Be careful with how you consider this article, it doesn’t have as much confirmation bias as the title would lead you to believe.
There are a few nuggets in this article that I think should be considered:
- The open source projects that were tested have an average age of 10 years, the article states that AI code assistance is better for new projects and less efficient when used with existing projects.
- The projects used have strict style guidelines.
- The developers were very well versed on their projects, meaning they were extremely efficient and code changes.
- the article itself claims that AI is probably better utilized by newer developers. Of course experienced devs who have years tied into their projects are going to be very well equipped to execute on their projects.
Big shocker. AI isn’t a panacea. FOMO is ruining the industry (even before AI was a thing)
<shocked pikachu.jpg>
(Fun fact, it’s an official emoji on Lemmy-UI. Start typing
:pikachu:and select it from the drop-down. Doesn’t work on mobile UIs.)I promise to use it for good. And some evil.
:pikachu:
Great read, thanks RSS Bot! This is a fair and nuanced take, and seems to suggest that senior devs aren’t sped up, especially in big open source projects where they have years of domain expertise. But for small and new repos, less experienced devs, and well designed plans - it’s still probably a net win
Of course. It’s a new tool and too many people think everything is a nail to its hammer. Any tool used improperly will hinder productivity. Hell, any tool that has any amount of learning curve will hinder productivity to some extent.
So, the question that people should be seeking an answer for is “what are the actual nails for this particular hammer?”
