Developers and employers should explain how the AI System functions and whether certain forms of transparency may exacerbate security risks.
In plain English this means that end users should understand:
- how the AI System makes decisions (inc. the algorithmic logic) and how outputs may be used
- the relevant capacities of the AI System to detect and address anomalies and unexpected performance.
- how to enforce their legal rights and what would amount to legal/illegal uses of the system
|