Software Development AI Gains Are Real, But Governance Lags Behind

The report, based on 578 survey responses from engineering and product leaders, confirms that AI is now a fixture in the software development lifecycle. Roughly 84% of teams utilize AI during the build phase, with 83% of respondents noting a meaningful reduction in shipped code defects. However, this shift has created a paradoxical bottleneck: 67% of developers acknowledge that AI-generated code requires more rigorous testing than traditional human-written code, effectively shifting the workload from raw creation to validation and oversight.

Brian Jackson, principal research director at Info-Tech, warns that speed should not be mistaken for discipline. Currently, only 37.4% of teams report having formal AI governance procedures in place. This lack of structure, compounded by the fact that 51% of developers cite legacy code as a major friction point, leaves many organizations vulnerable to unmanaged risks. As AI adoption scales, the study suggests that veteran developers are increasingly pivoting toward code review roles, underscoring the need for companies to formalize production-readiness criteria rather than relying on ad hoc implementation.