In May 2025, three management scholars at MIT’s Sloan School, David Wingate, Barclay Burns, and Jay Barney, published a paper whose title arrived like a cold shower in the middle of the loudest conversation in business: Why AI Will Not Provide Sustainable Competitive Advantage. Their argument was not that AI is overhyped, or that its disruptions are exaggerated. It was something more unsettling: that the race every company in the world is currently running is, structurally, a race to a tie.
I have been thinking about that paper a great deal since because it names something that most AI strategy discourse tends to avoid. I am a Lebanese lawyer working across a region where the pressure to adopt AI is near-total, where the topic permenates boardroom conversations, and where the question of whether to adopt has seemingly been settled in favor of how fast. The MIT Sloan argument shifts the question again. And this time, the shift matters.
The gold rush has a catch
The case for adoption is not in dispute. According to global management consulting firm McKinsey’s 2025 State of AI survey, which drew responses from nearly 2,000 participants across 105 countries, 88 percent of organizations now use AI in at least one business function, up from 78 percent the year before. Research by the London School of Economics and consulting firm Protiviti, published in October 2024 and based on surveys of nearly 3,000 workers, shared their compelling findings that professionals using AI save an average of 7.5 hours per week.
But Wingate, Burns, and Barney identify a structural problem that no amount of enthusiasm about these numbers resolves. Every serious technical advance ultimately becomes equally accessible to every company. Algorithms commoditize and open-source models erode proprietary offerings within months of their release. The MIT Sloan paper’s argument is that talent is plentiful, hardware competition is fierce, and what may be proprietary at time of release becomes table stakes within months.
“Far from being a source of differentiation,” Wingate, Burns, and Barney write, “artificial intelligence will be a source of homogenization.” When everyone runs the same engine, the engine is no longer the race.
The homogenization trap
This is the paradox that most AI strategy fails to confront directly. The universality that makes AI valuable as a category makes it worthless as a differentiator. If your competitor has access to the same models, the same automation capabilities, and the same tools, the advantage does not accumulate to either of you, and your market ranking remains more or less the same.
What rises in value when the tools flatten out? As Harvard Business School’s Institute for Business in Global Society argued in September 2025, AI cannot reliably distinguish good ideas from mediocre ones. It cannot guide long-term business strategy. It cannot replicate the kind of cultural intelligence and contextual judgment that determines whether a decision is right for a specific market, a specific organization, a specific moment in time.
The economists Ajay Agrawal, Joshua Gans, and Avi Goldfarb, writing in IMF Finance & Development in June 2025, argue that AI is fundamentally a prediction machine. It processes inputs and generates probabilistic outputs at extraordinary scale and speed. But between the prediction and the decision sits judgment, the weighing of values, context, uncertainty, and stakes that cannot be reduced to pattern recognition.
As prediction becomes cheap, judgment becomes scarce. The antidote to AI homogenization is not more AI. It is better human thinking.
Automating mediocrity at scale
The problem is that most organizations are doing the opposite.
The RAND Corporation, in a report published in August 2024, found that more than 80 percent of AI projects fail to reach meaningful production deployment, at almost twice the failure rate of conventional IT projects, with the primary cause traced not to technology but to the broken organizational foundations underneath.
Nicholas Carr mapped the deeper cost of this dynamic in The Glass Cage, his 2014 study of automation across aviation, medicine, and financial trading. His argument echoes one prominent concern around AI adaptation, which is that when machines absorb skilled tasks, humans lose the capacity to perform those tasks independently. The pilot who delegates to autopilot loses situational awareness; when the system fails, the judgment needed to recover has already atrophied. The financial analyst who defers to algorithmic outputs loses the interpretive muscle that once gave those outputs meaning. Competence, Carr demonstrates, is something that must be practiced. Practices that go unused erode. And in the current race to automate as much as possible as fast as possible, entire categories of human capability are going unpracticed.
For this region, the stakes are specific. McKinsey’s 2025 research on the GCC shows that close to 90 percent of CEOs report using GenAI, above global averages. Yet only 11 percent are what McKinsey calls “value realizers”: organizations that have genuinely scaled AI and can attribute meaningful earnings to it. The gap between adoption and value creation is a readiness problem, and underneath the readiness problem is a human capital problem that no amount of software procurement resolves.
The cultural intelligence, relational depth, and contextual market knowledge that define competitive advantage in this part of the world cannot be generated by any model. They were built over decades. They are precisely what AI cannot replicate, and precisely what is most at risk in an undiscriminating adoption sprint.
Human-Centered AI Is Strategy, Not Philosophy
The phrase “human-centered AI” has acquired the texture of a values statement, the kind of language that appears in sustainability reports between carbon targets and inclusion metrics. It is neither. It is the only AI strategy that holds up structurally.
The companies that will extract durable advantage from AI are not the ones with the most tools. They are the ones that use AI to do more of what only humans can do, not less. That means fixing the process before automating it. It means investing in the quality of human judgment before deploying the tools that will amplify it. It means treating creative culture, contextual expertise, and institutional knowledge not as costs to be optimized away, but as the moat that gives AI outputs their value in the first place.
AI amplifies whatever it finds. A company with sharp human judgment and strong creative culture, deploying AI, becomes exponentially more capable. A company with broken processes and atrophied thinking, deploying the same tools, becomes exponentially more broken, faster, and at lower cost.
The questions for every business leader in this region is what, exactly, is the AI going to amplify? Is what you have built worth amplifying? And what essential skills might be lost in the process?
