A common argument against gen-AI bans in many circles, but especially in open source software is that it’s unenforceable.

Let’s look at laws! Laws don’t necessarily stop anything. In theory, you can still do the things they ought to ban or stop. You just risk jail time.

It is the usual argument of (right-) libertarians, who make up 99% of AI executives, against all forms of regulations, and their alternative is that people should just boycott the offending corporations, and make competing businesses that will give those big monopolies a big fat reality check. Unfortunately it often ends up in the coconut island situation instead, but that doesn’t mean every situation is always like that. We even have David vs. Goliath situations working out relatively well within the open source community.

The same people are trying to give us the same orders as they are giving it to lawmakers and people concerned about deregulation. “Use the AI better than the prompt kiddies.” “Use AI to speed up your progress.” “Use a self-hosted model.” “Read the output, maybe rewrite it completely to make it yours.” “Edit the generated image to make it non-slop.”

Yet if anything, I’ve seen just as many people eventually getting AI psychosis from moderated AI usage as from vibe coding and other stuff. Even seen “hand written, AI-assisted” contributions going off of the wrong specifications completely, essentially being no better than “actual” vibe coding. AI-assisted art usually barely above quality-wise the completely AI stuff.

We shouldn’t let these techbros get to our minds and open our commons to be destroyed (I swear that’s their end goal). Just because sometimes people can sneak some AI slop into stuff, we shouldn’t be okay with AI slop, nor that we need to find a usecase for AI by yesterday.