SENTINEL Framework Reduces Hallucinations In Multimodal Large Language Models

SENTINEL Framework Reduces Hallucinations In Multimodal Large Language Models

Researchers developed a new system, called SENTINEL, that substantially reduces fabricated content in image-captioning artificial intelligence by identifying and correcting errors at the very beginning of the text generation process, achieving over 90% reduction in hallucinations and outperforming existing methods.

Focussed language models resolve Gen AI hallucinations

Last year, “hallucinations” produced by generative artificial intelligence (Generative AI [GenAI]) were in the spotlight in court, in court again, and certainly, all over the news. More recently, Bloomberg News said that in their 2024 annual reports, “Goldman Sachs Group Inc., Citigroup Inc., JPMorgan Chase & Co. and other Wall Street firms are warning investors about new risks from the increasing use of artificial intelligence, including softwa…