The artificial intelligence models underlying popular chatbots and content moderation systems struggle to identify offensive, ableist social media posts in English—and perform even worse in Hindi, new Cornell research finds.
This post was originally published on this website.
