top of page
Search

Measuring leadership skills with AI agents


Overview

We introduce a novel, scalable method for measuring leadership skills by evaluating individuals’ ability to lead teams of AI agents. Performance on this “AI leadership test” strongly predicts the causal impact leaders have on human teams, which we identify by randomly assigning leaders to multiple groups of human followers and measuring team performance. Our findings indicate that AI agents can be effective proxies for human participants in social experiments, which could greatly simplify the measurement of leadership and teamwork skills.


Key Empirical Findings

  • Variation in leader quality has a large and significant impact on group performance, for both the AI leadership test and in human teams

  • Leaders’ scores on the AI leadership test was strongly correlated with their causal impact on human teams (disattenuated correlation = 0.81).

  • Successful leaders of both humans and AI agents ask more questions and engage in more conversational turn-taking

  • Successful leaders score higher on measures of social intelligence, fluid intelligence, and decision-making skill, but do not differ in gender, age, ethnicity or education

 
 
 

Comments

bottom of page