Purpose: To ensure that AI systems and their outputs are understandable, traceable and open to scrutiny but those who use them, are affected by them or are responsible for their governance.
| Organisational / Technical | Measure |
|---|---|
| A. Openness about tool use and purpose | |
| ORG | There is clear guidance on the existence, purpose and intended use of the AI tool |
| ORG | Information about why the tool is used and how it may affect individuals is made available in accessible non-technical language |
| ORG | Responsibility for the tool/system governance and oversight is clearly communicated |
| ORG | Obligations are reviewed whenever the tools scope or use changes |
| B. Accessibility of information | |
| BOTH | Information about the tools functioning, inputs, outputs and safeguards is accessible to all relevant users |
| ORG | Communication channels and formats are adapted to different roles, languages, digital literacy and accessibly needs |
| ORG | Users know where to find information and whom to contact with concerns about the tool |
| BOTH | Transparency is communicated clearly in non-technical language and without intent to overwhelm the reader |
| C. Explainability of outputs and decisions | |
| TECH | The system provides explanation of outputs at a level appropriate to their use context |
| BOTH | Explanations enable users to understand key factors, limitations and uncertainty associated with outputs |
| ORG | Explanations support meaningful review, contestation or justification of AI-supported decisions |
Source: AIOLIA deliverable 3.1