If you don't have a filtering mechanism moderators will burn out and more interesting projects may never be seen. If you've run a bunch of LLM agent sessions concurrently you run into the human review bottleneck problem. That's all this is on a larger scale.
Agreed on the human bottleneck part, I've faced it myself too. but outright banning projects just because AI was used seems wild. I've seen genuinely useful, projects with pretty well thoughtful architectures built using AI. There needs to be some actual consideration of quality rather than just how it was coded. Reddit already has moderation tools and multiple mods, surely better filtering/review mechanisms or some sort of ai scoring could be worked out instead of blanket bans.