How to use the AI commentary for SFIA skills

The commentary is designed to be read alongside the SFIA skill description and level descriptions, not as a replacement for them.

The examples are illustrative rather than exhaustive. They do not predict exactly how work will change or prescribe particular tools. They also do not indicate that an employer has permitted a particular use of AI. Any use remains subject to the organisation’s policies, controls, technologies, ways of working and circumstances.

Throughout the commentary, may identifies a possibility. It does not establish that a tool can perform an activity reliably, that an organisation permits its use or that a practitioner should use it.

How the four headings work together

Read together, the four sections answer complementary questions:

• Potential uses of AI in this skill: Where might AI assist the work?
• Why AI-generated output does not prove someone has this skill: Why is the resulting output not evidence of competence by itself?
• What practitioners remain responsible for: What must practitioners still understand, judge, decide and take responsibility for?
• How AI may affect work at each responsibility level: How might the balance of work change at the levels where the skill is defined?

What to expect under each heading

Potential uses of AI in this skill

This section describes activities that AI may assist within the scope of the skill. Depending on the skill, these may include:

• collecting, organising, classifying or synthesising information
• generating candidate content, designs, models, code, plans or other working material
• comparing options and identifying patterns, anomalies, dependencies or possible gaps
• supporting analysis, simulation, testing, monitoring, workflows and reporting.

The section considers more than faster production. It may identify how AI could broaden the options explored or shift attention towards validation, interpretation and decisions.

In some skills, AI is also part of the subject being addressed. Practitioners may need to evaluate, design, introduce, govern or monitor AI-enabled technologies and ways of working, as well as use AI to assist their own work.

Illustration – Project management PRMG: AI may help prepare and compare project plans, identify dependencies and exceptions, and draft progress reports. The practitioner can then concentrate more attention on checking the information and deciding what action is needed.

Why AI-generated output does not prove someone has this skill

This section distinguishes an output from the professional capability needed to produce, assess and use it appropriately. “Output” may include a generated work product, analysis, recommendation, classification, automated result or action supported by AI.

A plausible output does not by itself demonstrate that someone:

• understood the need, context and constraints
• selected suitable information, methods and tools, and checked their assumptions
• established the accuracy, completeness, source, reliability and relevance of the material
• interpreted findings, risks and consequences correctly
• engaged appropriate stakeholders and exercised the judgement, responsibility or authority required by the skill.

This distinction is important for recruitment, assessment and certification. Evidence of competent performance needs to show what the practitioner understood, decided, checked and took responsibility for, not merely what a tool produced.

Illustration – Project management PRMG: An AI-generated plan or dashboard does not demonstrate that the approach suits the project, that constraints have been agreed or that risks and exceptions have been judged correctly.

What practitioners remain responsible for

This section identifies the responsibilities that continue to define competent performance when AI is used. The details vary by skill, but commonly include:

• framing the work and selecting an appropriate approach
• validating inputs, assumptions, generated material and results
• interpreting exceptions, risks, trade-offs and wider consequences
• applying relevant standards, policies and controls
• collaborating, reaching agreement and making decisions within the practitioner’s remit.

The section therefore goes beyond a generic instruction to “check the AI”. It connects practitioner responsibility to the purpose, decisions, relationships and outcomes of the skill.

Illustration – Project management PRMG: Practitioners remain responsible for selecting an appropriate approach, balancing scope, cost, timescales and quality, involving stakeholders and deciding how to respond to risks and changes.

How AI may affect work at each responsibility level

This section considers how AI may change the mix and emphasis of work at the levels where the skill is defined. AI may reduce some routine preparation, production or consolidation while increasing the attention given to checking, interpretation, professional judgement, communication and control.

The analysis remains anchored in the SFIA levels of responsibility. It reflects differences in autonomy, influence, complexity, business skills and accountability. AI use does not transfer responsibility for supervising others, providing authoritative advice, setting standards, allocating resources, defining strategy or authorising action.

This is not an AI proficiency scale. It does not imply that senior practitioners must use more sophisticated AI or that using AI moves someone to a higher SFIA level. The level continues to depend on the responsibility exercised in the work.

Illustration – Project management PRMG: AI may support planning and monitoring at several levels, but the responsibility exercised remains different. Managing delivery within agreed tolerances, directing complex projects and setting organisational strategy continue to represent different SFIA responsibility levels.

Benefits of the commentary

The common structure helps readers:

• consider practical opportunities for AI assistance in the context of real work
• distinguish changes to tasks from changes to skills or responsibility levels
• recognise the limitations of generated outputs as evidence of competence
• compare possible effects consistently across skills
• support informed discussions about work design, development, assessment and workforce change.

How employers may use it

Employers can use the commentary as a starting point to:

• review how AI may alter tasks within jobs while identifying responsibilities that must be retained
• update role descriptions, operating procedures and accountability arrangements
• identify where assurance, review, escalation, approval or authorisation is needed
• strengthen recruitment and assessment methods that rely on work products as evidence
• target learning and support planning across the range of skills needed in the workforce.

The commentary should be interpreted for the employer’s technologies, risks, policies, regulatory obligations and ways of working. It is not a ready-made AI policy, control framework, job description or workforce plan.

How individuals may use it

Individuals can use the commentary to:

• explore where AI could support their current work
• identify what they still need to understand, verify, decide and take responsibility for
• discuss changing expectations with managers, clients or professional assessors
• plan development and build stronger evidence through their decisions, checks, actions and outcomes
• consider progression by focusing on increased responsibility rather than tool use alone.

How to read the commentary

Start with a skill relevant to your work and read all four sections together. Compare the potential uses with the responsibilities that remain. Where a skill is defined at several levels, compare the commentary with the corresponding SFIA level descriptions.

For jobs involving several skills, examine the relevant commentaries together to identify common responsibilities, differences and interactions. Use the commentary as a basis for discussion and analysis. Local decisions about AI use, work design, assessment and accountability still require organisational and professional judgement.