Higher AI Accuracy vs Model Transparency
Embed explainability requirements and decision-traceability logs into AI design from inception to satisfy EU AI Act transparency and auditability obligations.
CyberTRIZ analysis · SmartCity contradiction C13-SC012 · one of 8,235 worked contradictions published by CyberTRIZ.AI
Regulations
Business Context
Advanced Artificial Intelligence models often achieve higher prediction accuracy by using increasingly complex algorithms and large-scale data processing. However, highly sophisticated models may become difficult for municipal officials, regulators, and citizens to understand or explain. Municipalities must improve AI performance while maintaining transparency and accountability in public decision-making.
Smart CityTRIZ Resolution
Rather than prioritizing predictive accuracy alone, municipalities should incorporate explainable AI techniques, model documentation, decision traceability, and transparent governance practices that allow users to understand how recommendations are generated.
Applicable TRIZ Principles
Principle 26 – Copying presents simplified explanations of complex analytical processes.
Principle 13 – The Other Way Around designs AI systems around explainability from the beginning.
Principle 23 – Feedback continuously evaluates the clarity and reliability of AI explanations.
Expected Outcome
Greater AI transparency
Improved regulatory compliance
Increased public trust
Better decision quality
Decision Indicators
Early indicators that AI transparency requires improvement include:
Decision logic cannot be adequately explained.
Auditors require extensive manual analysis.
Citizens question AI recommendations.
Regulatory reviews identify insufficient documentation.
Operational staff struggle to interpret AI outputs.
Monitoring these indicators supports trustworthy AI deployment.