A METR study revealed that experienced developers using AI coding assistants took 19% longer to complete tasks compared to those who did not use them. Despite the significant increase in time taken, developers using AI believed they had improved their productivity by 20%. The study involved 16 developers well-versed in open-source projects, observing that those without AI tools spent more active coding time, while AI-assisted developers spent more time reviewing, prompting, and waiting for AI responses. This unexpected outcome has raised questions about the effective utilization of AI in development workflows.
Experienced developers using AI coding tools took 19% longer to complete issues than those not using generative AI assistance. Despite longer task completion times, AI-assisted developers believed their productivity had increased by 20%.
The study recruited 16 developers experienced with large, open-source repositories. Participants were assigned to use AI assistance or not, leading to different coding times and productivity perceptions.
Collection
[
|
...
]