I'm looking for some insights on how to effectively review pull requests (PRs) and merge requests (MRs) when a significant portion of the code has been generated using AI. Honestly, I've never been too enthusiastic about code reviews, especially when they involve long changes across multiple files. With AI's rise, the situation feels even more daunting. How are others in the same boat handling this? Also, if you have recommendations for articles or resources discussing this topic, I'd greatly appreciate it!
1 Answer
Are you approaching this in a workplace or open-source context? In a workplace, someone higher up should ideally have a policy in place regarding AI contributions. For open source, it can be more of a conversation among contributors—ranging from avoiding AI contributions due to review complexity to only accepting AI code that's already been thoroughly validated by the submitter.
It's in the workplace for me, and I feel like the quality of the code is slowly decline. AI makes it hard to keep track of all the changes, especially since others might approve the reviews too. It's tough to find the motivation to do thorough reviews and catch when AI is going off track without getting bogged down in minutiae.