Leaders who work with AI tend to lead higher-performing teams because AI prompts more team reflection and broadens members' confidence to take on roles; older leaders gain the largest benefits.

Main Finding

Leader-AI collaboration improves team performance indirectly by enhancing team reflexivity and team role-breadth self-efficacy; the evidence for reduction in team role overload as a mediator is mixed. Older leaders strengthen the positive effects of leader-AI collaboration on team reflexivity and role-breadth self-efficacy.

Key Points

• Theoretical framing: builds on the model of work role performance and frames leader-AI collaboration as supporting collective role-based functions that matter for team outcomes.
• Proposed role-based mechanisms:
  • Reducing team role overload (resource/coordination relief).
  • Enhancing team reflexivity (team-level reflection and adjustment).
  • Improving team role breadth self-efficacy (team confidence to perform broader role demands).
• Empirical results:
  • Team reflexivity and team role-breadth self-efficacy mediate the positive effect of leader-AI collaboration on team performance.
  • The mediating effect of reduced team role overload is inconsistent across studies.
  • Leader age moderates effects: older leaders amplify the positive relationship between leader-AI collaboration and both team reflexivity and team role-breadth self-efficacy.
• Implication about leadership composition: older leaders may realize greater team-level gains from AI collaboration in the digital workplace.

Data & Methods

• Design: Two independent studies using multi-source, multi-wave survey designs at the team level (i.e., repeated measures and multiple respondents per team).
• Measurement: Leader-AI collaboration (reported at leader/team level), team role-based constructs (role overload, reflexivity, role-breadth self-efficacy), and team performance (multi-source ratings).
• Analysis: Mediation and moderation tests to assess indirect effects of leader-AI collaboration on team performance through role-based functions, and moderation by leader age.
• Limitations of the methods (implicit from abstract): observational survey data limit causal claims; sample details and contexts not reported in the abstract, so generalizability and potential common-method concerns should be checked in the full paper.

Implications for AI Economics

• Productivity complementarities: Evidence that AI tools paired with human leaders can raise team productivity indirectly by improving coordination (reflexivity) and expanding teams’ self-efficacy to take on broader tasks — a form of human-AI complementarity that affects organizational output.
• Heterogeneous returns to AI investment: The moderating role of leader age implies heterogeneity in returns to AI adoption across managerial characteristics; firms should expect variable productivity gains depending on leadership demographics and capabilities.
• Adoption and training policy: To maximize economic returns from workplace AI, firms should invest not only in AI tools but in leader-centered integration practices and training that promote team reflexivity and broadened role efficacy. Targeted support for different age cohorts may yield higher ROI.
• Labor and organizational structure: By increasing teams’ willingness and capability to take on broader roles, leader-AI collaboration could shift task allocations, change job designs, and influence demand for different skill types — relevant for forecasting labor reallocation and wage effects.
• Research and measurement needs for AI economics:
  • More causal evidence (field experiments, randomized rollouts) to quantify productivity gains and cost-benefit ratios of leader-AI systems.
  • Firm-level and sectoral studies to map how managerial characteristics (age, experience, digital skill) mediate AI returns.
  • Longitudinal data linking leader-AI adoption to objective performance metrics (revenue, output, error rates) to estimate economic impact at scale.
• Caution for policymakers and economists: Survey-based findings point to promising channels but do not yet establish generalizable causal magnitudes; heterogeneity and context matter when translating these results into investment or regulation decisions.