Sustainability
Balance environmental impact with public value when choosing AI tools for government services.
Why sustainability matters
AI speeds up conservation work and makes public services more accessible. But these benefits must not damage the environment.
The UK Government expects you to assess environmental impact before making AI decisions and choose less energy-intensive options when possible.
Government requirements
UK Government AI Playbook
Build “safe, secure and robust AI solutions” that remain sustainable throughout their life cycle. AI training and deployment use significant energy and resources.
For more information, see Artificial Intelligence Playbook for the UK Government
Sustainable-by-Design principles
Principle 5: “Take a greener approach to AI and software tools” requires you to:
- Choose the right model size - Pick the smallest or pre-trained model that meets your needs
- Use AI only when necessary - Apply AI tools only when they are the best option and measure their energy use
- Adopt green practices - Use shared tools, low-carbon data centres, and remove unused software
For more information, see Greener Service Principles
How to make sustainable choices
- Use smaller, shared or open models when possible
- Pick models trained in low-carbon regions
- Reuse existing models rather than training new ones
- Schedule heavy workloads for times when the electricity grid uses more renewable energy
- Cache results to avoid repeated processing
- Remove unused resources regularly
- Monitor and measure energy consumption to identify improvement opportunities
- Choose cloud providers with renewable energy commitments
- Document sustainability decisions to help other teams make similar choices
- Share trained models across government departments to avoid duplicate work
