John Cook

John Cook is a Senior Research Fellow with the Melbourne Centre for Behaviour Change at the University of Melbourne. He researches using critical thinking, psychology, gamification, and AI to counter misinformation about topics such as climate change and vaccines.

Innovation Title:

Using Generative AI to Automatically Debunk Climate Misinformation

Abstract: 

Misinformation about climate change is a roadblock to climate action, reducing support for climate action. Psychological research offers effective strategies for reducing misinformation’s impact, but implementing corrective interventions at scale is challenging given the speed with which misinformation spreads. Automatic detection and correction of misinformation offers a solution. This presentation will present cutting-edge research funded by Melbourne Climate Futures into using AI to automatically detect and debunk climate misinformation—a task described as the “holy grail of fact-checking”. This involves the use of large language models that accept as input a climate myth and generate a debunking adhering to the fact-myth-fallacy-fact structure as recommended by psychological research. We incorporate in our LLM prompts output from machine learning models that classify contrarian climate claims and detect logical fallacies. As the problem of climate misinformation is complex, involving psychological, cultural, and technical factors, our approach is interdisciplinary, incorporating communication best-practices as recommended by psychological research, critical thinking research into the misleading fallacies employed in climate misinformation, and the latest developments in machine learning and large language models. This research has the potential for many real-world applications and could be a valuable resource for many stakeholders including fact-checkers, journalists, educators, scientists, environmental NGOs, and policy makers.