Companies are racing to appoint chief AI officers, launch AI centers of excellence, and pour money into scaling AI initiatives. Use cases are multiplying. Yet McKinsey's latest research suggests that nearly two thirds of organizations still aren't reporting a positive, enterprise-level earnings impact from AI.

But here's a question worth asking: Who's actually responsible for the decisions AI is supposed to improve?

Consider a familiar dilemma within a manufacturing outfit. AI predicts that a critical machine is likely to fail. Maintenance recommends shutting it down. Production has customer orders to fill. Supply chain warns that replacement parts won't arrive for three days. Finance wants to know which option costs less.

Everyone has a reasonable argument. And everyone is optimizing for a different outcome.

Now imagine an AI agent that evaluates those trade-offs and recommends the best course of action for the business. Great. Who makes the call?

This is where industrial AI gets interesting, and where organizational accountability becomes more important than another algorithm.

We've spent years investing in systems that tell us what's happening and what might happen next. AI is increasingly capable of helping us determine what to do about it.

But AI cannot decide who has the authority to act, change competing departmental incentives, or establish accountability for enterprise-wide outcomes.

That's a leadership responsibility. The CIO, CDTO and chief AI officer can build the capabilities. Operational leaders must retain clear authority over plant-level decisions, particularly those involving safety. But when competing priorities require changing how the enterprise operates, the CEO must ultimately own that transformation.

You can appoint a chief AI officer. But you can't outsource the CEO's responsibility for what AI does to your business.

Kumar Dronamraju, independent executive advisor