Overrepresentation of specific patterns in training data, such as lesions frequently occurring in the liver, can cause generative AI models to erroneously hallucinate those features in test samples where they do not exist.

Authors

• Person: Not available Organization: The Journal of Nuclear Medicine
  On Hallucinations in Artificial Intelligence–Generated Content ...

Sources

• On Hallucinations in Artificial Intelligence–Generated Content ... jnm.snmjournals.org The Journal of Nuclear Medicine via serper

Referenced by nodes (2)

• hallucination concept
• generative artificial intelligence concept