Model learning & next decisions

The AI partner integrates experimental evidence into its evaluation or learning workflow and selects the model and retraining environment for the collaboration.

INPUT

Evidence package and the partner’s model evaluation plan.

WHAT HAPPENS

Review data suitability, update the model where appropriate and compare with the prior version.

OUTPUT

Evaluation results and an updated prediction set.

↳ Review whether the evidence reduced the targeted uncertainty.

Close the loop by choosing the next action based on the evaluation, including when to request another experiment.

INPUT

Updated predictions, evaluation results and remaining evidence gaps.

WHAT HAPPENS

Review progress against the original objective and prioritize the next hypothesis.

OUTPUT

A documented decision and the next evidence request, if needed.

↳ Feed the next question back into hypothesis generation.