I created an annotation to keep track of what code...

# ai

[Horatio Thomas]

I created an annotation to keep track of what code I am using that is generated by AI

[Michal Harakal]

We probably need also an annotation for human written code too 😇

[Horatio Thomas]

Are we at the point where AI written code is more reliable than human written code?

[Horatio Thomas]

I posted this on reddit and someone said they hated it because it reminds them of the @author javadoc tag

[Michal Harakal]

There are many points of view on the topic, you brought the quality. Transparency can be another. Mine the first idea was more pragmatic: as an OSS maintainer should I refuse AI PRs completely or planning more time for code reviews etc .

[Anonymike]

I do like the RequiresOptIn nature of this in that it could be restricted to dev branches etc and not be enabled by default. The complication is that this requires honesty from the developer that they haven't thoroughly reviewed it 😂

Regarding the broader topic, my first thought is why is code that hasn't been thoroughly reviewed by a human ever entering a production codebase at all? We are definitely not at a point where I believe that's even remotely safe for anything serious (and yes, I use these tools every day), but even for non-serious code we're just bloating search results with junk that's not reviewed and likely not even its author's specs accurately. But hey, we can make some AI companies a lot of money, waste a lot of open source reviewers' time, and make crazy claims about what "we" or AI built, filled with red flags, on social media so I guess that's the goal for many these days. 🤷 🤦