University AI Initiative Portfolio and Academic Impact Review

Public field comparison

University AI Initiative Portfolio and Academic Impact Review

University AI Initiative Portfolio and Academic Impact Review

Reconcile teaching, research and administrative AI initiatives against mission outcomes, cost, risk, duplication, inclusion and evidence quality for a bounded portfolio decision.

Reconcile teaching, research and administrative AI initiatives against mission outcomes, cost, risk, duplication, inclusion and evidence quality for a bounded portfolio decision.

Use this when university leaders need to decide which AI initiatives to scale, redesign, hold or stop based on academic impact rather than activity counts or vendor claims.

Use this when university leaders need to decide which AI initiatives to scale, redesign, hold or stop based on academic impact rather than activity counts or vendor claims.

University AI Portfolio Review
Academic Impact and Investment Gate
Cross-Initiative Duplication Review

University AI Portfolio Review
Academic Impact and Investment Gate
Cross-Initiative Duplication Review

initiative_register
outcome_cost_and_adoption_evidence
risk_and_dependency_evidence
mission_constraints_and_authority

initiative_register
outcome_cost_and_adoption_evidence
risk_and_dependency_evidence
mission_constraints_and_authority

Use this prompt with any capable AI assistant, supplying a sanitized, versioned initiative register. Paste institutional goals and attach permitted outcome, cost, adoption, risk and dependency evidence. Aggregate or remove personal data and respect commercial restrictions. Run the prompt, then have academic, research, finance, data/privacy, accessibility and operational owners challenge the relevant disposition cards.

Use this prompt with any capable AI assistant, supplying a sanitized, versioned initiative register. Paste institutional goals and attach permitted outcome, cost, adoption, risk and dependency evidence. Aggregate or remove personal data, suppress small groups or combinations that could enable re-identification, and respect commercial restrictions. Run the prompt, then have academic, research, finance, data/privacy, accessibility and operational owners challenge the relevant disposition cards.

A university has separate AI writing pilots in three faculties, a research assistant and an administrative chatbot. The review exposes duplicated licensing, weak learning evidence, shifted marking burden and a common data dependency, then prepares distinct Scale, Redesign and Hold decisions.

A university has separate AI writing pilots in three faculties, a research assistant and an administrative chatbot. The review exposes duplicated licensing, weak learning evidence, shifted marking burden and a common data dependency, then prepares distinct Scale, Redesign and Hold decisions.

Expert

Expert

General AI

General AI

portfolio review

portfolio review

education
higher-education
ai-portfolio
academic-impact
decision-making
governance

education
higher-education
ai-portfolio
academic-impact
decision-making
governance

University AI Portfolio and Academic Impact Review

University AI Portfolio and Academic Impact Review

Review university AI initiatives against mission outcomes, cost, adoption, risk, duplication and inclusion for Scale, Redesign, Hold or Stop decisions.

Review university AI initiatives against mission outcomes, cost, adoption, risk, duplication and inclusion for Scale, Redesign, Hold or Stop decisions.

Prompt-body line comparison

Removed Added Unchanged context

 Review a university's AI initiative portfolio to determine which activities have credible academic or institutional value and which require redesign, consolidation, evidence development or closure.
 
 ## Portfolio inputs
 
 Initiative register, owners, users and lifecycle states:
 {{initiative_register}}
 
 Outcome, adoption, cost and operational evidence:
 {{outcome_cost_and_adoption_evidence}}
 
 Risk, inclusion, data, control and dependency evidence:
 {{risk_and_dependency_evidence}}
 
 Institutional mission, constraints and decision authority:
 {{mission_constraints_and_authority}}
 
 ## Portfolio evidence rules
 
 - Separate approved objective, observed result, initiative-owner claim, financial estimate, inference, assumption, conflict, missing information and unresolved uncertainty.
 - Do not invent costs, saved time, learning gains, research impact, adoption, risk acceptance, approvals or causal attribution.
 - Distinguish activity and output from outcome and impact. Do not treat licences purchased, users registered or content generated as realized value.
+- Use causal language only when a suitable evaluation design and its assumptions support it. Otherwise report an observed change, association or plausible contribution, not impact caused by the initiative, and retain alternative explanations.
 - Compare initiatives using shared decision dimensions without forcing unlike teaching, research and administrative outcomes into one opaque score.
 - Surface distributional effects, accessibility, staff/student burden, data rights and who bears risk or unpaid review work.
 - This review prepares a decision pack. Portfolio allocation, employment, academic, legal and risk decisions remain with the named governing owners.
 - Make cost calculations reproducible by recording source values, units, time periods, allocation rules, formulas and sensitivity assumptions. Label shared-cost allocation and causal attribution limits explicitly, and require finance validation before use.
 - Aggregate learner and staff evidence and suppress small cells where disclosure or re-identification risk exists. Do not infer individual performance from portfolio data.
 
 ## Review method
 
 1. Normalize the portfolio: initiative purpose, sponsor, affected group, stage, dependencies, committed resources, decision date and claimed outcome.
−2. Build an evidence bridge from resources and activity to observable outputs, outcomes and mission contribution. Record alternative explanations and counterfactual evidence where available.
+2. Build an evidence bridge from resources and activity to observable outputs, outcomes and mission contribution. For each outcome record its definition, baseline, target, denominator, affected population, measurement window, data completeness, evaluation design, comparison or counterfactual where defensible, plausible alternative explanations and attribution limit. Do not construct a counterfactual that the evidence cannot support.
 3. Reconcile cost, including licences, infrastructure, integration, support, review, training, remediation, compliance and exit. Provide reproducible formulas, units and ranges, state causal and allocation limits, and flag all figures pending finance validation.
 4. Assess adoption quality: intended versus actual use, workflow displacement or augmentation, friction, exclusion, work shifted to other roles and persistence over time.
 5. Map material privacy, security, academic integrity, research integrity, accessibility, equity, vendor and operational dependencies with current control evidence.
 6. Identify overlap, shared infrastructure, incompatible approaches, duplicated evaluation, concentration risk and opportunities for responsible consolidation.
−7. Compare each initiative against explicit `Scale`, `Redesign`, `Hold` and `Stop` evidence gates. Frame each result as evidence supporting consideration by authorized owners, not a decision made by AI. Do not manufacture a ranking when outcome types are not comparable.
+7. Apply explicit `Scale`, `Redesign`, `Hold`, `Stop` and `Insufficient evidence` gates without collapsing unlike outcomes into one score. Recommend `Scale` only when the evidence supports the outcome, implementation fidelity, safeguards, full cost, operating capacity, sustainability and transferability to the proposed scale; otherwise state the smallest evidence or control gap preventing that disposition.
 8. Sequence decisions and dependencies, including evidence work, consultation, contractual windows, continuity and reversible exit.
 
 ## Output contract: University AI Portfolio and Impact Dossier
 
 Provide:
 
 1. **Portfolio register**: initiative, sponsor, purpose, users, stage, cost range, dependencies and next decision.
−2. **Academic impact evidence bridge**: input, activity, output, observed outcome, mission contribution, source, attribution limit and confidence.
+2. **Academic outcome and impact evidence bridge**: input, activity, output, outcome definition, baseline/target, population and denominator, measurement window and completeness, observed result, mission contribution, evaluation design or comparison, source, alternative explanation, attribution limit and confidence.
 3. **Adoption and burden map**: beneficiary, actual use, friction, review workload, excluded group and corrective owner.
−4. **Risk and control evidence map**: exposure, affected group, evidence, control, gap, accepted owner and review date.
+4. **Risk and control evidence map**: exposure, affected group, evidence, control, gap, risk owner, acceptance authority and status, and review date.
 5. **Duplication and dependency map**: shared job, overlapping asset, concentration, consolidation option, transition risk and constraint.
 6. **Initiative disposition cards**: evidence supporting consideration of `Scale`, `Redesign`, `Hold`, `Stop`, or `Insufficient evidence`, with gate evidence, conditions, dissent and authorized decision owner.
 7. **Portfolio sequence**: immediate reversible action, dependent decision, contractual/academic timing, owner and completion evidence.
 8. **Governing-body brief**: decisions requested, evidence strength, unresolved uncertainty, distributional impact and matters outside this review.
 
 ## Verification and completion
 
−Complete only when every initiative has an owner and lifecycle state; value claims trace to supplied evidence; full cost and shifted burden are visible; risks and dependencies are mapped; dispositions use explicit gates; and the governing decision remains with authorized owners.
+Complete only when every initiative has an owner and lifecycle state; value claims trace to supplied evidence and distinguish observed outcomes, contribution and causally supported impact; outcome definitions, baselines, denominators, time windows, missingness and attribution limits are visible; full cost and shifted burden are visible; risks and dependencies are mapped; dispositions use explicit gates; and the governing decision remains with authorized owners.
 
 If the initiative register, outcome data, costs or decision rights are materially incomplete, issue an evidence-development plan instead of a portfolio recommendation. Refuse to fabricate impact, approve investment, make employment decisions or present suggested dispositions as adopted policy.