Vectara Introduces Hallucination Corrector for Enterprise AI

Vectara has launched a new Hallucination Corrector to improve the reliability of enterprise AI systems, announced in a press release. This tool is designed to detect and mitigate unreliable responses from AI models, a common issue known as hallucinations.

Hallucinations occur when AI models provide false information with confidence. Vectara's Hallucination Corrector works alongside its Hughes Hallucination Evaluation Model (HHEM) to reduce these occurrences. The HHEM scores AI-generated responses against source documents to determine accuracy, and the Corrector provides explanations and corrections for any inaccuracies.

In initial tests, the Hallucination Corrector reduced hallucination rates in enterprise AI systems to about 0.9%. This feature is integrated into Vectara's platform and offers various options for users, including automatic corrections and detailed explanations for expert analysis. The Corrector aims to meet the high accuracy standards required in regulated industries such as finance, healthcare, and law.