Opinion: AI adoption may widen existing engineering blind spots

The piece identifies five specific blind spots where traditional ticket-based systems fail to capture AI-driven work: velocity theater, review debt, hidden work, quality drift, and unproven spend. For example, a Stanford case study mentioned in the text found that AI adoption lifted PR count by 14% but caused code quality to drop by 9%, with variance more than tripling. Additionally, a survey of engineering leaders found that 80% of teams spend at least 10% of their time on code review, with roughly one in ten spending more than 40%.

The author contends that engineering leaders must move beyond intuition and anecdotal evidence to prove AI ROI to finance and leadership. By observing actual codebase changes and deployment frequency against historical baselines, teams can distinguish between the "tuition cost" of learning AI and genuine systemic failures. The core argument is that leaders must first understand their current blind spots before investing further in AI tools.