Using a Large Language Model to Choose Effective Climate Change Messages

Thomas Benchetrit, Iris Kremer, Erik Hemberg, Aruna Sankaranarayanan, Una-May O'Reilly

Track Selection: My submission fits into both tracks and I will let the PC decide if it is accepted.

Keywords: large language model, climate change, computational argumentation, communication strategy, effective messaging

Abstract:

Prior research has demonstrated that climate change communication is
an effective way to increase public understanding and engagement.
However, an effective communication strategy can require an
extensive longitudinal study to segment an audience and conceive of
messages that might convince them. We assess the capability of
GPT-3.5-Turbo to associate an individual with a group and choose an
effective climate change message based on survey information about
an individual and prompt guidance. We observe that it, with a
significant bias, can match individual profiles to groups and select
messages based on an individual profile.

Submission Number: 7