CAROL LEE
Gen AI Developer
This project was led by Dr. Cat Hicks (I was the supporting author & lead statistician)
Problem: Developers, engineering managers, and leaders grapple with failure to thrive while adapting to the rapidly emerging changes and expectations associated with Generative AI-assisted software work. A key factor influencing this struggle is AI Skill Threat: developers' fear, anxiety, and worry that their current skills will quickly become obsolete as they adapt to AI-assisted coding.
Outcomes & Solutions: We developed and tested the first empirical framework of AI Skill threat to predict when and why AI Skill Threat emerges, as well as what engineering leaders, teams, and developers can do about it. Our findings show that engineers who maintain a strong belief in competition and the demonstration of “innate brilliance” are more likely to report AI Skill Threat, and that software teams’ investment in strong cultures of learning and belonging can strengthen developers’ resilience and teams' effectiveness as they transition to AI-assisted software development. To help teams build strong cultures of learning and belonging as they transition to AI-assisted software development, we also created and disseminated a Generative AI Toolkit.
Purpose & Method: Survey study examining the prevalence of AI Skill Threat, as well as the risk and protective factors exacerbating or mitigating its influence.
Outcome: A large proportion of developers experience AI Skill Threat (N = 2348).
A moderated serial mediation analysis showed that developers feel more anxious and worried about AI when they doubt their skill, intellect, and accomplishments as developers - a process that occurs because they believe that they must be innately brilliant and ruthlessly competitive to succeed.
A mediation analysis showed that building a culture of Learning and Belonging can both directly and indirectly lessen Contest Cultures. Specifically, when developers experienced greater Belonging, this directly lessened Contest Culture. When developers felt supported to learn at work, this indirectly lessened Contest Culture by boosting their sense of Belonging
A serial mediation analysis showed that building a culture of thriving extend beyond the individual developer. For example, developers who felt like they belonged and were supported to learn, were more productive, which in turn positively impacted their team’s effectiveness.
Our research also reveals equity gaps in developers’ experiences with AI-assisted coding. AI Skill Threat is higher for Racially Minoritized developers, who also rate the overall quality of AI-assisted coding outputs significantly lower. Both women and LGBTQ+ developers were significantly less likely to report plans to upskilling for new AI-assisted workflows. These and other emerging differences point toward a critical need to understand how organizations ensure that AI-assisted coding adoption is equitable and accessible, and that key insights from developers with important perspectives on the risks of AI-assisted coding are heard.