Larger AI chatbots often give incorrect answers over admitting uncertainty, study shows
A study of recent, larger versions of three major AI chatbots reveals that they are more likely to provide incorrect answers than to admit their mistakes. The results, published in Nature on Wednesday...