Page 62 - AI Ready – Analysis Towards a Standardized Readiness Framework

AI Ready – Analysis Towards a Standardized Readiness Framework

discipline utilizes sensors or roadside units. Thus, determining the interaction point via the actors
would point us to specific requirements of deployment scale and support needed.

Integral/add-on controls point to the nature of the AI integration in the actors needed in the
scenarios. External, overlay or add-on deployment of sensors or other types of hardware may
have specific requirements in terms of cost and regulations. The gradients of thresholds for
the limits of application of the scenarios, point to compliance and validation requirements.
For example, in accessibility scenarios, we aim to provide sign language to audio and video
translation, different set of thresholds may be applied based on dialects of sign languages.
Further study is needed, on a use case scenario basis to derive the thresholds and corresponding
validation requirements for each use case.
Finally, experiments for validation of the scenarios determine the tools, sandbox setups,
validation test suites and testing infrastructure needed in each scenario. Feedback from the real-
world deployments is more reliable and authentic. However, in rarely reproducible scenarios
such as massive natural disasters, it is important to use simulated sandboxes to validate and
prepare for such failure scenarios.

To conclude, a scenario-based study of the type of actors, type of controls, thresholds for
validation and compliance, and experiments for validation leads us to the relations between
specific parameters for the AI readiness factors and corresponding benefits which can be
obtained.

55