By the end of 2022, ChatGPT 3.5 unleashed a new AI era, and it became obvious to everyone that generative AI (GenAI) was a huge leap forward in technological innovation. The seamless generation of meaningful text, and subsequently pictures, sound, and videos, opened perspectives so deep that no one could completely describe what this change meant — least of all for enterprise data management.

For Chief Data Officers (CDOs), the reality is clear, and the stakes are urgent. If they can address two critical challenges and seize one significant opportunity, they will be able to push forward with strategic data agendas and accelerate AI initiatives for the enterprise.

2 critical challenges for traditional data management

CDOs need to be aware of two critical challenges in enterprise data management because if they are not addressed, the data teams inside enterprises will gradually lose relevance and competitive edge.

1. Data for AI is the opposite of data for analytics

The nature of data for analytics has been refined over several decades. The first experiments began with data warehouse architectures, evolved into data lakes and data lake houses, and culminated in completely decentralized data architectures based on data products and data contracts. However, throughout all these stages, the targeted outcome of all these activities was to refine data for analytics.

Data for analytics is organized in tables. It’s historical data that has been collected and aggregated to serve analytical use cases. Furthermore, it’s data that is semantically implicit, in the sense that a dashboard is intended for human interpretation.

In other words, we expect the recipient of the dashboard to be able to interpret what the dashboard represents, and if it contains errors, the human recipient can fix those errors. This way of working does not match the changing reality we are facing in the current era of AI.

It is absolutely essential that CDOs understand that data for AI is the opposite of data for analytics. Data for AI are chunks of unstructured data (text, events, and pictures) that are not necessarily organized in tables, as required in data for analytics. It also needs to be as close to real time as possible, unlike data for analytics that is historical by intention. And unlike analytics, which depend on aggregated summaries, AI needs massive amounts of raw data to perform well.

As explored in a recent webinar with CDO Magazine (at 16:40), the semantics around AI data cannot in any way be implicit; it needs to be explicit, which puts more pressure on metadata management. Finally, errors in data for analytics are adjustable by humans, as dashboard errors can be fixed quite easily.

However, errors in data for AI are of dramatic proportions. Hallucinations are mathematically impossible to avoid completely and may cause a plethora of problems, such as financial or reputational damage. The economic potential of AI-ready data dwarfs traditional analytics — but so does the cost of getting it wrong.

2. We all have to unlearn and learn — and we are all still needed

CDOs have spent decades building organizations that delivered data for analytics. That organizational reality now belongs in the past as AI is impacting all professional data roles. Data analysts and engineers are no longer standalone roles because AI makes dashboard creation table stakes. Someone in marketing or sales can build dashboards too. Data scientists face similar challenges.

However, outside of data, professional roles are changing across the board. Therefore, it is important for CDOs to not react spontaneously by shrinking their teams or thinking that certain professional roles belong in the past. It’s the way tasks are carried out that is changing. What is happening is that professional roles are converging.

Right now, no one has the answers for the best organizational setup to deliver data for AI. But in order to facilitate a fruitful conversion, everyone has to unlearn and learn — and all are needed. CDOs need to carefully manage the friction this transition creates. Teams that feel threatened won’t collaborate, and that internal conflict will slow the very AI initiatives the organization is counting on.

CDOs have an unexpected, strategic lever to address these two challenges and deliver on data for AI.

Data Intelligence is a big opportunity CDOs can’t miss

Consider the core premise of veteran data leader Laura Madsen’s landmark data governance book “Disrupting Data Governance”:

“Data governance is broken. There is no way to make incremental changes to fix it… The parochial idea of having everyone slow down long enough so we can define data and control its usage is nuts.”

Madsen’s analysis was correct for its time, but AI changes the equation. Today, the data analyst, data engineer, and data scientist see their professional roles challenged by AI. However, data governance teams face the opposite reality.

Data governance has always lacked budget, employees, and technological support. But with AI, we are facing a new era for data governance. One where we can finally fix data governance, because AI can take over many of the tasks no one had time to carry out before. And that is Data Intelligence: data governance performed in the era of AI, with the means of AI, for AI.

Data Intelligence proposes not only a new mindset toward data, but the technological means to work with governed data in a way that is incomparably better than what was possible in the past. This shift in reality is very concrete: tagging datasets, assessing data sensitivity and quality, monitoring data through observability, making data searchable and accessible.

These are all the things that good data governance requires. And with Data Intelligence, they can finally be unleashed.  

About the Author:

Ole Olesen-Bagneux is a globally recognized thought leader in metadata management and enterprise data architecture. As VP, Chief Evangelist at Actian, he drives industry awareness and adoption of modern approaches to data intelligence, drawing on his extensive expertise in data management, metadata, data catalogs, and decentralized architectures.

An accomplished author, Olesen-Bagneux has written The Enterprise Data Catalog (O’Reilly, 2023) and Fundamentals of Metadata Management (O’Reilly, 2025). He is currently writing the second edition of The Enterprise Data Catalog. With a PhD in Library and Information Science from the University of Copenhagen, his unique perspective bridges traditional information science with modern data management.

Before joining Actian, Olesen-Bagneux served as Chief Evangelist at Zeenea, where he played a key role in shaping and communicating the company’s technology vision. His industry experience includes leadership roles in enterprise architecture and data strategy at major pharmaceutical companies like Novo Nordisk.