This dissertation explores how emerging technologies and evolving economic conditions are reshaping work. It covers both the micro-level of the individual and team, and it covers the macro-level of labor markets. There are two central themes: (1) how artificial intelligence can augment human cooperation and performance, and (2) how artificial intelligence has altered and will alter the labor market.

Organized into three stand-alone essays, the dissertation first explores human–AI teaming from a team-level perspective. In Chapter 1, we design a network of AI agents that learn the mental states of their human teammates using a Bayesian theory-of-mind approach. In an experiment with 145 participants across 29 teams playing a cooperative game, these agents detect misalignments in group communication. They then propose targeted interventions that improve team performance. While unaided human teams achieve a 66% success rate on a cognitively challenging task, counterfactual teams augmented by AI agents surpass 77%. The addition of AI-recommended fixes increases that result to over 82%. These findings underscore AI’s promise in boosting group performance for complex problem-solving.

Chapter 2 investigates how seemingly small design choices shape human perceptions in a team context. Through a controlled experiment where teams of participants solved a puzzle with an AI agent, we find that AI teammates with human-sounding voices can heighten perceived anthropomorphism and animacy, but only if the AI’s contributions are consistently useful. This effect is entirely flipped when an AI teammate is unhelpful. Interestingly, the AI’s perceived intelligence and trustworthiness were solely determined by the agent's contribution quality rather than vocal humanness. This suggests that function often outweighs form in achieving high-performing human–agent teams, but that form still matters when it misaligns with user expectations.

In Chapter 3, we shift to macro-level patterns by analyzing extensive job posting data in the U.S. labor market. We focus on how generative AI adoption influences hiring for software developers. Using real-time vacancy data from Lightcast, we uncover a sharp decline in junior-level openings immediately after ChatGPT’s public release in November 2022. Senior roles were relatively less affected. Although generative AI can enhance entry-level productivity, our evidence suggests that organizations are reducing demand for inexperienced workers. While this reaction may be temporary while the labor market return to equilibrium, this has potentially far-reaching implications for newly trained developers, college enrollment decisions, and the future skill composition of the workforce.

These three chapters illustrate how the evolution of AI is simultaneously local and global. Understanding how to design AI that fosters teamwork and trust is vital. At the same time, integrating these technologies into the labor market requires new education, training, and economic policies. We must balance productivity gains with evolving employment pathways.