Beware of the "Trend Quagmire": When AI Makes Your Strategy Mediocre

AI Is Quietly Flattening Your Strategic Differences

When the board asks, "Should we enter a new market or deepen our core product?" more and more executives turn to ChatGPT for advice. AI can generate data points, strategic frameworks, and confident final recommendations in seconds, rapidly becoming the most powerful strategic thinking partner. But a recent study from Esade Business School reveals a hidden trap: if everyone uses the same frontier models, the advice they receive will inevitably converge—strategy is being mass-produced and becoming mediocre.

The Root of the "Trend Quagmire"

In a paper published in the *Harvard Business Review*, the research team presented thousands of business dilemmas to mainstream AI models. They found a consistent pattern: the models systematically favor strategies aligned with modern buzzwords—differentiation over low cost, cooperation over competition, long-term over short-term. The reason is simple: these models are trained on millions of blog posts, LinkedIn posts, and business reviews, inheriting the biases in the corpus—those "sexier" ideas are repeatedly praised, while classic strategies like cost leadership and focus are marginalized as "boring" or even "oppressive." The researchers call this the "trend quagmire."

Worse still, AI advice is unpredictable. Asking the same question twice may yield contradictory recommendations depending on the order of options. And prompt engineering (e.g., "please weigh the pros and cons") cannot reliably correct the inherent biases.

Classroom Experiment: AI Converges Student Answers

The author conducted an experiment in a strategy class: students analyzed whether a company should differentiate or reduce costs. Half were allowed to use AI, while the other half relied only on their own cognition. The students using AI almost all ended up recommending a hybrid path of "pursuing both differentiation and low cost simultaneously"—they used AI to deepen their thinking, but their answers became more similar as a result. This small scenario mirrors the big problem unfolding in boardrooms around the world.

Five Boundaries: Maintaining Strategic Uniqueness in the Age of AI

The conclusion is not to abandon AI, but to define what can be outsourced and what must remain as one's own thinking. Here are five boundaries managers must set:

1. Mine offline knowledge. AI currently only knows what has been written and published online. It doesn't know that your CFO is risk-averse, your big client prefers long emails, or your supplier is quietly considering an acquisition. The richest strategic context exists over client dinners, factory tours, and water-cooler conversations. When everyone uses AI, public data is fully commoditized. Managers who invest in tacit knowledge will bring richer, more proprietary inputs to their decisions.2. Resist “false rigor” from context. People tend to believe that feeding AI more numbers, cases, and internal research will yield more accurate answers. However, research on “AI sycophancy” shows that models often adjust their responses based on the assumptions, preferences, and framing in the prompts. More context may instead create “false rigor”—answers become more detailed and persuasive, but not necessarily more independent or correct. The solution is not to give AI more context, but to have the model construct the strongest argument for the options it rejected.

3. Form your own opinion before prompting. Once you see an AI’s confident suggestion, it’s hard to return to your original state of mind. Even if you disagree with the model, its framing can influence your thinking due to its persuasiveness. To prevent algorithmic anchoring, write down your preliminary views before opening the chatbot. Managers should use AI to challenge already established foundations, not to let AI build those foundations.

4. Decide what to say no to. AI hates saying no. When faced with two strategic options, its instinct is to compromise. In studies, models often recommend both contradictory strategies simultaneously, even when that makes no business sense. Managers must do what AI cannot: make a choice. True strategy requires trade-offs between conflicting interests, structural mismatches, and resource constraints. The way to avoid converging with competitors is to choose what you *do not do*.

5. Consider the unsexy route. If AI models are leading other managers toward a “trend swamp,” the biggest untapped opportunities lie precisely where AI tells everyone not to go. When AI recommends trendy options, leaders should ask: What would it look like to execute the “boring” strategy with extreme discipline? In an era where every executive is driven by the same tools, competitive advantage will increasingly come from the contrarian bets that AI quietly discourages.

Conclusion: Keep strategy “non-AI-able”

AI is not going away. Used well, it can summarize complex information, generate alternative scenarios, and expose weak arguments at a speed no human team can match. But the more leaders rely on the same set of models, the more strategies will converge. The antidote is not to abandon AI, but to use it with clear boundaries. In an era of “trend swamps,” the most valuable strategy may well be the one no chatbot would ever recommend.