Role and Identity Work of Software Engineering Professionals in the Generative AI Era

The adoption of Generative AI (GenAI) suggests major changes for software engineering, including technical aspects but also human aspects of the professionals involved. One of these aspects is how individuals perceive themselves regarding their work, i.e., their work identity, and the processes they perform to form, adapt and reject these identities, i.e., identity work. Existent studies provide evidence of such identity work of software professionals triggered by the adoption of GenAI, however they do not consider differences among diverse roles, such as developers and testers. In this paper, we argue the need for considering the role as a factor defining the identity work of software professionals. To support our claim, we review some studies regarding different roles and also recent studies on how to adopt GenAI in software engineering. Then, we propose a research agenda to better understand how the role influences identity work of software professionals triggered by the adoption of GenAI, and, based on that, to propose new artifacts to support this adoption. We also discuss the potential implications for practice of the results to be obtained.

💡 Research Summary

The paper investigates how the adoption of Generative AI (GenAI) reshapes the professional identities and identity work of software engineering practitioners, with a focus on role‑specific differences. It begins by highlighting that existing studies treat software engineers as a homogeneous group, overlooking the distinct self‑perceptions of developers, testers, architects, and other roles. Drawing on the concept of professional identity—defined as the constellation of attributes, skills, values, and experiences that individuals use to define themselves—the authors argue that identity work (the cognitive, discursive, physical, and behavioral activities people engage in to construct, revise, or discard self‑meanings) will be triggered by GenAI adoption, but in ways that depend on one’s role.

The literature review (Section 2) documents empirical evidence that roles differ not only in tasks but also in prestige and social standing. Architects are seen as “lone decision‑makers,” while testers often experience a “second‑class citizen” stigma, feeling excluded from decision‑making and undervalued. These hierarchical perceptions already shape identity formation and can become sources of conflict when work practices change.

Section 3 outlines how GenAI tools such as GitHub Copilot, Cursor, and conversational LLMs are already altering software development workflows. The “vibe coding” paradigm, where developers converse with a chatbot and spend most of their time inspecting generated code, exemplifies a shift from code creation to code validation. When combined with test‑driven development, GenAI may cause developers to focus more on testing the AI‑produced code, effectively blurring the line between developers and testers. This role convergence can generate role ambiguity, role conflict, and identity tension, which prior research links to reduced team productivity.

To address these gaps, the authors propose a detailed research agenda (Section 4) consisting of five research questions (RQ1‑RQ5). RQ1 and RQ2 aim to establish baseline professional identities for developers and testers through a systematic literature review (SLR). RQ3 investigates how GenAI adoption triggers identity work for each role, while RQ4 examines how it influences perceived relatedness among roles. Both RQ3 and RQ4 will be explored via qualitative interview studies (using “war stories”) and analysis of gray literature. Finally, RQ5 seeks prescriptive solutions—implemented through Design Science Research (DSR)—that embed identity‑aware practices, role‑redefinition models, and training/re‑skilling programs into GenAI adoption strategies.

The related‑work section (Section 5) integrates four theoretical lenses—social identity theory, critical theory, identity theory, and narrative theory—to provide a holistic view of identity work. It also references seminal models of professional identity transition (e.g., Ibarra’s three‑task framework) and longitudinal studies of identity construction in other domains (e.g., medical residency), arguing that similar mechanisms will apply in software engineering.

In conclusion, the paper emphasizes that identity work is especially fluid early in a career and during education, suggesting that curricula and corporate training must evolve to accommodate the new, AI‑mediated roles. At the organizational level, insights from the agenda could inform team structuring, hiring practices, and policies aimed at reducing role conflict. Overall, the study posits that understanding and supporting role‑specific identity work is essential for the successful, human‑centric integration of Generative AI into software engineering.