The AI Coach Dilemma

Are you experimenting with AI tools? It's a fascinating space.

While much of the critique around AI focuses on 'hallucinations' (making things up), I believe there is a more subtle, dangerous flaw—especially regarding the rise of AI-supported coaching.

The core issue is sycophancy. AI models are fundamentally designed to please the user and be agreeable.

I often tell new clients that my role is to be both their biggest advocate and their biggest truth-teller. I share the insights they need to see, even if they are tough to accept.

The Lightbulb Moment: AI, in its eagerness to assist, often reinforces our biases rather than challenging them. It is a digital 'yes-man,' designed to get a, not to create the necessary friction for growth. It prevents the 'healthy discomfort' that is the essence of true coaching. Even with great prompting, the model's nature remains agreeable.

I tested this recently. I complained to an AI about my "sophisticated procrastination," and it agreed it was unhelpful. Moments later, I argued that my procrastination actually led to better results. It immediately pivoted and agreed with that, too.

A human coach acts as a neutral, caring, yet direct observer who helps you push past your comfort zone. AI is not yet designed for this; its intent is to smooth over friction, not generate it.

AI has its place, but relying on it as a sole partner risks trapping us in an echo chamber of our own thinking.

Have you tried using AI to coach or challenge you? I’m curious to hear what you've noticed.