ChatClimate: Grounding conversational AI in climate science

Large Language Models (LLMs) have shown remarkable progress in question-answering, but challenges like hallucination and outdated information persist. This is particularly critical in climate science, where accurate and timely information is vital. This paper introduces ChatClimate, a conversational AI prototype that enhances GPT-4 by integrating the Sixth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC AR6). Evaluations by experts demonstrate that a hybrid ChatClimate approach, combining IPCC AR6 data with GPT-4's internal knowledge, produces more accurate responses than GPT-4 alone, showcasing the effectiveness of grounding LLMs in authoritative scientific sources.

Saeid Ashraf Vaghefi, Dominik Stammbach, Veruska Muccione, Julia Bingler, Jingwei Ni, Mathias Kraus, Simon Allen, Chiara Colesanti-Senn, Tobias Wekhof, Tobias Schimanski, Glen Gostlow, Tingyu Yu, Qian Wang, Nicolas Webersinke, Christian Huggel, Markus Leippold