AI for Government & Public Service Delivery

Course outline and learning expectations

This is a tutor-led course. The outline shows what your tutor will cover; teaching materials and examinations are provided directly to enrolled students.

What you will learn

• Assess where AI is suitable, unsuitable or unnecessary in public-service delivery
• Frame public-sector problems as measurable and responsible data or AI projects
• Prepare and evaluate government datasets for completeness, consistency, bias and fitness for purpose
• Build and evaluate models for predicting demand for health, education, water and other public services
• Apply transparent optimisation and prioritisation methods to constrained public resources
• Use anomaly-detection methods to flag possible fraud or leakage for human investigation
• Analyse citizen complaints, service ratings and other feedback while recognising language and context limitations
• Apply privacy, fairness, explainability, accountability and human-oversight principles to public-sector AI
• Build and present a public-service prediction or analysis tool with documented assumptions, limitations and safeguards

Modules

1. 1

   AI in Public Service Delivery

   Evaluate where AI can add value and where simpler approaches are safer or more effective

   Public-service problem framingRealistic versus hyped use casesAI suitability assessmentBaseline and non-AI alternativesDecision support versus automated decisionsHuman oversight and service-user impact
2. 2

   Predicting Public-Service Demand

   Develop models that estimate demand for essential services using administrative and contextual data

   Health-centre demandSchool enrolment and capacityWater-point usage and maintenanceTarget variables and featuresRegression and classificationTime-based validationError metrics and uncertainty
3. 3

   Resource Allocation Optimisation

   Prioritise limited public resources using transparent objectives, constraints and fairness checks

   Allocation objectivesBudget and capacity constraintsPriority scoringOptimisation basicsScenario analysisGeographic equitySensitivity analysisCommunicating trade-offs
4. 4

   Fraud and Leakage Detection

   Use anomaly screening to identify unusual public-programme records that require human investigation

   Fraud and leakage risk indicatorsRule-based controlsOutlier and anomaly detectionClass imbalanceFalse positivesInvestigation workflowAudit trailsHuman review and due process
5. 5

   Citizen Feedback Analysis

   Turn complaints, service ratings and public comments into useful service-improvement evidence

   Feedback collection channelsText cleaningComplaint classificationTopic analysisSentiment limitationsLocal-language and code-switching challengesTrend reportingPrivacy-preserving analysis
6. 6

   Government Data Quality and Interoperability

   Assess and improve the reliability of administrative datasets used for public-sector AI

   Missing and outdated recordsDuplicate and inconsistent entriesIdentifiers and data linkageSampling and reporting biasMetadata and data dictionariesData provenanceInteroperability across MDAs and local governmentsQuality monitoring
7. 7

   Ethics, Transparency and Accountability

   Design public-sector AI systems that respect rights, law, institutional responsibilities and public trust

   Uganda Data Protection and Privacy Act 2019African Union Continental AI StrategyPrivacy and data minimisationFairness and discrimination risksExplainabilityImpact assessmentsModel and dataset documentationAuditabilityAppeals, redress and human accountability
8. 8

   Capstone Public-Service Prediction Tool

   Build a tool that predicts or analyses which sub-counties may be underserved by health facilities using public administrative data

   Problem and stakeholder definitionOpen, synthetic or anonymised data selectionData-quality assessmentBaseline modelFeature engineeringValidation and error analysisGeographic comparison or mappingExplainable resultsSafeguards and limitationsPresentation to public-sector decision makers