The rising use of AI coding assistants like GitHub Copilot might suggest an uptick in productivity; however, they may actually be contributing to a decline in the overall quality of code.

This is the primary conclusion of a recent analysis from the software engineering platform GitClear. The report examined 211 million lines of code spanning from 2020 to 2024. GitClear’s findings revealed a significant drop in code reuse over the past year, raising concerns since reusing code is a fundamental practice for enhancing system reliability.

Recent surveys indicate that the performance of AI coding assistants is often inconsistent.

A survey conducted by software provider Harness found that many developers spend additional time addressing bugs and security issues in AI-generated code compared to their human-written counterparts. Meanwhile, a report from Google noted that while AI can speed up code review processes and enhance documentation, it may compromise the stability of deliveries.