Your AI Might Be Undermining Your Leadership

Part One of this series raised a critical question: what happens to human judgment when leaders begin thinking alongside machines? The answer depends on what those machines have been trained to think.

Most widely available AI tools are trained on the open internet, absorbing articles, opinions, frameworks, and advice written from vastly different levels of expertise. This gives them impressive range, but it doesn’t guarantee accuracy. A leadership model built from trending LinkedIn posts sits alongside peer-reviewed neuroscience. Viral frameworks carry the same weight as rigorous research. Confident opinions drown out nuanced evidence. The result is a system that sounds authoritative while reflecting the messy, contradictory nature of the internet itself.

This matters because leaders increasingly rely on these systems in real moments of leadership. You may already be using AI to draft feedback, shape a difficult conversation, or think through how to address a struggling employee. In those moments, the system is not simply helping you write faster. It is quietly influencing how you think about the situation and how you choose to respond.

Managing people is rarely what most leaders were originally hired to do, yet it consistently becomes the work that matters most. You are expected to make judgment calls about motivation, trust, accountability, and performance, often under time pressure and with incomplete information. When an AI system confidently suggests an approach, it is natural to assume that the confidence rests on solid ground.

But what if the guidance you are receiving is built on mixed-quality advice, recycled frameworks, and trending language rather than reliable knowledge about how people actually respond at work?

Consider a common scenario. You ask an AI tool to help draft feedback for an underperforming employee. The response sounds thoughtful, diplomatic, and well structured. It may even reference familiar leadership ideas such as “radical candor” or psychological safety. From the surface, it looks like exactly the kind of language a good leader should use.

But if that guidance was shaped by articles, blog posts, and thousands of performance reviews written by leaders of varying competence, the advice may not actually align with how people’s brains respond to difficult feedback. The language might inadvertently increase social threat at the exact moment the employee’s brain is least capable of receiving the message you are trying to deliver.

From your perspective, the conversation feels constructive. Yet the employee leaves the interaction more defensive, less trusting, and less open to change.

Weeks later, the real signal appears. Communication becomes more guarded, and engagement dips. The behavior you hoped to improve doesn’t change. And you are left wondering why the conversation didn’t land the way it seemed like it should.

This is the hidden risk of open-internet AI in leadership contexts.

This is where the conversation begins to shift from technology to leadership governance. Organizations carefully manage systems that influence financial decisions, legal exposure, and data security because leaders understand how those systems shape outcomes. Yet very few organizations have considered what happens when AI begins influencing leadership judgment itself. And once leaders begin asking where that intelligence comes from, the next realization is unavoidable: most AI available today was never designed for leadership at all.