Hmm. I hear where you’re coming from, though it seems optimal to address the causes - pressure to hype, toxic competitiveness, etc. - than the symptoms. It’s hard to believe LLMs wouldn’t (as you say) duplicate the hype bias, and harder to believe pressured researchers wouldn’t simply egg them on.