Thrilled (and frankly, exhausted in that great-bone-deep-I-made-something way) that I can now share this with you: the Developer Success Lab's latest empirical research bringing social science to developers, The New Developer, a co-authored project from me, Dr. Carol Lee, and Kristen Foster-Marks -- a true dream team that brought the social science of achievement and learning, the clinical science on wellbeing, behavior change, and anxiety, and practical engineering best practices expertise together.
In this open, freely available research, we share a deep technical paper on how AI adoption is hitting developers' wellbeing right now, paired with an action-oriented toolkit based on our research findings, which includes many resources for holding a learnathon and a pre-mortem to increase belonging and learning culture for software teams.
From the beginning of this research, we challenged ourselves to take a human-centered and multi-systems approach to developer experience & the transition to AI-assisted coding on software teams. There are many challenges and unanswered questions about how AI will be used in the world of software development. Ultimately, we believe it is developers themselves who will answer these questions as they move forward into the future of AI-assisted coding. Despite the uncertainty of this time, moments of change are also a transformational opportunity to examine our shared definitions of success, productivity, and knowledge work. Our focus was therefore on the key human sociocognitive factors that impact developers' thriving in this moment.
Our original empirical research with 3000+ developers introduces a new framework: AI Skill Threat. This framework describes pervasive stress, anxiety and uncertainty that a significant portion of developers (43-45% across our measures) are experiencing right now about how AI-assisted coding will change their professional identities. In The New Developer project, we explore not only what is happening with AI Skill Threat, we also explore why, and for whom: key difference in beliefs about software work success and brilliance lead to more AI Skill Threat, increase developers' imposterism, *and* dampen both individual productivity and overall team effectiveness: Contest Culture vs Thriving Culture.
Simply put, a developer who believes that they must be innately brilliant & ruthlessly competitive to succeed will feel particularly anxious & threatened when faced with the changes that AI brings to their work, and others’ changing evaluations. How we face this moment for our teams may have long-term impacts on career trajectories and developer wellbeing. We connect this model to previous large-scale findings on "the brilliance trap," a cycle of beliefs which have been shown to be a critical mechanism in how opportunity gaps perpetuate across STEM fields and areas that double down on field specific beliefs that success is only determined by innate brilliance, and must be demonstrated in constant technical contests.
And because of our deep investment in recruiting underrepresented perspectives across all of our projects, we are able to share evidence that there are critically important equity and opportunity gaps emerging across developers' experiences with AI-assisted coding. For example, 56% of Racially Minoritized developers report a negative judgment of the quality of AI outputs in coding work, compared with 28% for all developers, and report higher levels of AI Skill Threat. Women and LGBTQ+ people on software teams are less likely to report upskilling in AI-assisted coding, potentially signaling a lack of resources. It is vital to learn from these groups' insights as well as take action to ensure these gaps do not calcify into new, career-threatening barriers.
It is incredibly meaningful to me that we were able to build on our Developer Thriving framework in this project, and found increasing evidence for the importance of developers' learning and belonging in this moment. I've written about sense of belonging here, and I am thrilled that the Dev Success Lab continues to pilot the first software team specific measures of belonging, gathering evidence of its importance with real professional developers. I've recently written about the belonging element of our Dev Thriving framework and how leaders and orgs can impact it. In this project, we continue to find that belonging can significantly impact not only individual productivity, but overall team effectiveness, and is a powerful intervention point for organizations that want to mitigate AI Skill Threat.
