Why CIOs Don't Know Their Company's AI Strategy

It's understandable that rank-and-file employees lack clarity about what their company actually does with AI. For that confusion to reach the level of chief information officer (CIO) is harder to justify. Yet according to an article published this week on CIO.com, that's precisely what's happening in a significant number of organisations: senior technology leaders themselves can't find clarity about the direction AI is taking within their companies.

It's not that the technology is opaque or difficult to understand. The problem is structural, and in many cases, political.

Too many initiatives, too little coordination

One of the main causes highlighted in the article is the proliferation of AI projects launched from different business units simultaneously and without central coordination. Marketing experiments with content generation, operations pilots process automation, the legal team evaluates document review tools—and the CIO finds out, at best, only partially.

This phenomenon isn't new: it already happened with cloud adoption and earlier with SaaS software. The difference is that AI, especially large language models (LLMs), has security, privacy, and reputation implications that cannot be managed department by department without a common framework. When a team connects sensitive customer data to an external API without going through the systems area, the risk is not hypothetical.

The governance problem that arrives too late

Another pattern that emerges is governance strategies written after projects are already underway. Companies first experiment, then ask themselves who's responsible for what. This isn't necessarily a planning error: competitive pressure pushes many organisations to move before they have all the answers. But the cost of that inverted order is exactly what the article describes: executives who cannot clearly answer what their company's official position on AI is.

The situation worsens when AI oversight is split between the CIO, the CTO, and in some cases a newly created Chief AI Officer whose responsibilities overlap with those of the others. Without a clear line of responsibility, each assumes the other has the complete picture.

Who this affects

This diagnosis directly impacts several profiles:

• Technology teams in medium and large enterprises trying to prioritise which AI tools to adopt and needing to know whether a corporate policy exists to follow or if they have room to decide independently.
• AI consultants and solution providers entering conversations with enterprise clients only to find there's no single decision-making authority with real power.
• Business leaders who have launched AI pilots and wonder if they should scale them or wait for the company to define a common roadmap.
• Compliance and legal teams that need to know what data is being used, in which systems, and under what contractual conditions.

What can be done

The article doesn't offer a one-size-fits-all recipe, but the trend pointed out by several CIOs consulted suggests a clear direction: an honest inventory of all active AI projects is needed before designing any strategy. Without that map, any policy remains up in the air.

Some are betting on creating internal centres of excellence (AI Centers of Excellence) that act as a coordination point without blocking experimentation. Others opt for lightweight approval frameworks: not a committee reviewing each tool for months, but clear criteria about what types of projects need centralised review and which can advance autonomously.

What seems clear is that the model of "let each team try what they want and we'll sort it out later" has a limited shelf life. The more AI tools spread throughout an organisation, the more costly it becomes to reorganise afterward.

---

At ElephantPink, we continue to see in our clients and projects how the lack of a clear owner for AI decisions creates more friction than the technology itself. It's not a tools problem: it's an organisational structure problem that's best solved before, not after.