw the line in different places. My problem was that students could not read those differences, because every classroom described its expectations in language it had written for itself.

A vocabulary rather than a rule

A policy was the obvious shape for the work, but a policy is not what the team built. A single rule applied evenly across every discipline either bans AI in work where it genuinely belongs or permits it in work where the drafting is the point, and both failures land on faculty judgment instead of supporting it. I led a cross-functional faculty and staff team drawn from across the college. What we decided was to build a shared vocabulary rather than a rule, and to set that vocabulary at the level of the assignment.

Assignment level is the decision that carries the most weight, and it is the one I would defend first. A course-level policy cannot describe a course whose first essay allows no AI at any stage but whose final project asks students to design an AI workflow. Most courses in my experience turn out to have that shape once a faculty member looks at them assignment by assignment. Moving the scale down to the assignment moved the authority down with it, and faculty pick the level that fits the learning goal rather than the level the college hands them.

Adapted, not invented

We also decided not to invent a scale. ALIGN adapts the AI Assessment Scale (Perkins, Furze, Roe, and MacVaugh, 2024) to Montgomery College's context, so the team's work was translation rather than taxonomy. Though the six levels are recognizable to anyone who knows that literature, the work that mattered was making them usable inside this particular college.

The ALIGN scale: six levels of AI use in coursework, running from level zero, no AI use, to level five, innovation and exploration, with each level stepping further out along the scale.
The six levels, from no AI use at Level 0 to AI at the center of the work at Level 5.

What each level carries

Each level ships with three things attached rather than a definition standing on its own. A faculty member gets sample assignment language to paste into a syllabus, an example from a real Montgomery College course, and a statement of what visible student thinking looks like at that level. Level 0 allows no generative AI at any stage and keeps approved accessibility tools in place. Level 5 puts AI at the center and assesses the process documentation alongside the product. I co-authored the white paper, the practical faculty guides, and the assignment-level checklists that carry that language. The checklists are the piece I expect to see used most, since they are short enough to open while an assignment is being written.

What it refuses to be

The framework states its own limits on the page. ALIGN is not a mandate to use AI, and Level 0 is a full and legitimate choice rather than a floor. It is not a ranking, not a replacement for faculty judgment, and not a detection system. Five design principles hold it in place: faculty autonomy, transparency, human oversight, equity and access, and academic integrity. The site also carries a disclosure page describing how AI was used in producing the materials, which applies the framework's own transparency standard to the framework itself.

What this shows, and what it does not

I know what the team built and I know it is public. I do not know yet whether it changes what a student experiences across a semester, and I did not measure that. ALIGN v3.3b is a draft in its feedback stage, published with a feedback form beside it. That was deliberate, because in my experience a framework arriving as a finished mandate gets read as compliance. A companion tool, invite-gated for faculty, scores an assignment against the scale stage by stage, and my expectation is that the next round of evidence comes from there.

The next test

The next test is whether faculty reach for the language without being asked to. Ultimately the framework only works if a student can read one line in a syllabus, understand what is expected on that assignment, and carry that reading into the next course. The line the team put on the site is that clarity is a form of equity, and I would rather have my work judged on whether that holds up than on whether the framework is elegant.