Develop and manage AI systems
Using AI successfully depends on more than choosing the right technology.
Use good quality data
AI systems are more likely to produce reliable outputs when they use data that is:
- accurate
- complete
- relevant
- well managed
Understand how data is collected, managed and used within your service.
Protect information
Protect data from unauthorised access, misuse and loss.
Take extra care when using:
- personal data
- special category data
- externally provided AI tools and services
Follow your organisation's information security policies.
Involve the right people
Bring the right people into your project as early as possible.
This may include:
- policy teams
- service teams
- legal advisers
- data specialists
- information security specialists
- procurement professionals
- governance and assurance teams
Early collaboration helps identify risks, improve decisions and avoid unnecessary rework.
Build capability
Make sure people using AI understand:
- what the system can and cannot do
- its limitations and risks
- their legal and governance responsibilities
- how to review and challenge outputs
Provide training and support where needed.
Manage the full lifecycle
Plan for the whole lifecycle of the system, including:
- design
- testing
- deployment
- monitoring
- maintenance
- review
- updates
- retirement
Review systems regularly to make sure they remain effective, lawful and appropriate.