Pillar 2Governance Produces: AI Principles Statement

AI Principles

Turn borrowed values into institution-specific principles that actually decide cases.

Aligns withOECD AI PrinciplesNIST · GovernISO/IEC 42001

The idea

Principles are the values layer: a small set of commitments (fairness, transparency, accountability, human oversight, privacy) that later decisions are meant to honor. Most institutions can state them. The problem is that a statement copied from elsewhere changes little. This pillar is about principles that are local, rooted in mission, and tied to actual decisions.

“Transparency” does little until you define what it requires, say, when a unit deploys an undisclosed admissions chatbot. So the pillar is built from concrete parts: an institution-specific principles statement; decision norms that turn each principle into procurement review questions; a named forum to resolve tensions between principles; and enough visibility that people know the principles exist.

Why it matters

Principles connect mission (Pillar 1) to the binding rules of policy (Pillar 3). Without them, policy is arbitrary; with them, it has a rationale you can defend to a faculty senate. They also let the institution say no to a tempting tool, and explain why, without re-arguing its values every time.

Small-unit formAdopt the campus principles, plus 2–3 local ones that fit your work.

How you can tell

When it's working
  • The principles are specific to this institution, each with a one-line rationale tied to mission, not a generic copied list.
  • The principles were co-designed with faculty, students, and staff and ratified through shared governance, not issued by a single office.
  • Each principle maps to concrete review questions used in procurement and pilot decisions.
  • There is a named body and a known path for resolving conflicts between, say, innovation and equity.
  • A student, faculty member, or staffer can find the principles in under five minutes.
When it's missing
  • Principles exist as a webpage nobody consults; decisions never cite them.
  • Conflicts among innovation, access, academic freedom, and risk are settled by whoever is loudest.
  • Every contested case starts the values argument over from zero.
  • Most of the campus doesn't know the principles exist.

How to operationalize it

Rate your campus

Each practice is something you can build, assign, and evidence, with a first move, an owner, and what proof looks like. Move one practice up one level at a time.

The five levels: 1 Initial · 2 Emerging · 3 Defined · 4 Managed · 5 Optimizing. Each practice describes levels 1, 3, and 5. Level 2, Emerging: some level-3 behavior exists in pockets, but informally. Level 4, Managed: level-3 behavior is applied consistently, measured, and reviewed on a set cadence.

1

Institutional AI principles statement

Institution-specific responsible-AI principles, aligned to mission and recognized standards.

Not yet rated
1InitialNo principles, or a generic list copied from elsewhere with no local meaning.
2EmergingGood practice exists in pockets, but it is informal and has not been institutionalized.
3DefinedInstitution-specific principles are adopted, with rationale tied to mission and standards (OECD, NIST).
4ManagedThe practice is applied consistently, measured, and reviewed on a defined cadence.
5OptimizingPrinciples are versioned, periodically re-ratified, and cited in policy and procurement decisions.
Next moves
  1. Draft 5–8 principles with one-line rationales using the Pillar 2 template.
  2. Validate against OECD AI Principles and institutional mission; ratify through governance.
  3. Cite the principles explicitly in every new AI policy and major procurement.
Who owns it

Provost · Faculty senate · AI governance body

Evidence it exists
  • Ratified principles statement
  • Mapping to OECD/NIST
  • Citations in later policies
2

Principles-to-decision norms

A defined way principles get applied to real decisions, procurement, pilots, classroom use.

Not yet rated
1InitialPrinciples, if they exist, are decoration; decisions don't reference them.
2EmergingGood practice exists in pockets, but it is informal and has not been institutionalized.
3DefinedDecision checklists or rubric questions derived from the principles are in routine use.
4ManagedThe practice is applied consistently, measured, and reviewed on a defined cadence.
5OptimizingDecision records show principle trade-offs being weighed and documented as a matter of course.
Next moves
  1. Translate each principle into 2–3 concrete review questions.
  2. Embed those questions in procurement review and the Compass Ethical/Legal category.
  3. Keep brief decision records noting which principles were in tension and how it was resolved.
Who owns it

AI governance body · Procurement

Evidence it exists
  • Decision checklist
  • Completed reviews referencing principles
  • Decision log
3

Ethical tension management

A named forum and process for conflicts among innovation, equity, academic freedom, and risk.

Not yet rated
1InitialTensions surface as ad hoc disputes and are settled by whoever is loudest.
2EmergingGood practice exists in pockets, but it is informal and has not been institutionalized.
3DefinedA defined escalation path routes contested AI questions to a deliberative body.
4ManagedThe practice is applied consistently, measured, and reviewed on a defined cadence.
5OptimizingPrecedents are recorded and reused; the institution learns from each resolved tension.
Next moves
  1. Designate the body that adjudicates contested AI questions and publish its escalation path.
  2. Run contested cases through structured deliberation with stakeholder voice.
  3. Maintain a precedent log so similar cases get consistent answers.
Who owns it

AI governance body · Faculty senate

Evidence it exists
  • Escalation path
  • Deliberation records
  • Precedent log
4

Principles awareness & visibility

People affected by the principles know they exist and can find them quickly.

Not yet rated
1InitialAlmost no one outside the drafting group knows the principles exist.
2EmergingGood practice exists in pockets, but it is informal and has not been institutionalized.
3DefinedPrinciples are findable in under five minutes and included in onboarding.
4ManagedThe practice is applied consistently, measured, and reviewed on a defined cadence.
5OptimizingAwareness is measured (surveys, usage), and gaps trigger communication fixes.
Next moves
  1. Apply the five-minute findability test from a student, faculty, and staff starting point.
  2. Add the principles to new-employee and new-student onboarding.
  3. Survey awareness annually and act on the result.
Who owns it

Communications · HR · Student affairs

Evidence it exists
  • Findability test results
  • Onboarding materials
  • Awareness survey data

In practice

Worked example

Transparency stops being a slogan

A community college's principles say 'transparency.' A department quietly launches an AI chatbot answering admissions questions; when challenged, staff note the principles 'don't mention chatbots.' Rather than argue, the AI governance body uses this pillar's machinery: it had already translated 'transparency' into three review questions, including 'are users told when they're interacting with AI?' Embedded in the procurement checklist. The chatbot is paused, a disclosure banner added, and the precedent logged so the next case is decided in minutes. The principle held because it had teeth.

Watch for

  • Adopting a long, aspirational list nobody can apply, five to eight principles with rationales beats fifteen slogans.
  • Leaving principles disconnected from procurement and the Compass, where the decisions happen.
  • Never re-ratifying them, so they lose authority as leadership and context change.

Readiness & maturity questions

Readiness asks: are we prepared? Use these to surface blind spots before you build. Every “no” shows where the work is. Representative prompts for reflection, not a scored test. For the scored version, use the Maturity Assessment.

  1. Have institutional AI principles been created?
  2. Are the principles inclusive of ethical concerns (e.g., bias, equity)?
  3. Were diverse stakeholders involved in their development?
  4. Are the principles publicly available?
  5. Are faculty and staff trained on their application?
  6. Do they apply to both academic and administrative AI?
  7. Are they aligned with global standards (e.g., NIST, UNESCO)?
  8. Is there a review cycle for the principles?
  9. Are principles used in tool evaluation?
  10. Are violations or misuses addressed via these principles?

Apply this pillar