Training data
AIGP glossary · Last reviewed: · By Victor Humenhuk (AIGP certified)
Training data - The dataset the model learns from; must be representative, fair and compliant.
In the AIGP body of knowledge, Training data comes up under Module 8: AI Governance Vocabulary.
Training data in context
- Prompted by the Irish DPA, the European Data Protection Board harmonised how the GDPR treats AI models in three answers: when a model is anonymous, when legitimate interest is a valid basis (the ==Interest → Necessity → Balance== test), and how to handle unlawfully processed training data. (The EDPB Opinion on AI Models (2024))
- Key anchors: ==Bartz v Anthropic== on training data, and ==Thaler v Vidal== confirming only humans can be named inventors on a US patent. (Intellectual Property and AI)
- It brings five challenges (transparency limits, data sources, ownership, risky purposes, increased requirements) but four opportunities, chiefly ==sourcing your own training data for better transparency==. (Deploying a proprietary model)
Where Training data is covered in the AIGP study notes
Related terms
- Validation data
- Input data
- Ground truth
- Corpus
- Data quality
- Data provenance
- Data drift
- Synthetic data
- Variables
- Preprocessing
Test yourself on Training data
Recognising a definition is not the same as applying it in an exam scenario. Work through the free AIGP practice questions, or read the full AIGP study guide - every study note is free.