In watsonx.governance explainability, what is a contrastive explanation?
Select an answer to reveal the explanation.
Short Explanation and Infographic
Think of it this way: in real-world AI governance, an explanation that identifies the minimal changes to input features that would flip the model's prediction to a different outcome, also called a counterfactual explanation is exactly what teams reach for when they need to handle this scenario. Contrastive explanations, also called counterfactual explanations, identify the smallest perturbation to input features that would change the model's prediction to a different class. On the exam, remember that this falls squarely under the 4.0 Configure Evaluation and Monitoring domain.
Full explanation below image
Full Explanation
Contrastive explanations, also called counterfactual explanations, identify the smallest perturbation to input features that would change the model's prediction to a different class. This supports actionable recourse — for example, informing an applicant what changes to income or credit history would result in loan approval. watsonx.governance can surface contrastive explanations alongside LIME and SHAP outputs as part of a comprehensive explainability configuration. The correct answer, "An explanation that identifies the minimal changes to input features that would flip the model's prediction to a different outcome, also called a counterfactual explanation", directly addresses the scenario described because it aligns with the specific governance requirement in question. The incorrect options ("A comparison of SHAP feature importance values between two different models deployed on the same dataset", "A visualization contrasting training-time feature importance against runtime feature importance to detect explanation drift over time", "An explanation generated by running both LIME and SHAP and flagging features where the two algorithms produce conflicting attributions") may seem plausible but do not satisfy the core requirement. Understanding the distinction between these concepts is critical for IBM watsonx.governance implementations and is frequently tested in the 4.0 Configure Evaluation and Monitoring section of the certification exam.