I still remember the moment I realized my assumptions about AI in the workplace were dangerously narrow. It was during my conversation with Professor Robert Ployhart, the Bank of America Professor of Business Administration at the Darla Moore School of Business, University of South Carolina, when he casually mentioned a recruiting algorithm that worked beautifully in testing but systematically excluded qualified candidates in deployment. The algorithm wasn't broken, he explained—it was revealing something more unsettling. The system had learned to optimize for efficiency while quietly eroding the very human capital companies claimed to value most. That moment crystallized something I'd been circling around for months: we're not just adding AI to our organizations. We're fundamentally rewriting the social contract between technology, talent, and institutional memory.

The point that keeps me awake isn't about whether AI will transform human resources—that ship has sailed. What matters now is whether we understand what we're actually transforming. Ployhart spent decades consulting with Fortune 500 companies on talent management and leadership development, and his perspective cuts through the usual noise. Human capital, he reminded me, isn't just individual skills or credentials. It's the collective knowledge, relationships, and adaptive capacity that exists within and between people in an organization. When we deploy AI in recruitment, we're not simply speeding up resume screening. We're making decisions about which patterns of human potential we choose to recognize and amplify, and which we render invisible. The evidence from his research shows that AI systems excel at identifying efficiency gains—faster hiring cycles, reduced time-to-fill, better matching of keywords to job descriptions. But efficiency and effectiveness are not the same thing. An algorithm can find you a candidate who checks every box on a job description in half the time, yet completely miss the person whose unconventional background would bring the adaptive capacity your organization desperately needs for an uncertain future.

This distinction matters more than most leaders realize because of how human capital actually accumulates in organizations. During our conversation, Ployhart walked me through something that challenged my own thinking as a host focused on AI's revolutionary potential. He explained that when we talk about organizational capability, we're really talking about emergence—the way collective intelligence and adaptability arise from networks of people working together over time. You can't simply replace nodes in that network with algorithmic decisions and expect the emergent properties to remain intact. Every hiring decision, every performance review, every promotion shapes not just individual trajectories but the very substrate on which organizational learning happens. When AI makes those decisions optimizing for narrow metrics, we risk what he called "capability erosion"—a slow degradation in the organization's ability to sense, respond, and adapt that doesn't show up in quarterly reports until it's too late. The mechanism here is subtle but brutal: we trade short-term optimization for long-term resilience, and we don't realize we've made the trade until the environment shifts and we discover our carefully curated workforce can't improvise.

I pressed Ployhart on this during our discussion because I wanted to understand where the line is. If AI is genuinely better at predicting job performance based on historical data, isn't that exactly what we should want? His answer reshaped how I think about the ethics of deployment. Predictive accuracy, he noted, is always backwards-looking. It tells you who would have succeeded in your organization as it was, not who will help your organization become what it needs to be. The companies he works with are wrestling with this daily. They're discovering that the very characteristics that made someone promotable in a stable environment—consistent execution, pattern recognition, optimization of existing processes—may be precisely wrong for leading through discontinuous change. AI trained on historical promotion data will naturally favor the first set of characteristics because that's what success looked like in the training data. But the future demands the second set. This is why I've become convinced that the crucial question isn't whether AI can do HR tasks, but whether we're asking AI to optimize for the right outcomes. Most organizations haven't seriously grappled with that question because they're still thinking about AI as a tool that makes existing processes faster, rather than as a force that fundamentally changes what those processes select for.

What leaders should do next is both simpler and harder than implementing another platform. We need to get rigorous about what we actually want to preserve and cultivate in our human capital systems. That means being explicit about the trade-offs. Ployhart's work with Fortune 500 companies reveals that the most sophisticated organizations aren't asking "Should we use AI in talent management?" They're asking "For which decisions does AI's pattern recognition enhance our strategic intent, and for which decisions does it undermine the adaptive capacity we'll need?" They're conducting algorithmic audits not just for legal compliance but to understand what human potential their systems are systemically overlooking. They're experimenting with hybrid models where AI handles high-volume, low-stakes decisions while preserving human judgment for choices that shape organizational culture and capability. Most importantly, they're investing in building what Ployhart calls "algorithmic literacy" among HR professionals and line managers—the ability to interrogate what an AI system is actually optimizing for and whether that aligns with long-term human capital strategy.

As I reflect on this conversation, I keep returning to a realization that feels both obvious and radical: we're not in an era where AI is transforming work—we're in an era where our choices about AI deployment are revealing what we truly value about human contribution, and those revelations are uncomfortable. The future of work isn't being determined by technological capability; it's being determined by whether we're brave enough to design systems that preserve what makes organizations genuinely adaptive, even when that means sacrificing some efficiency in the short run.I still remember the moment I realized my assumptions about AI in the workplace were dangerously narrow. It was during my conversation with Professor Robert Ployhart, the Bank of America Professor of Business Administration at the Darla Moore School of Business, University of South Carolina, when he casually mentioned a recruiting algorithm that worked beautifully in testing but systematically excluded qualified candidates in deployment. The algorithm wasn't broken, he explained—it was revealing something more unsettling. The system had learned to optimize for efficiency while quietly eroding the very human capital companies claimed to value most. That moment crystallized something I'd been circling around for months: we're not just adding AI to our organizations. We're fundamentally rewriting the social contract between technology, talent, and institutional memory.

The point that keeps me awake isn't about whether AI will transform human resources—that ship has sailed. What matters now is whether we understand what we're actually transforming. Ployhart spent decades consulting with Fortune 500 companies on talent management and leadership development, and his perspective cuts through the usual noise. Human capital, he reminded me, isn't just individual skills or credentials. It's the collective knowledge, relationships, and adaptive capacity that exists within and between people in an organization. When we deploy AI in recruitment, we're not simply speeding up resume screening. We're making decisions about which patterns of human potential we choose to recognize and amplify, and which we render invisible. The evidence from his research shows that AI systems excel at identifying efficiency gains—faster hiring cycles, reduced time-to-fill, better matching of keywords to job descriptions. But efficiency and effectiveness are not the same thing. An algorithm can find you a candidate who checks every box on a job description in half the time, yet completely miss the person whose unconventional background would bring the adaptive capacity your organization desperately needs for an uncertain future.

This distinction matters more than most leaders realize because of how human capital actually accumulates in organizations. During our conversation, Ployhart walked me through something that challenged my own thinking as a host focused on AI's revolutionary potential. He explained that when we talk about organizational capability, we're really talking about emergence—the way collective intelligence and adaptability arise from networks of people working together over time. You can't simply replace nodes in that network with algorithmic decisions and expect the emergent properties to remain intact. Every hiring decision, every performance review, every promotion shapes not just individual trajectories but the very substrate on which organizational learning happens. When AI makes those decisions optimizing for narrow metrics, we risk what he called "capability erosion"—a slow degradation in the organization's ability to sense, respond, and adapt that doesn't show up in quarterly reports until it's too late. The mechanism here is subtle but brutal: we trade short-term optimization for long-term resilience, and we don't realize we've made the trade until the environment shifts and we discover our carefully curated workforce can't improvise.

I pressed Ployhart on this during our discussion because I wanted to understand where the line is. If AI is genuinely better at predicting job performance based on historical data, isn't that exactly what we should want? His answer reshaped how I think about the ethics of deployment. Predictive accuracy, he noted, is always backwards-looking. It tells you who would have succeeded in your organization as it was, not who will help your organization become what it needs to be. The companies he works with are wrestling with this daily. They're discovering that the very characteristics that made someone promotable in a stable environment—consistent execution, pattern recognition, optimization of existing processes—may be precisely wrong for leading through discontinuous change. AI trained on historical promotion data will naturally favor the first set of characteristics because that's what success looked like in the training data. But the future demands the second set. This is why I've become convinced that the crucial question isn't whether AI can do HR tasks, but whether we're asking AI to optimize for the right outcomes. Most organizations haven't seriously grappled with that question because they're still thinking about AI as a tool that makes existing processes faster, rather than as a force that fundamentally changes what those processes select for.

What leaders should do next is both simpler and harder than implementing another platform. We need to get rigorous about what we actually want to preserve and cultivate in our human capital systems. That means being explicit about the trade-offs. Ployhart's work with Fortune 500 companies reveals that the most sophisticated organizations aren't asking "Should we use AI in talent management?" They're asking "For which decisions does AI's pattern recognition enhance our strategic intent, and for which decisions does it undermine the adaptive capacity we'll need?" They're conducting algorithmic audits not just for legal compliance but to understand what human potential their systems are systemically overlooking. They're experimenting with hybrid models where AI handles high-volume, low-stakes decisions while preserving human judgment for choices that shape organizational culture and capability. Most importantly, they're investing in building what Ployhart calls "algorithmic literacy" among HR professionals and line managers—the ability to interrogate what an AI system is actually optimizing for and whether that aligns with long-term human capital strategy.

As I reflect on this conversation, I keep returning to a realization that feels both obvious and radical: we're not in an era where AI is transforming work—we're in an era where our choices about AI deployment are revealing what we truly value about human contribution, and those revelations are uncomfortable. The future of work isn't being determined by technological capability; it's being determined by whether we're brave enough to design systems that preserve what makes organizations genuinely adaptive, even when that means sacrificing some efficiency in the short run.