Why AI Struggles with Leadership Coaching

In 2016, Google introduced AlphaGo Zero, a groundbreaking AI that made history by defeating the best human players in Go, an ancient and highly complex board game. AlphaGo's journey to mastery offers a fascinating look at how AI excels in closed environments with clear objectives—yet it struggles when applied to more ambiguous, human-centered tasks like leadership coaching.

AlphaGo’s success came from two key stages of learning:

• Learning by Imitation: AlphaGo studied thousands of games played by expert human players to understand foundational strategies.

• Self-Improvement: It played against itself repeatedly, receiving rewards each time it won, allowing it to refine its strategies without human input.

In just 40 days, AlphaGo became the best Go player in the world. This was possible because Go is a closed environment with a simple “reward function”—you either win or lose. This clarity allowed AlphaGo to play millions of games, rapidly learning what worked.

The Challenges of Artificial Intelligence in Coaching

When we examine Artificial Intelligence in Coaching—particularly leadership coaching—the situation becomes much more complex. Large Language Models (LLMs) like ChatGPT initially learn by imitating expert coaches through exposure to thousands of coaching conversations. However, leadership coaching isn’t as clear-cut as Go; there’s no obvious “win” condition. Coaching outcomes are subjective, vary from one individual to another, and don’t fit into neat categories of success or failure.

This lack of a clear reward function is a critical limitation. Unlike AlphaGo, which could perfect its strategy through endless self-play, AI in coaching can’t simply “practice” its way to mastery. Leadership coaching involves subtle cues, empathy, and context—elements that don’t translate into binary outcomes AI can easily learn from.

Understanding Why Artificial Intelligence in Coaching Hits Roadblocks

Without a defined feedback loop, Artificial Intelligence in Coaching faces several technical barriers:

• Subjective Success: Coaching effectiveness is personal. What works wonders for one leader may fall flat for another. AI relies on patterns and probabilities, but lacks nuanced judgment to adapt in real-time.

• No Immediate Feedback: Go provides a clear result after each game. Coaching, on the other hand, unfolds over time, and outcomes aren’t immediately measurable—slowing AI’s learning process.

• No Self-Improvement Loop: AlphaGo thrived by playing against itself. In coaching, no equivalent self-play loop exists, making it impossible for AI to refine its “skills” independently.

Each coaching conversation is unique, shaped by human complexity and emotion, which AI can’t yet fully interpret or respond to with genuine insight.

Why Human Coaches Are Still Essential

AI offers valuable tools to support business processes, but leadership coaching requires nuance, empathy, and deep contextual understanding—areas where human intelligence still outperforms. Effective coaching depends on trust, relationship-building, and the ability to respond fluidly to unpredictable human dynamics. These are things no AI system has mastered so far.

Leaders who recognize the limitations of Artificial Intelligence in Coaching can make more informed decisions, balancing the efficiency of AI tools with the irreplaceable value of human connection.

Looking Ahead

Artificial Intelligence will continue advancing, opening new opportunities across industries. However, when it comes to leadership coaching—where success is defined by growth, trust, and human complexity—the human touch remains essential. While AI may eventually bridge some gaps, for now, leadership coaching is best guided by skilled human coaches who bring empathy, adaptability, and real-world understanding to the table.