Authors: Rafaela Martelo, Ruo-Qian Wang

Published on: March 05, 2024

Impact Score: 8.0

Arxiv code: Arxiv:2403.03188

Summary

• What is new: Introduces a GPT-4 powered AI Assistant for real-time flood forecasting to facilitate communication among decision-makers, public, and forecasters.
• Why this is important: The complexity of numerical flood models complicates decision-making and public understanding of flood risks.
• What the research proposes: A GPT-4 Large Language Model powered AI Assistant that provides immediate flood alerts, interprets flood maps, and offers risk management advice.
• Results: The prototype demonstrated effective communication of flood risks by evaluating its performance on relevance, error resilience, and context understanding.

Technical Details

Technological frameworks used: GPT-4

Models used: GPT-4 for natural language understanding and function calling

Data used: Real-time flood warnings, flood maps, social vulnerability data

Potential Impact

Disaster management services, emergency response technology providers, insurance companies, and city planning authorities

Want to implement this idea in a business?

We have generated a startup concept here: FloodGuard AI.