This last week I was watching some YouTube, as you do, and I came across a video by ThePrimeTime about this very topic I’m researching right now! In it, he goes over some research from GitClear along with the person who wrote the report.

Is it a good report though?

I don’t know if it’s good or not. I’m not familiar with GitClear and it’s interesting at least that their CEO would be creating the report. However it doesn’t contradict what I’ve seen so far in any great way. At least not in any way that my stupid brain can recall. So I would say that, at the very least, it’s worth taking note of for further study.

What does it say?

In the report they are seeing a rise in code churn and lines of code added with a corresponding decrease in lines of code removed. This, they claim, may indicate that AI is duplicating code rather than using common code elements. In doing so, this would reduce code readability and maintainability. These are not good attributes for software. The author doesn’t go so far as to say that LLMs need to be removed from the development process, but they do say that we will want to take this into account and put systems in place to prevent this code duplication.

Conclusion

So overall I would say, assuming this report is accurate, that LLM usage in software development is not as ideal for long-lived systems. Perhaps for standing up a new project or prototyping one could make good use of an LLM, but once the system is in place that either LLM usage is reduced or additional steps are taken to reduce code duplication, churn, and to improve removal of dead code.