New Plan: Pay to Prove Your AI Content Isn't Junk

The internet is about to be flooded with AI-written articles, comments, and research. How do we find the good stuff without banning all AI or hiring an army of human moderators? A new proposal suggests a pay-to-post system. Users pay a small fee, and AI moderators (like Claude or ChatGPT) review their content. The more you pay, the more thorough the review. A $1 post might only be seen by people who opt into 'the dregs.' A $20 post gets a jury of AIs checking it against a rubric. A $200 post gets a serious replication attempt.

The front page of this forum would show only the highest-rated posts. Lower tiers are linked but not promoted. Human moderators then step in to evaluate the best AI-approved content. These humans are paid by the hour, and their reputation grows when their promotions turn out to be good. This mixes economic incentives with social status to keep quality high.

The idea came from a discussion about AI alignment (making sure AI does what humans want). One person asked Claude to critique the idea, and Claude said it was bad unless the verification problem was solved. So the author focused on verification: how do you prove something is good when there's too much content for humans to check? The answer: let AI do the first pass, and let users pay for that AI time.

This isn't just for AI researchers. Any online community could use this model to fight spam and low-effort content. It could also help experts share knowledge without being buried by AI-generated noise. The catch? It might favor those who can pay, and AI moderators can still be fooled. But in a world of infinite AI content, some form of paid verification might be the only way to keep the internet useful.