Data Governance And Ai

Published on: September 02, 2024

Data Governance is more important for AI than Data Quality! This was the surprising conclusion -- for me at least -- from my MBA Thesis [https://lnkd.in/ed5wdz76] research on organisational readiness for AI. It's been almost 3 years now, and at the time of writing AI was mainly ML, Data Science, and advanced linear regressions -- LLM's existed, but more as science fair material. Obviously this all changed with the launch of ChatGPT, and everything is (Gen)AI and everybody is an AI expert nowadays, however the conclusion still stands.

The conclusion surprised me because for most of my career Data Quality was king, and Data Governance was often an afterthought, it was helpful, but not critical. And during programmes and projects it was often allocated a small slice compared to the enormous chunks of effort which went into getting Data Quality up to a workable level.

Does this mean you need to invest more in your Data Governance? Yes, but that was true anyway ;-) Does this mean you need to take those resources from Data Quality? No, you do definitely need Data Quality as well. Do you need to weigh the investment and required deliverables for Data Governance more heavily when you look at AI initiatives compared to looking at regular processes embeddings? Yes, definitely.

And why does Data Governance matter more than Data Quality? In a similar same way that discussions around technology move more towards organisational readiness, culture changes, upskilling, and employee acceptance. It's great if your data is high quality, but data governance will enable you to actually find it, and get you access to it, it will make your users capable and literate towards data, so they can actually use it for their AI usage.

And a nice bonus is that improved Data Governance also increases Data Quality 🎉