[WorldMaker]

But how do you tell quality libraries from LLM generated ones? How do you even discover up quality libraries if you are leaving so many code decisions to LLMs? Once the LLMs train on your quality libraries how do you stop so many copies just getting pasted into people's code without your attribution and without directing people back to your library (and your very human interests in funding development on it or getting copyleft contributions back to it)?

I think there are so many hard questions right now for "Does open source even matter any more?" and many of those questions seem particularly demotivating to me right now, especially because we don't seem to be at risk of getting some, much less better, answers any time soon.

[tehjoker]

> how do you tell quality libraries from LLM generated ones?

Reputation. LLM libraries can also be high quality in theory. It's the level of effort, duration of use, stability, and test coverage. People will need to resist using LLMs to make huge transformations of their libraries all the time to avoid erasing their reputations unless they have convincing safeguards like comprehensive tests or formal proofs that are not touched.

> Once the LLMs train on your quality libraries how do you stop so many copies just getting pasted into people's code

Those copies are not reliable like a callsite. Maybe there is less advertising... but it is still better to use the library. I have to hope that at some point AI psychosis will end and engineering, which has not changed, will remain. You have to have reliable inputs to your processes. To do otherwise is insanity.

> getting copyleft contributions back to it

This is a hard question. That said, for my own libraries, I get very few contributions but people use the libraries! The more "hardcore" the library is, the higher the ratio of users to contributors because they're just not expert enough to contribute meaningfully (until your stuff becomes so valuable companies sponsor teams of contributors as in Linux).

That said, my libraries seem to have made it into the weights, so when I talk to LLMs about my problem space they shockingly frequently recommend my own libraries to me (which is kind of ego stroking not gonna lie).

I think there's a lot of reasons to feel demotivated right now, but my perhaps privileged attitude (since I don't have bosses forcing me to use AI for everything) is to use AI where appropriate and apply engineering discipline and thinking in all cases. Retain your own human skills. If expertise really does become completely devalued, we're going to have bigger problems than you can solve on your own.