Maintainable Code Is About Responsibility, Not AI
Criticism around “unmaintainable AI code” is growing louder as coding assistants become commonplace. In her blog post, Anja Schirwinski, CEO and co-founder of undpaul, argues that the debate often misidentifies the problem. The issue, she suggests, is not the tool itself, but how teams define responsibility, standards, and review processes around its use.
Anja acknowledges valid concerns: hallucinated assumptions, architectural inconsistencies, security risks, and blind copy-paste practices. But she points out that poorly written code long predates AI. Under time pressure and without discipline, teams have always created technical debt. AI, in her view, simply amplifies what already exists—strong engineering cultures benefit from it, while weak ones expose their gaps more quickly.
Rather than framing AI as a threat to craftsmanship, she encourages leadership to treat it as a fast junior developer: useful, but always reviewed. The strategic risk is not AI-generated code itself, but unclear ownership and hidden usage without standards. For organisations deciding how to adopt AI responsibly, the article makes a broader argument: maintainability is a matter of professional accountability, not technological capability.


