Experts are warning against the use of anthropomorphic language when describing AI model failures. Terms like 'going rogue' or 'hallucinating' mask the reality that these systems are followi...