Podcast: How Will We Train Developers If AI Does the Routine Work: A Conversation with Scott Hanselman

The conversation between Michael Stiefel and Scott Hanselman regarding developer training in an AI-augmented world strikes at a crucial juncture for our industry. The traditional model of junior developers learning through tackling routine coding tasks – the scaffolding upon which proficiency is built – is demonstrably shifting. As AI agents increasingly handle those repetitive duties, the question of how we cultivate the next generation of engineers becomes significantly more complex. Hanselman’s suggestion of a preceptorship model, mirroring the nursing profession, offers a compelling and surprisingly elegant solution. It acknowledges that the core of software development isn’t just about writing code, but about understanding systems, debugging effectively, and collaborating within a team – skills that require mentorship and real-world observation, not just algorithmic execution. This echoes concerns raised in What’s behind the AI industry’s latest warnings of doom?, highlighting the broader anxieties surrounding AI's impact, and points to a proactive approach to navigate these changes.

The shift to a preceptorship model necessitates a reimagining of the developer role, moving away from a focus on individual coding output towards a more holistic understanding of software architecture and problem-solving. It implies a greater emphasis on senior engineers acting as guides and mentors, actively shaping the learning process and providing context beyond the immediate task. This isn't merely about assigning a senior developer to a junior; it’s about cultivating a structured learning environment where observation, questioning, and guided experimentation are prioritized. We’ve seen similar challenges arise in other fields embracing automation, where the focus must shift from performing tasks to optimizing processes and managing automated systems. Consider, for example, the parallels to how database administration is evolving—managing and optimizing Postgres, as detailed in Article: Implementing Durable Workflows on Postgres Without an External Orchestrator, now requires a deeper understanding of performance tuning and architectural considerations than simply executing SQL queries.

The implications extend beyond individual training programs. Companies will need to invest in cultivating strong mentorship cultures, providing senior developers with the training and resources to effectively guide junior colleagues. It also suggests a potential restructuring of teams, with a greater emphasis on knowledge sharing and collaborative problem-solving. The success of this model hinges on a clear understanding of what skills *remain* essential for human developers in an AI-driven landscape. These are likely to be higher-level skills like system design, strategic thinking, and the ability to translate business needs into technical solutions – areas where AI, at least for the foreseeable future, will struggle to replicate human expertise. Insight Partners’ Deven Parekh’s perspective on diversification, as outlined in Insight Partners’ Deven Parekh on why the firm is diversifying while everyone else bets the farm on OpenAI and Anthropic, underscores the importance of a balanced approach, recognizing that human ingenuity will continue to be a critical differentiator.

Ultimately, Hanselman's proposal isn’t about resisting the rise of AI; it’s about adapting to it. It's about recognizing that the tools have changed, but the fundamental purpose of software development—to solve problems and create value—remains the same. The question now is not whether AI will change how we train developers, but how quickly and effectively we can adapt our educational systems and professional practices to embrace this new reality. Will the industry proactively adopt preceptorship models, or will we see a widening skills gap as the traditional training ground disappears, leaving a generation of developers ill-equipped to thrive in an AI-powered world?