From pre-AI times to today’s aspirations

Long before the current hype of GenAI, data professionals were already deeply immersed in the world of machine learning (ML), skillfully injecting human feedback into ML-driven processes.

This approach was well aligned with complex business rules and the necessary explicit or implicit approvals for implementing actual changes in data management. These approvals were overseen by humans, with varied feedback gathered from data stewards and professionals. The purpose was to combine the rising power of ML analytics with factual, human-validated input, allowing a continuous circle of learning and the development of ML. Activities such as data labeling, data categorization, and the creation of robust filters for identifying poor data quality across a variety of world languages significantly enhanced our understanding of the impact of language phonetics and syntax on datasets.

This integration was not about who could do better –...