Leadership Talent Is Hiding in Plain Sight (And AI Can Spot It)

Are we missing great leaders because we're measuring the wrong things?

A fascinating new study from Harvard, MIT, and Wharton just discovered something that could change how companies find and develop leaders.

Here's the big idea: researchers found that how well someone leads human teammates strongly predicts how well they'll lead AI teams.

The connection is surprisingly strong, which almost never happens in studies about people.

I've seen firsthand how many companies struggle with finding the right leader for the job.

Finding good ones are expensive, complicated, and often based on gut feelings rather than data.

What the research is suggesting is there could be a simpler way to assess good leadership and make it available to more organizations.

#1. The Leadership Mirror: We've Been Missing A New Way to Test Leadership Skills

The most exciting finding from this research isn't just that AI tests predict human leadership.

It's what this tells us about leadership itself.

The experiment was clever: Have people lead teams of AI helpers to solve problems. Then see how those same people perform when randomly assigned to lead teams of real humans.

What's cool is that people who do well leading AI teams aren't just good with technology.

They show the same core skills that make them good with human teams.

This suggests something important about leadership: the basic skills that make someone a good leader work across different situations.

"This project really made me question what makes a good leader," says Dr. David Deming, one of the study's main researchers.

"The AI test seems to capture something fundamental about leadership ability."

Even better, the AI version of this test costs about 80% less than traditional methods that need multiple human participants.

This could make good leadership assessment available to organizations that previously couldn't afford it.

Quick Win: If you help develop leaders in your organization, try running a small AI based leadership test alongside whatever you're currently doing. You don't need to replace your existing methods right away, but this new approach might reveal leadership talent in unexpected places.

The researchers found no connection between leadership effectiveness and things like gender, ethnicity, or age.

This suggests AI based tests might help identify overlooked leadership talent and bring more diversity to leadership positions.

#2. The Power of Questions (Not Statements):
How Communication Style Predicts Leadership Success

What makes a leader successful with both humans and AI agents?

The answer challenges conventional wisdom about commanding presence.

The Harvard team conducted detailed analyses of communication patterns and found that effective leaders:

• Asked significantly more questions (rather than making statements)

• Engaged in more conversational turn-taking (not monopolizing discussion)

• Used more inclusive, plural pronouns like "we" and "us" (rather than "I" and "me")

Perhaps most surprisingly, the total volume of communication—simply talking more—showed no correlation with leadership effectiveness.

This flies in the face of the common association between leadership and dominant communication styles.

I've noticed this pattern myself when observing executive teams: the most impactful leaders often speak less but ask more penetrating questions.

They create space for others to contribute while guiding the overall direction through strategic inquiry.

The research also revealed an interesting difference between human and AI teams.

Expressing enthusiasm, encouragement, and optimism was strongly correlated with success in human teams but showed a weaker relationship with AI teams.

This suggests emotional leadership remains uniquely important for human collaboration.

Try This Now: Record your next three team meetings (with permission) and analyze your communication patterns:

1. Calculate your question-to-statement ratio

2. Measure how often you use "we/us" vs. "I/me" language

3. Track the distribution of speaking time among team members

Then in subsequent meetings, consciously increase your question rate by 50%, use more inclusive language, and ensure everyone gets roughly equal speaking time.

Monitor changes in team engagement, idea quality, and decision outcomes.

What's powerful about this approach is that asking better questions isn't just an assessment tool, it's a development strategy.

By practicing more inquiry-based leadership, you're simultaneously improving and demonstrating leadership skill.

#3. The Skills That Really Matter

The Harvard study shows which skills actually predict leadership success.

After controlling for job specific knowledge (what we might call "technical skills"), three factors stood out as powerful predictors of leadership success:

1. Fluid intelligence: The ability to solve new problems and think flexibly

2. Emotional perceptiveness: Being good at reading and responding to others' emotions

3. Decision making quality: Making good choices, especially with incomplete information

What's interesting is what didn't predict leadership success: things like gender, age, ethnicity, or education level showed almost no connection to leadership effectiveness.

This has big implications for how we develop leaders.

Many organizations focus mainly on technical expertise when promoting leaders.

This research suggests we've been looking at the wrong things.

Instead, the data points toward developing leaders by improving their decision making frameworks, emotional intelligence, and problem solving skills. These abilities seem to work well across different situations and team types.

I've seen this in my consulting work with leadership teams. The executives who succeed in challenging environments aren't necessarily the ones with the most industry experience, but those who can quickly adapt their thinking and build strong emotional connections with team members.

Smart Strategy: Create a leadership development program that focuses on these three skill areas:

• Decision making: Teach frameworks for making decisions with limited information and practice using them

• Emotional intelligence: Use assessments like the one in the study (called PAGE) to measure emotional awareness and help people improve

• Adaptive problem solving: Create challenges that require working across departments and synthesizing different types of information

Measure improvements in these areas and connect them to team performance over time. The Harvard research suggests these investments will pay off in better leadership effectiveness.

#4. The Confidence-Competence Gap:
How Self-Awareness Shapes Leadership Impact

Perhaps the most intriguing exploratory finding from the Harvard study relates to leadership emergence versus effectiveness.

The researchers discovered that willingness to take on leadership roles correlated strongly with overconfidence, participants who rated their abilities higher than their actual performance were much more likely to seek leadership positions.

However, there was an inverse relationship when it came to actual leadership performance.

Leaders with accurate self-assessments, those who correctly understood their own strengths and limitations, delivered significantly better team outcomes.

This creates what the researchers call "the leadership paradox": the people most eager to lead are often not the ones who would be most effective in those roles.

This mirrors what I've observed in executive selection processes.

The most charismatic and confident candidates often outshine more thoughtful, self-aware contenders in interview settings.

Yet those same self-aware leaders frequently deliver superior long-term results once in position.

The study suggests that organizations may be systematically promoting the wrong people—rewarding confidence over competence.

But it also points to a solution: using objective, performance-based assessments like the AI leadership test to identify leadership potential independent of self-presentation skills.

Leadership Opportunity: Implement a dual-track leadership identification system that separates self-nomination from performance-based assessment:

1. Create objective leadership simulations (potentially using AI-based assessments) that measure actual leadership capabilities

2. Compare these results with self and peer nominations for leadership roles

3. When discrepancies emerge, provide targeted coaching focused on self-awareness

4. Build regular feedback mechanisms that help leaders align their self-perception with reality

This approach doesn't eliminate confident candidates but ensures they also possess the fundamental skills necessary for leadership success.

#5. Why Finding Good Leaders Is So Hard

The task used in this study, called a "hidden profile" problem, reveals a lot about leadership.

In these scenarios, important information is spread among team members, with no single person having the complete picture.

Success requires leaders to gather and combine information from multiple people to make the best decision.

This mirrors real world leadership, where no single person can know everything.

Traditionally, measuring leadership effectiveness in these situations has been extremely difficult.

The Harvard approach, using AI as consistent, controllable team members, represents a breakthrough in how we can assess leadership.

What makes this valuable is that leaders must show actual leadership behaviors to succeed, not just talk about leadership concepts.

You can't fake effectiveness in a hidden profile task. You either get the critical information and make a good decision, or you don't.

The correlation with real world team leadership is strong. Leaders who do well on the AI test tend to do well with human teams, and vice versa.

This suggests we've found a measurement approach that captures something real about leadership ability.

Where to Start Tomorrow: Don't Overthink This

Pick one area where:

1. Leadership assessment currently feels subjective or inconsistent

2. You need to find leadership potential beyond the usual suspects

3. There's willingness to try innovative approaches

Start with a small test comparing AI based assessment results with known high performers in your organization.

Look for matches with current stars, but also pay attention to surprising results.

You may discover hidden leadership talent.

Use the results not just for selection but also for development feedback.

Share the specific behaviors that predict success and create practice opportunities for emerging leaders.

The Bottom Line: This research could change how we find and develop leaders. The implications go beyond academic interest.

They could reshape how organizations build their leadership teams.

The AI leadership test isn't replacing human judgment.

It's adding objective data about leadership capabilities that have been hard to measure.

The major reduction in cost and complexity makes sophisticated leadership assessment available to organizations that previously couldn't afford it.

Perhaps most importantly, this approach may help break the cycle where leadership positions go mainly to those who promote themselves rather than those who would be most effective.

By measuring actual leadership behaviors rather than confidence or charisma, we can build fairer pathways to leadership.

Organizations that adapt quickly to these insights will gain a significant advantage.

They'll find and develop leadership talent more effectively and efficiently than their competitors.

Start small. Test thoroughly. Scale thoughtfully.

The future of leadership development may have just arrived.

Never Stop Innovating,

Ben S. Cooper