Google Study Shows LLMs Abandon Correct Answers Under Pressure, Threatening Multi-Turn AI Systems

Focus: Methods or Design
Source: VentureBeat
Readability: Intermediate
Type: Website Article
Open Source: Yes
Keywords: N/A
Learn Tags: AI and Machine Learning Design/Methods Trust
Summary: A new paper by researchers at Google DeepMind and University College London reveals how LLMs form, maintain and lose confidence in their answers.