Data governance in the age of generative AI

This article discusses the importance of robust data governance strategies for successful implementation of generative AI applications in enterprises. It covers the key considerations and approaches for governing data used by large language models (LLMs) to ensure accuracy, relevance, privacy, and security of responses.

Specifically, the article covers:

• The need for a comprehensive data governance approach in generative AI pipelines, including managing structured and unstructured enterprise data sources
• Data governance steps in data pipelines like data cataloging, implementing data privacy and quality controls, and governing vector stores
• Integrating data governance controls into the user request-response workflows of generative AI applications, including access control, compliance checks, and data redaction
• Additional governance requirements for prompt engineering, model fine-tuning, and training foundation models
• Conclusion on the critical role of data governance in enabling responsible and trustworthy generative AI applications