Noisy
← this week  ·  Subscribe →

The world is being quietly rearranged by people who write very long documents.

arXiv · May 13, 2026
The title they went with Large Language Models for Causal Relations Extraction in Social Media: A Validation Framework for Disaster Intelligence Noisy translates that to

AI can find disaster causes in social media, but also invent them

What happened
Researchers tested if large language models could find cause-and-effect relationships in social media posts during disasters. It turns out the AI can identify real connections, but it also makes up connections that are not supported by evidence.
Why it matters
During a disaster, knowing what causes what is critical for emergency response. If AI tools are used to help make these decisions, emergency managers need to know if the information is reliable. This paper shows that AI can be helpful, but it also introduces a new risk: the AI might hallucinate causes that don't exist, leading to wasted resources or bad decisions.
The signal
Watch for emergency response agencies to publish guidelines on how to validate AI-generated insights from social media, or for new tools designed to cross-reference AI findings with official reports.
If you insist
Read the original →