More Training Data vs Citizen Privacy
Apply data minimisation, anonymisation, and federated learning by design to improve AI models while meeting GDPR purpose-limitation and privacy-by-design obligations.
CyberTRIZ analysis · SmartCity contradiction C13-SC013 · one of 8,235 worked contradictions published by CyberTRIZ.AI
Regulations
Business Context
Artificial Intelligence systems improve as they learn from larger and more diverse datasets. Municipal governments possess valuable information regarding transportation, utilities, public services, environmental conditions, and citizen interactions. However, collecting additional information to improve AI performance may increase privacy risks and regulatory obligations. Municipalities must improve AI capabilities while protecting personal information.
Smart CityTRIZ Resolution
Rather than collecting all available data, municipalities should adopt privacy-by-design principles, anonymization, federated learning, synthetic data generation, and data minimization strategies that improve AI performance without exposing sensitive citizen information.
Applicable TRIZ Principles
Principle 2 – Taking Out removes personally identifiable information before model training.
Principle 26 – Copying uses synthetic or anonymized data whenever possible.
Principle 30 – Flexible Shells and Thin Films protects sensitive information through privacy-preserving techniques.
Expected Outcome
Stronger privacy protection
Better AI performance
Improved regulatory compliance
Increased citizen trust
Decision Indicators
Early indicators that AI data practices require improvement include:
Privacy concerns increase.
Sensitive information appears in training datasets.
Data governance audits identify unnecessary collection.
Regulatory compliance issues become more frequent.
Public confidence in AI initiatives declines.
Monitoring these indicators supports responsible AI development.