Step 4: Model Prototyping, Tuning and Assessment

Select a decision point below to access further resources, tools, and guidance.

What are the accuracy / success metrics / thresholds and who determines these?
Who creates the model?
Have all risks been considered?

This Framework does not provide an exhaustive list of all decision points that are to be addressed in the process of an AI/ML project. Rather, it focuses on providing tools and guidance around the decision points which provide particularly strong opportunities for processes of co-creation. Below is a list of additional questions relevant to “Model prototyping, tuning, assessment” for consideration:

  • Can the model's outcomes/predictions be explained and validated?

  • Have potential error, bias, discrimination been considered?

  • Have contingency plans / safeguards been identified to prevent harmful consequences for stakeholders?

  • What is the appropriate extent of human oversight?