AI Systems (UK Definition)

The UK government defines AI systems by two key characteristics that create unique regulatory challenges:

Adaptivity: AI systems are 'trained' and operate by inferring patterns and connections in data that are often not easily discernible to humans. Through training, they develop the ability to perform new forms of inference not directly envisioned by their programmers.

Autonomy: Some AI systems can make decisions without the express intent or ongoing control of a human.

This functional definition focuses on capabilities rather than specific technologies, making it future-proof against technological evolution. The combination of adaptivity and autonomy creates challenges in:

• Explaining, predicting, or controlling system outputs
• Understanding the underlying logic of decisions
• Allocating responsibility for system operation and outcomes

Examples include:

• Natural language processing chatbots that generate responses based on training data
• Automated healthcare triage systems that recommend treatments
• Text-to-image generators that create content from prompts

This definition underpins the UK's context-specific regulatory approach, focusing on AI use and outcomes rather than blanket technology restrictions.