This step collects the improvement opportunities arising from management review, corrective action, monitoring and other sources, and runs them as improvement tasks, so that the AI management system keeps getting better. You receive the list of improvement opportunities, the improvement task plan, and what to feed back into next year's operation.
It collects improvement opportunities from the decisions of 14 Management Review, the results of 15 Nonconformity and Corrective Action, the observations and improvement recommendations of monitoring, internal audit and legal compliance checks, and the opinions of interested parties. On that basis, it sets improvement tasks that continually raise the suitability, adequacy and effectiveness of the AI management system.
After checking implementation and effects, the results are fed back into next year's operation (1 Context Analysis, 5 Risk Assessment, 6 Impact Assessment, 7 SoA, AI objectives).
Output
- List of improvement opportunitiesSource (management review, corrective action, observation, monitoring, interested parties' opinions) / content / expected effect / priority
- Improvement task planPerson in charge, deadline, resources, performance indicators (before and after) and the method of checking effects
- What to feed back into next year's operationUpdates to the risk assessment, impact assessment, SoA, AI objectives and training
What to do in this step, by scope of application
| Development | Supply | Use |
|---|---|---|
| Retraining, model improvement and improvement of the development process | Improvement of terms of provision, user documentation and customer support | Expanding or narrowing the scope of use, and improving usage guidelines and training |
How improvement is run
ISO/IEC 42001 10.1 requires continual improvement of the suitability, adequacy and effectiveness of the AI management system. Decisions on improvement opportunities, changes and resources from the management review outputs (9.3.3) are received and run as improvement tasks, and the results of their implementation are fed back as input to the next management review.
Viewpoints for judgement
| Perspective | What is examined |
|---|---|
| Suitability | Does the AI management system fit the organization's purpose, AI policy and AI objectives? |
| Adequacy | Does it have the documents, processes and resources that meet the requirements of ISO/IEC 42001 and legal requirements? |
| Effectiveness | Does it deliver the intended results: achievement of AI objectives, incidents and nonconformities, accumulation of operating records |
| Strategic alignment | Is the use of AI linked to business strategy (productivity, competitiveness, customer requirements)? |
Sources of improvement opportunities
- Management review decisions and improvement recommendations (9.3.3)
- Internal audit observations and good practices: good practices are rolled out horizontally to other departments and AI systems
- AI objectives (KPI) not met: objective, result, judgment (achieved, in progress, not met) and the reasons for falling short
- Nonconformities and items to be confirmed from legal compliance checks
- Follow-up after corrective action: going beyond preventing recurrence to making things better
- Incident and monitoring trends
- Improvement suggestions from employees and departments
- Technological progress, regulatory changes and changes in the business environment: periodic review of policies, procedures, guidelines and forms (at least once a year)
Operating flow
- ① Identify improvement opportunitiesCollect them by source and assign a linking number (management review decision, audit observation, KPI)
- ② PrioritizeEffect, urgency, feasibility, risk
- ③ Plan the improvement taskPerson in charge, deadline, resources (linked to the resource decisions of the management review), performance indicators (current → target)
- ④ Implement
- ⑤ Check the effectCompare indicators before and after
- ⑥ Standardize and roll out horizontallyReflect in policies, procedures, guidelines and forms, and spread good practices
- ⑦ Report to the next management reviewAs "results of implementing previous actions", raising AI objectives after the first certification and establishing a half-yearly review
Common improvement tasks (examples)
Regular accumulation of operating records (evidence of implementation) · consistency between documents (baseline values, individual SoA controls, dates, approvals) · tracking completion of AI awareness training · regular evaluation of external AI providers · registration rate and quarterly checks of the AI system inventory · automation of repetitive tasks
Improvement operation indicators
| Indicator | Target | Formula and cycle |
|---|---|---|
| Improvement task completion rate | 100% | Tasks completed (including effect check) out of those due / quarter |
| Improvement task on-time rate | 90% or more | Completed within deadline / completed tasks |
| Share of AI objectives (KPI) achieved | Improvement over the previous year | Items achieved / total items / half-year |
| Recurrence of nonconformities with the same cause | 0 cases | Number of recurrences / year |
Output: improvement judgment (10.1) · degree of achievement of AI objectives · list of improvement opportunities · improvement task register · horizontal rollout of good practices · periodic review · proposals for adjusting AI objectives · what to feed back into next year's operation · input to the next management review (including Excel sheets).
Only the method has been generalized from ISO/IEC 42001 10.1 and actual customer operation cases (management review reports and procedures); company names and results are not included. Decisions and approvals are made by your top management.
Next steps
Previous steps: 14 Management Review · 15 Nonconformity and Corrective Action · 2 AI Policy, Roles and Objectives · 9 Operational Control · 11 Monitoring, Change and Revision Control · 12 Internal Audit · 13 Legal Requirements Identification and Monitoring · 10 Documented Information Control · The results of this step feed into: 1 AI Context Analysis · 2 AI Policy, Roles and Objectives · 3 AI System Selection · 5 AI Risk Assessment · 6 AI System Impact Assessment · 7 Control Implementation and Statement of Applicability (SoA) · 14 Management Review
