When AI-produced code goes bad

The same generative artificial intelligence tools that are supercharging the work of both skilled and novice coders can also produce flawed, potentially dangerous code. Multiple studies have shown that more than half of programmers are using generative AI to write or edit the software that runs our world—and that number keeps rising. In 2022, GitHub found that developers who used its AI coding assistant worked 55 percent faster than those who didn't. But the productivity gains come with a price. One study from Stanford found that programmers who had access to AI assistants "wrote significantly less secure code than those without access to an assistant." Another study from researchers at Bilkent University in 2023 found that 30.5 percent of code generated by AI assistants was incorrect and 23.2 percent was partially incorrect.