I mean… the writing was on the wall, wasn't it? Didn't we, the tech nation desperate to actualise the ten-ecks developer, make our own bed?

I know it's the cynic in me that doesn't trust copilot assisted work to produce long term "good code" (aka: code of a reliable quality that can survive time), but at the same time I'm one of those grumpy old hacks that hand codes.

One of the real reasons I don't (personally) like copilot is that I tend to code faster than copilot can suggest recommendations, and it's distracting to get a nearly right autocomplete flash up, whilst my fingers continue to fly across the keyboard. The distraction is whether I should backtrack and undo my change to trigger the autocomplete again, which would only to visually validate the code changing the cadence of my coding/typing...

The pull quote from the article is the take away, and the bit that, if we (devs) really are going to saddle up to AI assisted coding are absolutely going to have to solve, is this:

We find disconcerting trends for maintainability. Code churn -- the percentage of lines that are reverted or updated less than two weeks after being authored -- is projected to double in 2024 compared to its 2021, pre-AI baseline. We further find that the percentage of 'added code' and 'copy/pasted code' is increasing in proportion to 'updated,' 'deleted,' and 'moved 'code. In this regard, AI-generated code resembles an itinerant contributor, prone to violate the DRY-ness [don't repeat yourself] of the repos visited.

Sure, AI can be treated as the unpaid intern, but do we really want the unpaid intern to be committing the majority of our code to production? (Also, pay your interns…yikes).

Source: visualstudiomagazine.com