AI is helping coders save time — but it’s leading to more slop content than ever before | Daily Reports Online
- Coding is more accessible and developers are submitting more lines of code than ever
- Reviewers face more work without a clear understanding of whether AI was involved
- The Linux Foundation calls for “governance, transparency and trust mechanisms”
A new Linux Foundation report has confirmed what we’ve all been thinking – while AI has proven to cut code writing time down considerably, it’s not actually improving the quality.
Researchers found that artificial intelligence can reduce the amount of time it takes to write code by a staggering 52%, and yet a quarter (25%) of pull requests are still being described as low-quality ‘AI slop’.
While there are clear benefits (71% of maintainers agree AI has helped to lower the barrier to entry), AI is ultimately pushing problems further downstream instead of offering a net boost.
Latest Videos FromTechRadar
AI-generated code doesn’t mean faster net code generation
While AI allows developers to generate code at a quicker rate, maintainers still need to verify its quality and security, but with more code being written, they’re facing more human reviews than ever.
The technology is starting to prove its usefulness in code review for some easier tasks, however one in five (21%) respondents actually said using AI increases the time needed for more complex tasks.
AI-generated reports have already been seen to generate false positives and duplicated submissions, wasting human reviewers’ time. Developer burnout used to be a topic we heard a lot about a few years ago, but this is now shifting to maintainer burnout.
The Linux Foundation also noted difficulties in handing code over to a reviewer from a writer – just three in five (61%) say they conduct detailed code reviews before submitting it, and not all developers actually disclose their use of AI to their next-in-chain.
Ultimately, the report acknowledges that AI is here to stay within coding, with Research SVP Hilary Carter calling for projects to “[set] clear expectations for GenAI, putting responsibility for testing and disclosure on contributors and giving maintainers better tools to manage incoming pull requests.”
“The answer is not to slow innovation, but to build the governance, transparency and trust mechanisms that allow AI-assisted contributions,” AI & Infrastructure GM Mark Collier added.
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds.
