The Hidden Costs of Coding With Generative Al

Generative Al can boost coding productivity, but careless deployment creates technical debt that cripples scalability and destabilizes systems.

- By Edward Anderson, Geoffrey Parker, and Burcu Tan

Generative AI can be a powerful productivity booster in coding — but only when deployed thoughtfully. Used carelessly, it can cripple scalability, destabilize systems, and leave companies worse off.

Generative AI is growing explosively across knowledge work, particularly in software development. OpenAl's GPT-4.1 focuses heavily on enhancing coding capabilities and is a step toward full automation. Organizations adopting these tools are anticipating major gains. And early research supports their optimism: GitHub has reported that programmers using Copilot are up to 55% more productive, and McKinsey has found that developers can complete tasks up to twice as fast with generative Al assistance.

But these positive indicators come with a major caveat. The studies were conducted in controlled environments where programmers completed isolated tasks — not in real-world settings, where software must be built atop complex existing systems. When the use of Al-generated code is scaled rapidly or applied to brownfield (legacy) environments, the risks are much greater and much harder to manage. As part of our ongoing research on the strategic management of Al-augmented software development, we conducted interviews with individuals involved in developing software — ranging from junior developers to lead software engineers and CIOs — across a diverse set of industries, including insurance, web hosting, social media, defense, management consulting, and fintech. Drawing on insights from these interviews, a review of the trade press, and our own economic modeling, we have identified several strategic trade-offs that companies should consider when adopting generative Al for software development.

Why Technical Debt Grows Faster With AI