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.
AI models often fail to identify ableism across cultures
