Artificial intelligence is gradually becoming integrated into companies’ tools, processes, and decision-making. But its adoption is no longer simply a matter of choosing a technology or launching a few isolated use cases. As AI impacts business functions, data, legal teams, human resources, marketing, and customer support, one question becomes central: Who should decide its role within the organization? 

For Adia Damso, the answer cannot come solely from technical or data teams. AI governance must be approached from a cross-functional perspective, with representatives capable of conveying the actual needs of each department. A tool that works well for one team may be unsuitable for another. And without a common framework, disparities in access, training, and usage can quickly lead to a multi-tiered AI system within the company. 

The challenge, therefore, is not only to structure AI at the strategic level, but also to ensure that this strategy remains connected to the front lines. This means giving employees the time to test the tools, identify useful use cases, document what works, and help define the rules governing the use of AI. 

When a company seeks to align AI with its overall strategy, who should be involved in the decision-making process? 

In my view, we need to involve representatives from the company’s various departments. AI now affects so many levels of the organization that it would be difficult to determine its role within the company based solely on a technical perspective. 

Therefore, it is important to have representatives from the tech and data teams, as well as from legal, production, customer support, human resources, product, marketing, and other departments, depending on the company’s structure. 

The goal is to take a broad enough view to understand the needs of each department and identify relevant use cases. The way AI is used will differ between a technical team, a support team, and a legal team. And a tool that works very well for one department may be completely unsuitable for another. 

In my view, AI governance must therefore be approached from a cross-functional perspective, with representation from various departments and tailored to the realities of each one. 

How can we ensure that decisions related to AI aren't left solely in the hands of technical or data teams? 

Precisely by involving the other departments from the very beginning. 

I think we need to have representatives and testers on the various teams. They can experiment with the tools, identify their own use cases, and, most importantly, provide feedback on what works and what doesn't in their day-to-day work. 

This also helps avoid favoring one department over another. You might have a tool that works particularly well for technical teams, but that doesn't meet the needs of support or administrative teams at all. 

The idea isn't necessarily to have a single tool for the entire company. Some departments may have very specific needs and require complementary solutions. But these choices must be made based on actual usage, not just from a technical standpoint. 

Why is data governance also a matter of corporate culture? 

Because data governance also depends on the behavior of the people who work at the company. 

If employees do not feel invested in the company’s operations, its development, or the consequences of their actions, it is difficult to ensure effective data governance. They may be less careful about how they use, share, or protect company information. 

Conversely, in a company where employees feel engaged, understand the issues at stake, and are eager to contribute to the organization’s growth, they will naturally be more mindful of security and the proper use of data. 

In my view, governance cannot, therefore, rely solely on rules or tools. We must also foster a culture of accountability. 

What risks arise when AI decisions are made without sufficient diversity in terms of profiles, expertise, or business perspectives? 

The first risk is creating a two-tier AI system within the company. 

If certain categories of employees have access to the best tools and are trained in how to use them, while others have to make do on their own with free or less suitable tools, disparities quickly arise. 

And this can also become a data security and governance issue. An employee who does not have access to a tool approved by their company may be tempted to use an external solution or a free version to meet their needs, potentially entering work-related data that should not be entered there. 

There is, therefore, a twofold challenge: to ensure that everyone has access to tools tailored to their profession, while providing a clear framework for their use. 

The diversity of the roles involved in AI decisions makes it possible to identify these different needs and avoid approaching AI solely from the perspective of a single category of employees. 

How can we further involve the relevant business units, users, and teams in defining AI use cases? 

I think we need to give them a meaningful role in the pilot program. 

Having representatives and testers in the various departments allows us to gather use cases directly from the field. They are the ones who can say, “The AI really saved me time on this task,” or, conversely, “In this specific case, it’s not relevant.” 

It’s also important because we often discover the best use cases through experimentation. We may start with an initial idea of what AI will offer, only to find that its true value lies elsewhere. 

We must therefore allow ample room for experimentation and create a system in which employees can test ideas, document their feedback, and share what they’ve learned with the rest of the company. 

How can you tell if an organization treats AI as a strategic capability rather than as a series of isolated projects? 

In my view, you can tell when AI is truly part of the company's strategic goals and becomes a cross-functional priority. 

This isn't just a project led by a team or a few people identified as "AI experts." AI-related goals must apply to the entire organization, with input from various departments. 

This also means that the company empowers employees to get involved: by giving them time to learn, test the tools, identify use cases, and share their feedback. 

Once AI becomes a skill that all employees can develop and apply in their work—rather than a series of one-off projects—that’s when we can truly start talking about strategy. 

What advice would you give to a company that wants to establish a framework for its AI governance without slowing down innovation? 

I would advise him to make it a project in its own right. 

We really need to give employees the time to learn, experiment with the tools, test use cases, and understand how AI can benefit their work. 

If teams are asked to do all of this in addition to their day-to-day work, without being given the time or resources to do so, AI risks remaining a theoretical concept or relying solely on a few particularly motivated individuals. 

We therefore need to establish a dedicated framework, with designated personnel, time for experimentation, and the ability to report use cases. 

In my view, setting aside time for experimentation doesn't slow down innovation. On the contrary, it allows us to move faster afterward, because we're testing the right use cases, with the right people, and in a controlled environment. 

Ultimately, an organization’s maturity isn’t measured solely by the number of AI tools it deploys. It’s evident in its ability to engage the right people, establish a clear framework, and turn experimentation into collective learning. 

AI becomes a strategic capability when it is no longer the sole domain of a few experts. It truly takes its place within the company when business units, technical teams, management, and the relevant users all participate together in the decisions that guide its use.