If the purpose of a metric is to show adoption, the metric can be defined in a way to show adoption. Could be just an effect of promo driven culture, AI push and good’ol Goodhart’s law.
Like, how do you even measure when code is ai authored and when not. If you insert 25% of a variable name and the autocompleter guesses the rest of the name correctly, are the remaining 75% AI generated?
If the purpose of a metric is to show adoption, the metric can be defined in a way to show adoption. Could be just an effect of promo driven culture, AI push and good’ol Goodhart’s law.
Like, how do you even measure when code is ai authored and when not. If you insert 25% of a variable name and the autocompleter guesses the rest of the name correctly, are the remaining 75% AI generated?