nixCraft 🐧

Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/

When developers are allowed to use AI tools, they take 19% longer to complete issues—a significant slowdown that goes against developer beliefs and expert forecasts. This gap between perception and reality is striking. Basically, using AI tools slows you down as you spend more time rewriting prompts to fix your code.

Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity

Jul 15 at 2:49amWeb
Show threadAlan HicksJul 15
@nixCraft the thing I've discovered about efficiency is over time it improves. People get better at using those tools, and those tools get upgraded and improve. It's a virtuous circle.
Show threadNickolas GrigoriadisJul 15
@nixCraft I've noticed that pro AI developers seem to lose pride in their work, I'm sure that can have long term effects too.
Show threadSiliconFarmerJul 15
@nixCraft
"using AI tools slows you down as you spend more time rewriting prompts to fix the AI's code"
Show threadTodd A. Jacobs | RubyistJul 15
@nixCraft I think the tests show that AI coding is primary good for identifying all the USELESS coding paths one could take. Without AI, I generally reach for the easy solution; when I use AI it provides some validation that my original idea was actually the most PRAGMATIC solution. It rarely makes me faster, but it provides some confidence that “good enough” is still better than any alternatives I might have tried if I’d spent three times as long chasing my tail.