A handful of dominant AI companies have been quietly manipulating one of the most influential public leaderboards for chatbot models, potentially distorting perceptions of model performance and undermining open competition, according to a new study.
1 minute read