Working with AI WWAI
Using AI responsibly and critically within agreed work, risk, decision-making and accountability boundaries.
THIS IS A PROTOTYPE
Let us know: Is this a useful resource? What else would be useful?
Guidance notes
Working with AI progresses from following guidance to shaping organisational practice.
It involves:
- selecting appropriate AI tools and uses, considering the task, risk, cost, quality and expected outcome
- working within approved policies, tools, data boundaries and delegation or decision limits, including work delegated to autonomous AI
- providing appropriate context, instructions and source material, including through prompts where applicable
- checking outputs against evidence, standards, tests, sources and stakeholder needs
- deciding when to accept, question, modify or reject AI outputs or actions and when to pause or escalate work, according to risk, reversibility, impact and delegated authority
- protecting sensitive, personal, confidential and regulated information
- identifying uncertainty, limitations, bias, assumptions and potential misuse
- maintaining required traceability of AI-related inputs, outputs, significant decisions and their rationale
- exercising professional judgement and retaining accountability for AI-supported work, decisions and recommendations
- evaluating and improving AI-related practices, safeguards and measures.
Working with AI combines digital confidence, critical thinking, domain knowledge, professional judgement and responsible conduct. It does not replace role-specific expertise or accountability for outcomes.
As responsibility increases, the focus moves from safe use in assigned tasks to taking responsibility for one’s own AI-enabled work, enabling others, ensuring effective practice within an area of responsibility, shaping organisational practice and setting the organisational approach.
Level 1 - Follow
- Uses approved AI for simple tasks when instructed and within defined boundaries.
- Seeks help and does not proceed when outputs are unclear, incomplete or unexpected.
Level 2 - Assist
- Uses approved AI on assigned tasks within established methods and boundaries, exercising limited discretion.
- Checks outputs against supplied sources, examples or instructions before use.
- Protects sensitive information. Questions unexpected outputs and escalates uncertainty, errors or policy concerns.
Level 3 - Apply
- Takes responsibility for their own use of AI within agreed methods, decision limits and accountability boundaries.
- Validates outputs against evidence, standards and stakeholder needs. Questions, modifies or rejects outputs that do not withstand validation and escalates concerns beyond their authority.
- Records significant assumptions, AI contributions and decision rationale where required. Considers whether AI use is proportionate to the task, risk and expected outcome.
Level 4 - Enable
- Enables and reviews responsible use of AI within a team or defined area of work.
- Guides team members in choosing and applying approved AI tools, recognising their limitations and using suitable safeguards. Supports adoption of agreed practices in team processes.
- Reviews team use of AI for quality, traceability, risk and proportionality.
- Helps team members exercise judgement, use agreed review and escalation routes and challenge AI outputs or actions when concerns arise.
Level 5 - Ensure, advise
- Takes accountability for establishing and maintaining effective and responsible use of AI within their area of responsibility.
- Establishes appropriate decision rights, review checkpoints, intervention thresholds and expectations for evidence, verification, explainability and accountability.
- Provides authoritative guidance where required and reviews AI-related working practices for quality, risk, effort, outcomes and opportunities for improvement.
Level 6 - Initiate, influence
- Leads the development and implementation of organisational practices, decision rights and accountability arrangements for AI-enabled work.
- Ensures these are aligned with governance, risk, data, privacy, security, ethics and assurance requirements. Influences how AI-supported work moves through review, intervention and decision routes across organisational functions.
- Takes accountability for the effective implementation and operation of these arrangements across their area of responsibility.
Level 7 - Set strategy, inspire, mobilise
- Sets the organisational approach for working with AI across strategy, operating model, risk, decision authority and accountability.
- Engages executive stakeholders on the implications for productivity, quality, trust and organisational responsibility.
- Holds accountability at the highest level for the organisation’s approach.