Microsoft executives are warning that the rise of artificial intelligence coding tools could exacerbate a skills gap in software engineering, potentially hindering the development of future technical leaders. The concerns, detailed in a paper titled “Redefining the Engineering Profession for AI,” stem from observations that AI currently boosts the productivity of senior engineers while simultaneously reducing the productivity of those early in their careers.
Mark Russinovich, CTO of Microsoft Azure, and Scott Hanselman, VP of Developer Community, co-authored the report, which argues that a failure to proactively address this disparity could lead to a “hollowing out” of the next generation of software engineers. According to the paper, AI coding assistants can produce code with significant bugs, implement inefficient algorithms, and introduce inconsistencies throughout a codebase. These issues require experienced engineers to identify and correct, a task that junior developers are currently less equipped to handle.
Russinovich, in a prior discussion of the issue, noted that the premise – AI’s divergent impact on senior and junior developers – is a recurring theme in conversations with Microsoft’s customers. “They all say they see it at their companies,” he stated, as reported by The Register.
The paper highlights a specific example where an AI agent attempted to resolve a race condition by inserting a delay using Thread.Sleep, a solution that merely masked the underlying problem. Only an engineer with a strong understanding of synchronization code could recognize the inadequacy of this approach.
The authors suggest that organizations may respond to this dynamic by reducing their hiring of early-in-career (EiC) developers, focusing instead on experienced engineers who can effectively leverage AI tools. This, they contend, would be a short-sighted strategy with long-term consequences.
Hanselman, in a LinkedIn post promoting the paper, advocates for a program of deep investment in the EiC pipeline, emphasizing the necessitate for formal mentorship – modeled after the preceptor system used in nursing – to cultivate a strong base of future senior engineers. He stated the paper was published in the Communications of the ACM, Association for Computing Machinery.
One commenter on Hanselman’s LinkedIn post, Toby Kohlenberg, raised concerns about the potential difficulty of implementing such a mentorship program within existing company cultures, citing examples of competitive environments where seeking help from senior engineers is not always encouraged. He suggested tying mentorship to the review process, explicitly recognizing “human impact” alongside product impact.
Hanselman and Russinovich also collaborate on several developer-focused video podcasts, including “Scott and Mark Learn To” and “Azure Friday,” where they explore new technologies and share insights with the developer community. Hanselman also hosts the “Hanselminutes” podcast, featuring interviews with technology leaders.
