AI and Future of Work Training Course

Participants can distinguish major AI approaches and match them to appropriate business problems.

• Machine learning classifications, forecasts and anomaly detection for operational decisions
• Natural language processing supports document extraction, search and case triage
• Large language models generate, summarise and classify unstructured business content
• Computer vision identifies objects, defects and events from image data
• Rules engines, robotic process automation and AI compared systematically
• Use-case patterns for customer service, fraud, procurement and workforce planning

Participants can map a process and identify suitable AI intervention points.

• SIPOC diagram defines suppliers, inputs, process steps, outputs and customers
• Swimlane maps show handoffs, decision rights and recurring delays
• Process mining concepts reveal variants, rework loops and bottlenecks
• Identify repetitive, rules-based and information-intensive tasks for AI assessment
• Exception analysis separates standard cases from high-risk specialist decisions
• Service blueprint links front-stage interactions with supporting operational work

Participants can assess whether available data is fit for an intended AI use case.

• Data inventory records source systems, owners, fields, access and retention
• Completeness, accuracy, timeliness, consistency and uniqueness measure data quality
• Training, validation and production data require documented separation
• Data lineage traces business records through transformations to AI outputs
• Sensitive personal data requires purpose limitation and access controls
• Representative samples reduce distorted outcomes from incomplete historical records

Participants can define safe, useful generative AI applications for knowledge work.

• Prompt structure specifies role, task, context, constraints and output format
• Retrieval-augmented generation grounds responses in approved organisational knowledge
• Hallucination risks require source checking, confidence thresholds and human review
• Prompt libraries standardise recurring drafting, analysis and summarisation tasks
• Output evaluation checks accuracy, relevance, tone, completeness and citations
• Confidentiality rules govern material entered into public AI services

Participants can assess AI effects on tasks, roles, capacity and workforce demand.

• Task inventory records frequency, effort, complexity, discretion and error consequences
• Human-machine allocation assigns automation, augmentation and human-only activities
• Role impact heatmap compares task change across affected job families
• Capacity modelling estimates hours released, redeployed and newly required
• Skills taxonomy identifies technical, analytical, ethical and interpersonal requirements
• Workforce scenarios test adoption rates, attrition assumptions and service demand

Participants can redesign AI-supported work while preserving accountability and job quality.

• Human-centred design balances user needs, worker experience and service outcomes
• Decision rights define who approves, overrides and escalates AI recommendations
• Cognitive load assessment identifies overload from alerts, reviews and exceptions
• Job crafting options reshape tasks, autonomy, feedback and skill variety
• Accessibility requirements include assistive technology, language and inclusive interfaces
• Standard operating procedures document AI-supported steps and handoff rules

Participants can establish governance principles and accountabilities for AI use.

• OECD AI Principles inform accountability, transparency, fairness and human oversight
• ISO/IEC 42001 concepts structure AI management system responsibilities
• AI policy defines approved uses, prohibited uses and escalation routes
• RACI matrix assigns business, technical, risk and oversight responsibilities
• Model inventory records purpose, owner, data, version and control status
• Governance forums review material risks, incidents, changes and performance

Participants can identify AI risks and specify proportionate preventative and detective controls.

• NIST AI RMF Govern, Map, Measure and Manage functions applied
• Risk register captures harms, likelihood, impact, owners and treatment actions
• Bias testing examines disparate outcomes across relevant user groups
• Privacy impact assessment identifies collection, access, retention and disclosure risks
• Red teaming tests misuse, prompt injection and unsafe model behaviour
• Control design combines technical safeguards, procedures, training and monitoring

Participants can evaluate AI suppliers and translate requirements into procurement controls.

• Requirements specification defines intended use, users, data and performance thresholds
• Vendor questionnaires examine model provenance, security, testing and support
• Contract clauses address data use, intellectual property and audit rights
• Service-level measures cover availability, response time, incident reporting and fixes
• Third-party risk assessment evaluates subcontractors, hosting and cross-border transfers
• Proof-of-value criteria establish evidence required before scaled deployment

Participants can select an operating model and define delivery responsibilities for AI initiatives.

• Centralised, federated and hub-and-spoke AI operating models compared
• Product team roles include sponsor, owner, users, data and risk specialists
• Stage-gate governance moves concepts through testing, approval and release
• Backlog prioritisation balances value, risk reduction, dependencies and effort
• Decision logs record assumptions, approvals, exceptions and unresolved issues
• Communities of practice share patterns, controls and reusable assets

Participants can plan stakeholder engagement and behaviour change for AI-enabled work.

• ADKAR model diagnoses awareness, desire, knowledge, ability and reinforcement gaps
• Stakeholder mapping identifies influence, impact, concerns and engagement actions
• Change narrative explains business purpose, worker implications and safeguards
• Manager toolkits support local conversations, demonstrations and feedback collection
• Resistance analysis distinguishes capability gaps, trust concerns and incentive conflicts
• Adoption metrics track active use, quality, workarounds and support demand

Participants can design role-based AI learning pathways linked to changed work.

• Skills gap analysis compares current capabilities with future task requirements
• Role-based curricula differentiate users, managers, specialists and control functions
• Learning pathways combine awareness, applied practice, coaching and assessment
• AI literacy covers limitations, verification, privacy and accountable use
• Performance support embeds checklists, prompts and escalation guidance into work
• Competency assessment uses observed tasks, scenarios and evidence criteria

Participants can plan a controlled AI pilot and scale decision.

• Pilot charter defines hypothesis, scope, users, duration and decision criteria
• Minimum viable process identifies essential workflow, controls and integration needs
• Phased roadmap sequences discovery, pilot, evaluation, release and scale-up
• Dependency mapping identifies data, technology, legal, procurement and workforce constraints
• Pilot cohorts select representative users, cases and service conditions
• Scale decision criteria combine benefits evidence, safety results and readiness

Participants can define measures that track AI value, quality, risk and adoption.

• Benefits map links AI outputs to operational, customer and financial outcomes
• Baseline measurement establishes pre-implementation cost, quality and cycle-time performance
• Balanced scorecard combines customer, process, people, financial and risk measures
• Model performance monitoring tracks accuracy, drift, latency and failure rates
• Human oversight metrics record overrides, escalations, corrections and review times
• Benefit realisation reviews compare expected outcomes with observed operational results

Participants can adapt AI work transformation methods to sector-specific constraints.

• Public-service casework balances eligibility automation with appeal and review rights
• Banking fraud detection combines model alerts, investigator queues and customer treatment
• Development programme targeting tests data quality, inclusion and community accountability
• Healthcare scheduling illustrates capacity gains, safety checks and clinical oversight
• Procurement analytics identifies spend anomalies, supplier risk and approval controls
• Case comparison identifies transferable patterns and sector-specific control requirements

Participants can prepare response processes for AI failures, complaints and performance changes.

• Incident taxonomy distinguishes safety, privacy, bias, security and service failures
• Triage protocol assigns severity, containment actions, owners and communication routes
• Root cause analysis uses five whys and fishbone diagrams
• Correction workflow updates prompts, data, rules, models or operating procedures
• Post-incident review records lessons, control changes and accountable approvals
• Continuous monitoring schedules performance reviews, audits and user feedback

Participants can build, review and present a complete AI work transformation case with value, workforce, governance and implementation plans.

• Problem statement defines service, users, pain points and intended outcomes
• Process map identifies tasks, decisions, handoffs and proposed AI interventions
• Use-case assessment scores value, feasibility, data readiness and material risks
• Workforce plan specifies role impacts, skills actions and change responsibilities
• Governance design assigns controls, decision rights, monitoring and escalation routes
• Implementation roadmap sequences pilot, evaluation, release, measures and review
• Executive presentation communicates recommendation, evidence, trade-offs and next decisions