The increasing popularity of AI coding assistants, such as GitHub Copilot, may seem to be enhancing productivity, but in actuality, they may be contributing to a decline in overall code quality.

This is the primary conclusion drawn from a recent study published by GitClear, a software engineering platform, which examined 211 million lines of code from 2020 to 2024. The analysis by GitClear revealed a significant decrease in code reuse in the previous year, a trend that raises concerns given that code reuse is a common practice for building redundant systems.

Several surveys conducted recently have yielded inconsistent results regarding the effectiveness of AI coding assistants.

A survey by Harness, a software vendor, discovered that the majority of developers spend more time debugging AI-generated code and addressing security vulnerabilities compared to code written by humans. Conversely, a report by Google found that AI can accelerate code reviews and improve documentation, albeit at the expense of delivery stability.