As they acquire in recognition, AI coding assistants reminiscent of GitHub Copilot might seem like boosting productiveness. However in actuality, they may very well be inflicting total code high quality to say no.
That’s the top-line discovering from a new report launched by software program engineering platform GitClear, which analyzed 211 million code strains from 2020 to 2024. In keeping with GitClear’s evaluation, there was a exceptional decline in code reuse final 12 months — a possible trigger for concern, provided that code reuse is a typical observe to assist construct redundant techniques.
A number of latest surveys have proven that AI coding assistants have a tendency to supply combined outcomes.
One from software vendor Harness discovered nearly all of devs spend extra time debugging AI-generated code and safety vulnerabilities in comparison with human-written contributions. A Google report, in the meantime, discovered that AI can quicken code critiques and profit documentation, however at the price of supply stability.
