Researchers at the University of Oxford developed a method to identify when generative AI 'hallucinates' by making up incorrect answers.
The advanced communication nature of AI models can present false information as facts, causing concerns, particularly with serious topics like healthcare and law.
The University of Oxford's research distinguishes between AI models being certain about answers or 'making something up'; Dr. Farquhar emphasizes the difficulty in discerning certainty.
Collection
[
|
...
]