Purpose: To ensure that AI systems support rather than replace, human decision making.
| Organisational / Technical | Measure |
|---|---|
| A. Human control by design | |
| TECH | The AI System is designed as a decision-support tool rather than an autonomous decision-maker |
| BOTH | Clear points of human intervention are defined for high-risk or sensitive uses of the system |
| ORG | The organisation has specific which decisions must involve human judgment |
| TECH | Automated actions can be paused, overridden or reversed by humans |
| ORG | Human oversight is not static but involves active review, validation or challenge of system outputs |
| ORG | Oversight responsibilities are clearly defined and embedded into every day work flows |
| B. User autonomy and informed interaction | |
| ORG | Checks are in place to ensure individuals using the tool retain meaningful choices and are not coerced into accepting AI outputs |
| BOTH | Users show they understand when they are interacting with AI and when outputs are AI generated |
| ORG | Organisation offers users opportunities to question, challenge or deviate from AI recommendations |
Source: AIOLIA deliverable 3.1