Fair

Fairness means AI must treat people equally and without bias. This includes making sure that the data used to train AI does not discriminate against certain groups when we are using AI. Fair AI helps us to build trust and avoids causing harm by being inclusive and respectful to everyone.

How will we use this principle?

• We will ensure that our datasets are inclusive and representative by implementing string data governance processes and correcting imbalances that could lead to unfair outcomes.

• We will promote equity in access to AI systems by removing reasonable technical, economic and language barriers that exclude users.

• We will provide clear feedback and grievance mechanisms so people can report perceived unfairness or harm from AI decisions, so we can take corrective actions.

• We will assess fairness at every decision point by explicitly asking “Who benefits and who might be harmed?” and will make this reflection a routine part of AI design and deployment, adoption, and governance.

• We will teach people to recognise and challenge bias by offering materials that help people to spot and address bias in data, algorithms, and outputs.