AI: A Giant Mirror
Ask the Same Question with the Same Data and Get the Same Answer
“He fell in love with his own reflection in the water, not knowing it was merely an image. Without thought of food or rest, he could not be drawn away from gazing at his own reflection.” (Ovid’s Metamorphoses, on Narcissus)
I’m constantly asked about how to integrate proprietary data into VSTRAT, our AI-powered strategic planning platform. The conversation usually goes like this: “Can we upload our company data?”
Our simple answer is yes we can absolutely do this; use the Azure API we already use to upload a firm’s data. The better answer? We shouldn’t. And here’s why this popular impulse might be sabotaging your strategic thinking.
The Narcissus Problem in Strategic Planning
Your own data is, by definition, almost entirely backward-looking. It reflects where you’ve been, what you’ve done, and how you’ve competed in the past. When you feed this data into an AI system and ask for strategic insights, you’re essentially asking the machine to fall in love with your own reflection to generate recommendations based on the very patterns that may have led you into your current competitive position.
This creates what I call the “Narcissus Problem” in strategic planning: the dangerous attraction to our own organizational mirror. Just as Narcissus couldn’t tear himself away from his reflection, companies become mesmerized by their own data, believing that more of the same information will somehow yield breakthrough insights.
During my decade working with Chan Kim and Renée Mauborgne on Blue Ocean Strategy, Professor Kim consistently reinforced a crucial principle: don’t benchmark against your competitors, or you’ll remain trapped within your industry’s existing competitive boundaries. The same logic applies to your internal data. When you feed AI systems primarily your own historical information, you’re essentially benchmarking against yourself which is even more limiting than benchmarking against competitors.
The Comfort of Worn-Out Jeans
Your proprietary data might feel comfortable, like a pair of well-worn jeans. It’s familiar, broken-in, and fits exactly how you expect. But comfort isn’t the same as effectiveness. Those jeans might be threadbare and outdated, no longer serving their purpose. Sometimes you need to try on something new.
The current AI discourse seems obsessed with feeding more data into systems. “We need more data! Better data! Our data!” But consider this: you have access to silicon brains stuffed with enormous amounts of diverse, global information spanning countless industries, cultures, and contexts. Why would you want to constrain this vast intelligence to the microcosm of your own organizational experience?
It’s like having access to the entire Library of Alexandria but choosing to read only the books written by people from your hometown. You’re artificially limiting the system’s ability to make novel connections and generate unexpected insights.
Einstein Meets Blue Ocean Strategy
Einstein allegedly said that insanity is doing the same thing over and over while expecting different results. Kim and Mauborgne’s Blue Ocean Strategy takes this further: make your competitors irrelevant by reconstructing industry boundaries and rejecting the false trade-off between value and cost.
Both insights point to the same strategic imperative: breakthrough thinking requires breaking out of established patterns. When you feed AI systems the same data you’ve always analyzed, using the same frameworks you’ve always employed, you’re essentially automating your existing thought patterns. You might get faster analysis, but you won’t get fundamentally different insights.
The Hallucination Question
In our work with VSTRAT and generative AI, we’ve invested heavily in preventing hallucinations; those moments when AI systems generate false or nonsensical information. But I always clarify something important with our users: when the system generates an unexpected insight, don’t immediately dismiss it as a hallucination.
Instead, ask yourself: why is the system “seeing” what it’s seeing? What patterns is it detecting that your biological brain might be missing? The insight might indeed be a hallucination, but it could also be a genuine strategic revelation that emerges from the AI’s ability to process vast amounts of diverse information without the cognitive biases that limit human analysis.
This is where the magic happens when AI systems, drawing from their broad training data, identify patterns and possibilities that transcend your industry’s conventional wisdom. But this magic disappears when you constrain the system to your own historical data.
The Path Forward
So what should you do instead? Embrace the discomfort of the unknown. Let AI systems work with their full breadth of knowledge. Ask questions that force you to look beyond your own reflection:
What would a successful strategy look like if we weren’t constrained by our industry’s current boundaries?
How are completely different industries solving similar problems?
What assumptions about value creation might we be taking for granted?
If we started this business today, with today’s technology and market conditions, what would we do differently?
The goal isn’t to ignore your data entirely it’s to prevent that data from becoming a prison. Use your proprietary information to understand your current position, but don’t let it define your future possibilities.
Breaking the Mirror
Narcissus couldn’t break away from his reflection and ultimately wasted away. Don’t let your organization suffer the same fate. The most powerful strategic insights often come from outside your existing frame of reference, from patterns and possibilities that your own data simply cannot reveal.
Your data tells you where you’ve been. AI’s vast knowledge base can help you imagine where you could go. The choice is yours: will you fall in love with your own reflection, or will you look up and discover new horizons?
The future belongs to organizations brave enough to break their own mirrors.