Practice position
AI assists. People decide.
Use approved systems, minimise information, check important output and keep accountable professional judgement human-led.
Ownership must be visible
For each use, define who understands the task, approves the information, checks the output and signs off the final work. ‘The AI said’ is not an accountable rationale.
Review should be meaningful. A hurried glance at fluent text is not effective human oversight, particularly when the reviewer lacks time, evidence or authority to challenge it.
Match oversight to consequence
A spelling suggestion and a service-eligibility decision do not require the same controls. Consequential decisions affecting care, safety, liberty, employment or access to services should remain human-led and use established professional processes.
Keep an adequate record
For significant uses, record the purpose, system, information, AI contribution, verification, changes and responsible person. The record should support learning and transparency without creating unnecessary personal data.
Practical checklist
- Name the responsible person
- Define the human-review method
- Keep consequential decisions human-led
- Record significant AI assistance
- Review incidents and model changes
Written by Tony D’Agostino / TD Consultancy. Reviewed . This page provides general professional education, not legal or clinical advice.