At Prophet’s latest Uncommon Intelligence event in Abu Dhabi, there was no shortage of optimism in the room. Bringing together CMOs and senior business leaders from across industries including government, healthcare, hospitality and banking, the discussion reflected genuine excitement about what AI could make possible.

But that optimism came with a dose of realism: AI’s promise is enormous, but realizing it is proving more complicated.

Leaders are under pressure to move faster, automate more and avoid being left behind. Yet beneath that urgency sit harder questions: Where is AI creating meaningful value? What is the ROI? And why, despite rapid experimentation and adoption, does its impact so often fall short of expectations?

These are not uniquely Middle Eastern challenges. They reflect a much broader issue facing leaders around the world. Closing the gap between AI’s potential and its real-world impact requires organizations to look beyond it as another productivity tool.

That means moving from efficiency to effectiveness and, ultimately, enhancement. There was broad agreement that creating average work faster isn’t the goal. 

1. Start with the Outcome, Not the Tool

Perhaps the biggest danger surrounding AI right now is starting in the wrong place.

With so many slick new tools, agents and use cases emerging almost daily, there is an understandable temptation to ask, “How can we use this?” A better question is: “What are we trying to achieve?”

Productivity gains alone don’t guarantee competitive advantage. Organizations need to identify where AI can create real value, with a clear line of sight to business impact and ROI.

That requires leaders to resist chasing every new platform. With new technology emerging constantly, the challenge is increasingly about selecting the right capabilities for the right problem. In other words: Impact first, tools second.

2. AI Transformation is About Leading From the Front

If AI is going to change how an organization creates value, it cannot simply be delegated to an innovation team or technology function. Leaders need to own the outcome.

That means establishing priorities, allocating resources, and modeling the behaviors they want others to adopt. AI adoption isn’t only a technology transformation; it changes workflows, roles and ways of working.

This is particularly apparent in marketing, where there is significant opportunity across efficiency, effectiveness and enhancement, but functional expertise remains critical. Not every activity should become AI-first. Some areas are better suited to AI-powered collaboration, while others still depend heavily on practical human knowledge.

The onus on leadership is to understand the difference and set the direction.

3. Keep Human Intelligence Firmly in the Equation

There was also a realistic and pragmatic view about AI’s limitations. It isn’t a silver bullet, and speed is not the only measure of success.

AI can process, analyze and create, at remarkable scale but people still bring the expertise, judgment, context and creativity that determine whether the output is useful. This is where the idea of augmented intelligence becomes important: not AI replacing human intelligence, but the two working together to produce better outcomes. This means thinking about human and artificial intelligence as part of one extended workforce, each bringing different strengths to the task.

The opportunity lies in finding the right balance. Where can AI meaningfully improve the work? Where does human intelligence matter most? And where can combining the two deliver something genuinely better?

4. Interrogate AI, Don’t Evangelize it

We bring this to life with AI twins, which can be used to gather insights, test ideas, and pressure-test decisions. Rather than simply demonstrating another AI capability, the aim was to provoke a different question: What would you ask your digital twin?

When grounded in the right data, digital twins can challenge existing assumptions, surfacing perspectives that are direct, unexpected, and sometimes uncomfortable.

That is part of their value. Leaders don’t need AI to tell them what they already think, they need it to help uncover blind spots, challenge assumptions and provide new perspectives that can lead to better decisions.

It points to a broader mindset shift. The goal shouldn’t be to evangelize AI or adopt technology for technology’s sake. It should be to interrogate it. Where can it help us see something we couldn’t see before? Where should we challenge its output? And where can human judgment and AI together lead to a better decision?