I'll never forget the moment I realized my most important job wasn't to have all the answers—it was to know which questions hadn't been asked yet. Years ago, before I started hosting conversations about AI and strategy, I watched a seasoned executive lose millions because he asked his team to optimize the wrong thing. They had built a beautiful model, elegant algorithms, dashboards that would make any data scientist weep with joy. But they had optimized for speed when the business actually needed resilience. The technology worked perfectly. The strategy failed catastrophically. That memory rushed back during my conversation with Rick Marchese, whose career arc from Navy officer to finance principal to ESSEC Business School professor has given him a rare vantage point on what actually breaks when humans and machines try to make decisions together.

Rick's insight cuts against everything Silicon Valley wants us to believe about AI transformation. We're told that competitive advantage flows from having the most sophisticated algorithms, the largest datasets, the fastest compute. But what Rick learned commanding people at sea—where bad decisions don't just cost money, they cost lives—is that the critical skill isn't technical prowess. It's the ability to frame the problem correctly before you ever touch a keyboard. "In the Navy, you learn very quickly that the right answer to the wrong question gets people killed," he told me, and the weight of that experience infuses everything he does now in private capital and teaching. When you've made decisions where the stakes are literally life and death, you develop what Rick calls "question discipline"—the practice of forcing yourself and your team to articulate not just what you're solving for, but why that particular problem matters and what you'll sacrifice to solve it.

This matters more now than ever because AI is exceptionally good at answering questions and catastrophically bad at questioning answers. I've watched this pattern repeat across industries: organizations deploy AI tools that optimize beautifully for metrics that turn out to be tangential to actual business outcomes. They automate processes that shouldn't exist in the first place. They generate insights at scale that nobody knows how to act on. The technology performs exactly as designed while the strategy slowly hemorrhages value, because no human with sufficient context and authority stopped to ask whether they were solving the right problem. Rick's experience in finance reinforced this lesson from a different angle. Managing capital means living with irreducible uncertainty, making bets with incomplete information, and accepting that some percentage of your decisions will be wrong no matter how sophisticated your models. The goal isn't perfect prediction—it's building systems resilient enough to survive your inevitable mistakes.

What strikes me most about Rick's framework is how it inverts the typical AI implementation process. Most organizations start with the technology and work backward: we have machine learning capabilities, now what should we apply them to? Rick starts with the strategic question—what decision would change our trajectory if we got it right?—and only then asks whether AI adds value to that specific decision-making process. Sometimes it does. Often it doesn't. But the discipline of framing the question first, with rigor and specificity, before you deploy any technology, is what separates AI initiatives that transform businesses from AI initiatives that generate impressive demos and then quietly disappear. This isn't about being anti-technology. Rick is deeply engaged with AI-driven finance and teaches the next generation of business leaders how to work with these tools. It's about understanding that algorithms amplify judgment rather than replacing it, and if the underlying judgment is aimed at the wrong target, amplification just means you'll fail faster and at greater scale.

The practical implication for leaders is uncomfortable: you cannot delegate the question-framing work. You can delegate the modeling, the data engineering, the technical implementation. But the strategic framing—what are we actually trying to achieve, what constraints matter most, what tradeoffs are we willing to make—requires someone who understands both the business context and the human systems that will have to live with the answers. Rick's experience teaching at ESSEC has shown him that young professionals coming up now often have impressive technical capabilities but lack the contextual judgment to know which problems are worth solving. That's not their fault. They've been trained in a world that valorizes optimization and efficiency, but strategy isn't optimization. Strategy is choice under uncertainty, which means the quality of your questions determines the ceiling of your outcomes.

This is why Rick's Navy background remains relevant decades after he left active service. Military organizations, whatever their other flaws, have centuries of experience training people to make consequential decisions under pressure with incomplete information. They've learned through painful iteration that you need robust processes for surfacing assumptions, challenging prevailing wisdom, and ensuring that junior voices can flag problems that senior leaders might miss. In contrast, many corporate environments actively suppress dissent and reward people for confidently executing on poorly-framed problems. Add AI to that environment and you've just turbocharged your organization's ability to implement bad strategy at scale. The most valuable thing Rick brings to his private capital work and his teaching isn't his technical knowledge—it's his insistence on creating space for the uncomfortable questions before anyone starts building solutions.

The truth is, most AI failures aren't technical failures. They're failures of imagination and interrogation. We implement AI when we should be restructuring the underlying process. We automate decisions when we should be rethinking the strategy. We optimize for speed when resilience would serve us better. The algorithm works fine. We just asked it to solve the wrong problem. And the only way to catch that mistake is to have humans in the room—preferably humans with Rick's kind of cross-domain experience—who have the judgment, the courage, and the organizational authority to stop the momentum and ask: are we solving the right problem? The most important person in the room isn't the one with the most sophisticated algorithm. It's the one who knows what question to ask before the algorithm runs.I'll never forget the moment I realized my most important job wasn't to have all the answers—it was to know which questions hadn't been asked yet. Years ago, before I started hosting conversations about AI and strategy, I watched a seasoned executive lose millions because he asked his team to optimize the wrong thing. They had built a beautiful model, elegant algorithms, dashboards that would make any data scientist weep with joy. But they had optimized for speed when the business actually needed resilience. The technology worked perfectly. The strategy failed catastrophically. That memory rushed back during my conversation with Rick Marchese, whose career arc from Navy officer to finance principal to ESSEC Business School professor has given him a rare vantage point on what actually breaks when humans and machines try to make decisions together.

Rick's insight cuts against everything Silicon Valley wants us to believe about AI transformation. We're told that competitive advantage flows from having the most sophisticated algorithms, the largest datasets, the fastest compute. But what Rick learned commanding people at sea—where bad decisions don't just cost money, they cost lives—is that the critical skill isn't technical prowess. It's the ability to frame the problem correctly before you ever touch a keyboard. "In the Navy, you learn very quickly that the right answer to the wrong question gets people killed," he told me, and the weight of that experience infuses everything he does now in private capital and teaching. When you've made decisions where the stakes are literally life and death, you develop what Rick calls "question discipline"—the practice of forcing yourself and your team to articulate not just what you're solving for, but why that particular problem matters and what you'll sacrifice to solve it.

This matters more now than ever because AI is exceptionally good at answering questions and catastrophically bad at questioning answers. I've watched this pattern repeat across industries: organizations deploy AI tools that optimize beautifully for metrics that turn out to be tangential to actual business outcomes. They automate processes that shouldn't exist in the first place. They generate insights at scale that nobody knows how to act on. The technology performs exactly as designed while the strategy slowly hemorrhages value, because no human with sufficient context and authority stopped to ask whether they were solving the right problem. Rick's experience in finance reinforced this lesson from a different angle. Managing capital means living with irreducible uncertainty, making bets with incomplete information, and accepting that some percentage of your decisions will be wrong no matter how sophisticated your models. The goal isn't perfect prediction—it's building systems resilient enough to survive your inevitable mistakes.

What strikes me most about Rick's framework is how it inverts the typical AI implementation process. Most organizations start with the technology and work backward: we have machine learning capabilities, now what should we apply them to? Rick starts with the strategic question—what decision would change our trajectory if we got it right?—and only then asks whether AI adds value to that specific decision-making process. Sometimes it does. Often it doesn't. But the discipline of framing the question first, with rigor and specificity, before you deploy any technology, is what separates AI initiatives that transform businesses from AI initiatives that generate impressive demos and then quietly disappear. This isn't about being anti-technology. Rick is deeply engaged with AI-driven finance and teaches the next generation of business leaders how to work with these tools. It's about understanding that algorithms amplify judgment rather than replacing it, and if the underlying judgment is aimed at the wrong target, amplification just means you'll fail faster and at greater scale.

The practical implication for leaders is uncomfortable: you cannot delegate the question-framing work. You can delegate the modeling, the data engineering, the technical implementation. But the strategic framing—what are we actually trying to achieve, what constraints matter most, what tradeoffs are we willing to make—requires someone who understands both the business context and the human systems that will have to live with the answers. Rick's experience teaching at ESSEC has shown him that young professionals coming up now often have impressive technical capabilities but lack the contextual judgment to know which problems are worth solving. That's not their fault. They've been trained in a world that valorizes optimization and efficiency, but strategy isn't optimization. Strategy is choice under uncertainty, which means the quality of your questions determines the ceiling of your outcomes.

This is why Rick's Navy background remains relevant decades after he left active service. Military organizations, whatever their other flaws, have centuries of experience training people to make consequential decisions under pressure with incomplete information. They've learned through painful iteration that you need robust processes for surfacing assumptions, challenging prevailing wisdom, and ensuring that junior voices can flag problems that senior leaders might miss. In contrast, many corporate environments actively suppress dissent and reward people for confidently executing on poorly-framed problems. Add AI to that environment and you've just turbocharged your organization's ability to implement bad strategy at scale. The most valuable thing Rick brings to his private capital work and his teaching isn't his technical knowledge—it's his insistence on creating space for the uncomfortable questions before anyone starts building solutions.

The truth is, most AI failures aren't technical failures. They're failures of imagination and interrogation. We implement AI when we should be restructuring the underlying process. We automate decisions when we should be rethinking the strategy. We optimize for speed when resilience would serve us better. The algorithm works fine. We just asked it to solve the wrong problem. And the only way to catch that mistake is to have humans in the room—preferably humans with Rick's kind of cross-domain experience—who have the judgment, the courage, and the organizational authority to stop the momentum and ask: are we solving the right problem? The most important person in the room isn't the one with the most sophisticated algorithm. It's the one who knows what question to ask before the algorithm runs.