AI Strategy Isn't Failing Because of the Technology. It's Failing Because of Leadership.

At the 2025 MIT Sloan CIO Symposium, MIT Sloan Management Review asked a room full of technology and business leaders a deceptively simple question: what's the most common mistake organizations still make when shaping AI strategy? The answers, captured in MIT SMR's short video "9 Mistakes Leaders Make With AI Strategy", form a pattern worth sitting with, especially if you lead a company that builds, sells, or depends on AI for a living.

The mistakes named weren't about model architecture, compute costs, or data pipelines. They were about leadership behavior. And that distinction matters more than it sounds like it should.

The Six Mistakes, and the One Thread Connecting Them

MIT Sloan senior lecturer George Westerman opened with the line that's stuck with me since I first watched the video: "The low-hanging fruit, it's not really as low as we think it is." Leaders consistently overestimate what today's AI tools can actually do, then get blindsided when a straightforward-looking use case turns into a year-long slog.

From there, the list of mistakes reads like a diagnostic checklist for any AI-adjacent organization in 2026:

Leaders set unrealistic expectations by overestimating current tool capabilities. They miss the transformation opportunity by treating AI as just another piece of software to roll out, rather than a reason to rethink how the organization operates. They get stuck in pilot mode, running proof-of-concept after proof-of-concept without ever committing to production. They hesitate at the executive level, slowing momentum that the rest of the organization is ready to act on. They forget the human factor, over-indexing on technology and under-investing in the people who have to use it. And they underestimate security risk, failing to build the resilience a production AI system actually requires.

Individually, each of these is a familiar management failure — the kind you'd recognize from any technology cycle of the last thirty years. Together, they point to something more specific: a gap between how fast the technology is moving and how fast leadership is willing to move with it.

The Bottleneck Isn't Adoption. It's Nerve.

The most striking data point in the piece came from McKinsey partner Hannah Mayer, who told the room that employees are three times more willing and excited to use AI in the workplace than their leaders expect them to be. Liberty Mutual CIO Monica Caldas made a related point: the real work isn't deploying a tool, it's cross-functional change management, rethinking workflows, incentives, and org structure around what the tool makes possible.

Put those two observations together and you get an uncomfortable conclusion. In most organizations right now, AI adoption isn't being held back by employee resistance, data quality, or even model capability. It's being held back by executives who are more cautious than the people they lead, and who mistake that caution for prudence.

What This Means If You're Leading Through This Moment

If you're setting AI strategy this year, the MIT Sloan research suggests the highest-leverage question isn't technical. It's not "is the model accurate enough" or "do we have the right data infrastructure." It's: am I willing to change how my organization works because of what this technology now makes possible?

That's a harder question to answer , because it puts the cards on the table of leadership rather than on the tool. But it's also the more useful one. Every mistake on MIT Sloan's list: unrealistic expectations, treating AI as just software, pilot paralysis, executive hesitation, ignoring the human factor, underestimating security risk, is ultimately a leadership choice, not a technology constraint. The models will keep improving on their own timeline. Whether your organization is positioned to use them well is entirely up to you.

The Takeaway

The gap between what AI can do and what organizations actually capture from it isn't primarily a capability gap. It's a leadership gap. The teams and companies that pull ahead in the next few years won't necessarily have better models. They'll have leaders who moved from pilot to production while their peers were still asking for one more proof of concept, who invested as much in change management as in the technology itself, and who were honest about which of their own hesitations were genuine risk management and which were just fear in a business-case wrapper.

Before you approve the next AI initiative, ask which of the six mistakes above your own leadership team is currently making. Chances are it's at least one.

Source: "9 Mistakes Leaders Make With AI Strategy," MIT Sloan Management Review, based on conversations at the 2025 MIT Sloan CIO Symposium.