Other comments have pointed out that packages on PyPi might not be the best metric and posted other countering evidence like spikes in GitHub contributions or m

Other comments have pointed out that packages on PyPi might not be the best metric and posted other countering evidence like spikes in GitHub contributions or mobile app submissions or even mobile app revenue. However I think open source package numbers are still worth watching as an inverse measure of AI adoption. That is, I expect the numbers (at least the frequency of downloads, if not the number of new packages) to go down over time as AI makes generating functionality easier than hunting down and adding a dependency. The number of new packages could still go up as people may still open-source their generated code, for street cred if not actual utility. But it's not clear how much of those incentives apply if the code is not very generally useful and the effort put into is minimal.