Skip to main content
    Skip to main contentSkip to navigationSkip to footer
    Artificial Intelligence
    (Kontrafaktische Erklärung)

    Counterfactual Explanation

    Updated: 2/10/2026

    Explanation method that shows what minimal input change would have led to a different model outcome.

    Quick Summary

    Counterfactual explanations show the smallest input change for a different outcome – the most intuitive and GDPR-compliant XAI method.

    Explanation

    "Your loan application was rejected. If you had €5,000 more annual income, it would have been approved." Counterfactuals are intuitive and actionable.

    Marketing Relevance

    Counterfactual explanations are particularly GDPR-relevant and the most human-friendly form of AI explanation.

    Common Pitfalls

    Multiple counterfactuals possible – which to show? Unrealistic suggestions ("Become 20 years younger"). Instability with small changes.

    Origin & History

    Wachter et al. formalized counterfactual explanations in 2017 in the GDPR context. DiCE (Microsoft, 2020) made generating diverse counterfactuals practical. The method gained further importance through the EU AI Act.

    Comparisons & Differences

    Counterfactual Explanation vs. SHAP

    SHAP shows feature contributions to current prediction; counterfactuals show what would need to change for a different outcome.

    Counterfactual Explanation vs. Feature Importance

    Feature importance ranks features by influence; counterfactuals provide concrete, actionable change suggestions.

    Related Services

    Related Terms

    👋Questions? Chat with us!