logoalt Hacker News

Systemerror7A69today at 6:37 AM0 repliesview on HN

It's also relying on the assumption that the checking LLM only ever corrects wrong statements and never incorrectly "corrects" an already correct statement, which might not always be the case as well.