Question 1

Which article of the EU AI Act contains specific requirements for data quality and data management?

Accepted Answer

Article 10 of the EU AI Act contains extensive requirements for data governance and data quality. It requires providers of high-risk AI systems to manage training, validation, and test data according to quality criteria.

Question 2

What is the main requirement for training data according to Article 10?

Accepted Answer

Article 10 requires that training, validation, and test data be 'relevant, representative, error-free, and complete.' These quality criteria are essential to prevent bias and develop reliable AI systems.

Question 3

Why is 'representativeness' of training data crucial?

Accepted Answer

Representative data means that all relevant user groups, scenarios, and use cases are adequately represented in the training data. This prevents the AI system from performing worse for certain groups (bias) or situations.

Question 4

What must happen if relevant training data contains special categories of personal data?

Accepted Answer

Article 10(5) allows special categories of personal data (such as health data, ethnic origin) to be processed for bias monitoring and correction, provided appropriate safeguards such as technical limitations, pseudonymization, or anonymization are applied.

Question 5

What is 'data lineage' in the context of AI governance?

Accepted Answer

Data lineage is the ability to trace the complete history of data: where it comes from, which transformations were applied, how it was used. This is crucial for transparency and accountability in AI systems.

Question 6

What responsibility does a provider have for data used to retrain an AI system?

Accepted Answer

Article 10 also applies to data used to retrain or update AI systems. Providers must continue to ensure relevant, representative, error-free, and complete data with each iteration.

Question 7

What is the difference between 'data quality' and 'data fitness for purpose'?

Accepted Answer

Data quality measures objective properties such as correctness and completeness. Fitness for purpose assesses whether the data is actually suitable for the specific purpose of the AI system - for example, whether medical training data is representative of the target population.

Question 8

What documentation about data governance must be included in the technical documentation according to Article 11?

Accepted Answer

The technical documentation must contain extensive information about data governance: data sources and selection criteria, preprocessing and cleaning, bias detection and mitigation measures, quality controls and validation methods.

Data Governance Quiz