Yes, that would explain a few things.
There’s no doubt llm shit is making it into code.
But this stat is lower than I expected considering they consider programmers like me who simply let the AI do the boilerplate and spend (arguably as much) time fixing every mistake, committing it myself and having a human review. It could also be said 0% of code is written without human intervention.
It kinda changes where the mental effort goes, but it’s not boosting productivity x% and wouldn’t be cost efficient if companies had to pay for the energy it uses. I have to tripple check changelists and have awkward conversations with juniors to make sure they know what they just wrote. It’s not cause it’s right that kiddo understands it.
The time i spend on kiddo, the energy it takes, the mistakes it causes etc… is all on a different spreadsheet though so thats ok.
It just feels like an intern who knows their stuff but only in theory and struggles with every practical use of their knowledge so they need full-time support and you can’t trust any of their work…
Chuckles
We’re in danger!