A new study has found that artificial intelligence systems struggle to distinguish between objective facts and subjective beliefs. This limitation poses risks as these technologies enter high-stakes fields like medicine and law.