Then: do.capture.iterate.
A method for building AI literacy in institutions under uncertainty.
The Method
Anchored Adaptation™ is a method for building AI literacy in institutions when no fixed implementation can stay current. Rather than fixing implementation choices in advance, the institution commits to a disciplined cycle of experimentation and learning, held steady by a fixed conceptual anchor. The method treats institutional uncertainty as material to be worked, not a problem to be solved once.
In a fast-moving field, any specific implementation of AI literacy (the tools, the policies, the trainings) dates quickly. A strategy that fixes those implementation choices in advance is therefore mismatched to the problem by design: it is built to settle a question that will not hold still. Anchored Adaptation separates what should stay fixed from what should keep moving, and disciplines the movement so that adaptation produces compounding knowledge rather than drift.
One Commitment, Then a Three-Beat Loop
Small, real experiments rather than a single rollout. Each cycle takes on a genuine teaching, research, or operational problem, small enough to finish and real enough to matter.
Every cycle produces durable, reusable evidence: assessments, documented cases, templates, artifacts. Capture is what separates adaptation from churn.
Each pass closes with a decision to scale, revise, or stop, made on current evidence and given a revisit date. The cadence of re-deciding is the method.
The anchor holds; everything downstream adapts. A stable framework defines what AI literacy means, in terms durable enough to outlast specific tools. The method also knows its limits: legal, procurement, accessibility, and safety-critical questions take a fixed answer, and when cycles stop changing a decision, that question has settled.
Freely Shared
The canonical statement of the method: premise, loop, boundaries, lineage, and references. Permanently archived at DOI 10.18130/uva-open/61.
Download the definition CC BY-NC-ND 4.0 · share, don't modifyThe six steps for leaders and conveners together: who does what, the first semester week by week, common objections answered, and what should stay fixed. Permanently archived at DOI 10.18130/uva-open/62.
Download the playbook CC BY 4.0 · adapt with attributionCite the definition as: Lo, L. S. (2026). Anchored adaptation: A method for building AI literacy under uncertainty (Version 1.0) [Definition of record]. University of Virginia. https://doi.org/10.18130/uva-open/61
Cite the playbook as: Lo, L. S. (2026). The anchored adaptation playbook: Building AI literacy at your institution when no fixed plan can stay current. LibraOpen, University of Virginia. https://doi.org/10.18130/uva-open/62
For Academic Leaders
Provosts and chief academic officers sponsor the method rather than run it: adopt an anchor, sign the charge, and own the decisions the evidence feeds. The playbook addresses this sponsorship layer directly, alongside the convener’s work, so both readers hold the same document.
The method’s first application is underway at the University of Virginia through the AI Literacy and Action Lab. For advisory conversations, working sessions, or adoption questions, write to the author below.
Lineage
The method draws on traditions that manage uncertainty by holding something steady: Deming's PDSA learning cycle, developed from Shewhart's work; Mintzberg and Waters's umbrella strategy; agile's cadence of delivery and reflection; Bryk and colleagues' improvement science; and adaptive management, the closest predecessor. Its contribution is a combination and an institutional specification: it holds the definition of the capability itself stable, and binds each cycle's evidence to a documented, reversible institutional decision with a revisit date.