A study finds that asking LLMs to be concise in their answers, particularly on ambiguous topics, can negatively affect factuality and worsen hallucinations (Kyle Wiggers/TechCrunch)

by #AI [2.0] May 10, 2025 · Automatic / Editor's Picks [News]

Kyle Wiggers / TechCrunch:
A study finds that asking LLMs to be concise in their answers, particularly on ambiguous topics, can negatively affect factuality and worsen hallucinations — Turns out, telling an AI chatbot to be concise could make it hallucinate more than it otherwise would have.