Principles for using AI
Using AI does not reduce your responsibility for the decisions made by your service.
Understand the system
Understand:
- what the system does
- how it produces outputs, where possible
- its limitations
Check that outputs are:
- accurate enough for your purpose
- reliable
- appropriate for the people affected
Consider whether the system could introduce bias or create risks to privacy, security or equality.
Do not assume AI outputs are correct. Review and validate them before using them.
Keep people responsible
People, not AI systems, remain responsible for decisions.
Where AI could affect individuals or public services, make sure there is meaningful human review.
People responsible for the service should be able to:
- understand AI outputs
- challenge them
- override them where necessary
Do not rely on AI where decisions cannot be explained or properly reviewed.
Be open about using AI
People should understand when AI is being used and how it may affect them.
Be able to explain:
- why AI is being used
- what it is used for
- how outputs contribute to decisions
- what safeguards are in place
Being open helps build trust and accountability.
Meet legal and ethical requirements
Make sure your use of AI complies with relevant legal, regulatory and organisational requirements.
For example:
- data protection law
- equality duties
- human rights obligations
- information security requirements
- procurement requirements
- your organisation's governance processes
Consider how AI could affect different groups and take steps to make outcomes fair and proportionate.