How do we validate the correctness of the industrial AI models?

The Industrial Analytics Task Group members posed and answered this question in the Industrial IoT Analytics Framework (IIAF). What do you think?
IIAF|352: Industrial analytics must satisfy a higher level of accuracy in its analytic results. Any system that interprets and acts on the results must have safeguards against undesirable and unintended physical consequence.
IIAF|414: Industry subject matter experts have a good understanding of the context and condition of the process and assets, and can interpret and validate the readings and recommend cleansing filters. It is this combination of data science and subject matter expertise that produces the best results.
IIAF|417: Ensure analytics results when applied in the context of the application and environment will not harm people or property. This function should be independent from the core analytics processing and act as a governor.