Certificate Course in AI Leadership

Participants can frame AI opportunities as testable business use cases.

• Problem statements separating customer need, process failure and technology preference
• Jobs-to-be-Done interviews identifying high-friction decisions, tasks and handoffs
• Use-case canvases defining users, decisions, inputs, outputs and constraints
• Value hypotheses linking AI outputs to revenue, cost, quality or risk
• Process mapping reveals bottlenecks, exception volumes and decision latency
• Human-in-the-loop requirements specified for consequential decisions and exceptions

Participants can rank AI initiatives using transparent investment and risk criteria.

• Weighted scoring models balancing value, feasibility, risk, urgency and strategic fit
• Impact-effort matrices identify quick wins, strategic bets and deferred proposals
• Portfolio dependencies across data platforms, processes, vendors and shared capabilities
• Sequencing pilots before scale based on uncertainty, reversibility and control maturity
• Kill criteria establish when experimentation should stop or be redesigned
• Investment horizons distinguish immediate productivity gains from longer-term capability building

Participants can assess whether available information is suitable for an AI proposal.

• Data inventory identifies source systems, owners, sensitivity and permitted uses
• Data-quality dimensions measure completeness, accuracy, timeliness, consistency and validity
• Lineage mapping traces training, retrieval and operational data to source
• Privacy impact assessment triggers identified before personal data enters AI workflows
• Retention schedules and deletion controls applied to prompts, outputs and logs
• Data access models distinguish least privilege, role-based access and privileged access

Participants can question technical choices without needing to build AI models.

• Supervised learning predicts labelled outcomes from historical examples and features
• Unsupervised learning groups patterns where labels or outcomes remain unknown
• Generative models produce text, images, code and structured content probabilistically
• Retrieval-augmented generation grounds responses in approved organisational knowledge sources
• Model performance metrics distinguish accuracy, precision, recall, calibration and error costs
• Training, validation and test sets prevent misleading claims about model performance

Participants can specify controlled generative AI services for workforce use.

• Prompt patterns define role, task, context, constraints and output format
• Grounding strategies reduce unsupported responses through approved retrieval sources
• Human review thresholds match output consequences, complexity and uncertainty
• Content controls address confidential information, harmful outputs and intellectual property
• Shared-service models govern enterprise assistants, approved tools and support channels
• Usage policies distinguish permitted experimentation, sanctioned work and prohibited activities

Participants can construct a decision-ready business case for an AI initiative.

• Benefits maps connect AI capabilities to measurable operational and financial outcomes
• Baseline measurement establishes current cost, volume, quality and cycle-time performance
• Total cost of ownership includes data, integration, licences, controls and support
• Sensitivity analysis tests benefits against adoption, accuracy and volume assumptions
• Stage-gate funding links investment release to evidence, controls and adoption milestones
• Benefit owners accept accountability for measurement, process changes and realised value

Participants can identify and treat material risks associated with AI use cases.

• NIST AI RMF functions Govern, Map, Measure and Manage applied
• Risk scenarios cover bias, privacy, security, safety, misinformation and exclusion
• Impact assessments distinguish affected groups, severity, likelihood and reversibility
• Fairness testing compares error patterns across relevant demographic or operational groups
• Risk treatment selects avoidance, mitigation, transfer, acceptance or redesign actions
• Residual-risk statements record accountable approval and conditions for deployment

Participants can design governance structures that maintain accountable AI oversight.

• ISO/IEC 42001 management-system concepts applied to AI governance responsibilities
• Decision-rights matrices allocate sponsor, owner, developer, reviewer and approver roles
• Three-lines model clarifies management, risk oversight and independent assurance responsibilities
• Lifecycle control gates govern design, testing, deployment, monitoring and retirement
• Model inventories record purpose, owner, data, version, risk and approval status
• Escalation protocols define incident thresholds, response authority and stakeholder notification

Participants can translate applicable AI obligations into practical management controls.

• EU AI Act risk categories inform prohibited, high-risk and transparency decisions
• Data-protection principles guide lawful purpose, minimisation and accountable processing
• Consumer-protection risks arise from misleading, discriminatory or opaque automated decisions
• Records management preserves prompts, decisions, approvals, incidents and audit evidence
• Sector rules shape AI controls in banking, healthcare, public services and employment
• Internal policies connect acceptable use, procurement, security and disciplinary processes

Participants can evaluate AI suppliers and establish enforceable oversight requirements.

• Vendor due diligence examines architecture, data practices, security and financial viability
• Request-for-proposal criteria test performance evidence, transparency and implementation support
• Contract clauses address data ownership, model changes, audit rights and liability
• Service-level measures track availability, response quality, incident handling and support
• Supplier assurance evidence includes certifications, test results, policies and penetration reports
• Exit plans preserve data portability, business continuity and alternative service options

Participants can set leadership expectations for AI security and operational resilience.

• Threat modelling identifies assets, attackers, attack paths and mitigating controls
• Prompt injection risks addressed through input controls, isolation and tool permissions
• Adversarial testing probes harmful outputs, data leakage and instruction bypasses
• Identity controls govern users, service accounts, application programming interfaces and privileges
• Incident playbooks coordinate containment, investigation, communication and service recovery
• Business continuity plans address vendor outage, model failure and degraded manual operations

Participants can oversee AI delivery from discovery through controlled deployment.

• Product discovery validates user needs, workflow fit and measurable problem definitions
• Minimum viable products test assumptions without committing premature enterprise-scale investment
• Experiment design specifies comparison groups, success metrics and decision thresholds
• Acceptance criteria combine functional quality, safety, usability and control requirements
• MLOps practices manage versioning, deployment, monitoring and rollback of models
• Release governance authorises deployment only after evidence, training and control readiness

Participants can establish monitoring arrangements for deployed AI systems.

• Key risk indicators track drift, incidents, override rates and control failures
• Key performance indicators track quality, adoption, productivity and customer outcomes
• Data drift detection identifies changing input distributions and deteriorating model relevance
• Model drift reviews compare current outcomes with approved performance baselines
• Feedback loops capture user corrections, complaints, exceptions and unintended consequences
• Retirement decisions address obsolete models, unsupported vendors and unacceptable residual risk

Participants can lead workforce transition for AI-enabled changes to work.

• Role-impact assessments identify task shifts, decision changes and capability gaps
• Stakeholder mapping distinguishes sponsors, users, affected groups, critics and control owners
• Change narratives explain purpose, limits, safeguards and expected working practices
• Capability academies combine AI literacy, role-specific practice and manager reinforcement
• Consultation plans address worker concerns, professional standards and representative bodies
• Adoption dashboards track usage, proficiency, sentiment, workarounds and support demand

Participants can present AI decisions clearly to executive and oversight audiences.

• Executive decision papers state recommendation, value, risks, options and required approvals
• One-page dashboards combine portfolio status, benefits, risk exposure and decisions needed
• Assumption logs separate evidence, estimates, uncertainties and unresolved dependencies
• Scenario planning compares best case, expected case and adverse-case outcomes
• Governance meeting agendas focus discussion on exceptions, thresholds and accountable actions
• Plain-language communication explains model limitations without overstating certainty or capability

Participants can build, review and present a complete AI initiative leadership plan.

• Use-case portfolio selects priorities using documented value, feasibility and risk criteria
• Business case quantifies benefits, costs, assumptions, funding gates and accountable owners
• Governance design assigns decision rights, lifecycle controls and escalation routes
• Risk assessment records NIST AI RMF findings, treatments and residual-risk approvals
• Implementation roadmap sequences data, technology, workforce, supplier and assurance actions
• Executive presentation defends recommendations against stakeholder questions and challenge