The AI Is Emotionally Self-Aware. Is the Boss?
Best-selling author Dan Goleman says new AI models are being trained in an emotional-intelligence skill that many leaders can’t—or won’t—master.
Daniel Goleman is author of the international best-seller Emotional Intelligence and Optimal: How to Sustain Personal and Organizational Excellence Every Day. He is a regular contributor to Korn Ferry.
This winter, Anthropic released what it's calling Claude's Constitution—a lengthy and detailed document outlining the values, character, and ethical commitments it wants its AI to embody. The authors made a choice: Rather than writing a list of rules for the AI to follow, they described virtues for it to develop. Rules, they reasoned, tell you what to do in anticipated situations while character is what guides you when no rule covers the moment.
This raises an uncomfortable question for many leaders: Have you done this work yourself?
According to organizational psychologist Tasha Eurich, 95% of people believe they are self-aware—meaning that they think they know what they value and how those values impact their emotions and behaviors. But the real number of self-aware people is somewhere between 10% and 15%, which means there's a striking gap between how people think they are and how they actually are.
Self-awareness is external and internal. Externally, it involves understanding how others perceive you. Internally, it points to something deeper: knowing what you value, what drives you, and how your emotions shape your behavior. It’s an important distinction, mostly because a person can be skilled at one and not at the other. For example, a leader may know how they land with their team, but have no idea how their internal world functions. They may see the disruption they cause but have no idea how to better manage their emotions or consistently align their actions with what they say they stand for.
Values have always been at the core of how humans organize themselves in decisions, partnerships, communities, cultures, and religions. They are also one of the most impactful mechanisms by which self-awareness becomes actionable. When you know what you stand for, you have a reference point for your own behavior — away to ask: Am I actually living this? Without that anchor, even good intentions go sideways.
Research has continually pointed to something counter intuitive: The more power and experience a leader accumulates, the less self-aware they tend to become. Not because they stop caring, but because they get so high up in the hierarchy that the feedback loops stop working. The more senior you are, the less likely the people around you are to tell you the truth.
What's notable about Claude's Constitution is not just that it asks the AI to hold values, but it also asks it to hold those values with humility; to remain open to being wrong; to update when presented with good reasons; and to approach ethical questions with curiosity rather than defensiveness.
This is a standard most leaders struggle to meet. Emotional self-awareness, the foundation of all emotional intelligence, requires not just knowing your values but also being honest enough to see when your behavior contradicts them. That kind of honesty demands the very thing that becomes harder as leaders rise: the willingness to receive feedback without defending against it.
The irony is stunning. We are building machines capable of moral self-correction, which is not necessarily a skill humans and organizations have already mastered.
The stakesare not abstract. Research shows that working with colleagues who aren’t self-aware can cut a team’s success in half. Which raises a question worth asking: Will AI's success be cut in half too, if it can't actually develop the self-awareness its constitution demands? Writing values down—for humans or machines—is the easy part. Living by them, especially under pressure, is another matter entirely.
Co-written by Elizabeth Solomon
Click here to learn more about Daniel Goleman's Building Blocks of Emotional Intelligence.
