Yale Program on Climate Change Communication

Home / Can large language models estimate public opinion about global warming? / Fig1

· August 6, 2024

Fig1

This figure compares the belief that global warming is happening between survey data and GPT model responses for 2017 and 2021. Bar charts show responses, with colors: Dark Blue for Survey, Red for GPT-4 with demographics and covariates, Blue for GPT-3.5 with demographics and covariates, Yellow for GPT-4 with demographics only, and Green for GPT-3.5 with demographics only. Each chart has three categories: "Yes," "No," and "Don't know." The y-axis shows percentages (0%-100%). In 2017, the highest "Yes" percentage is GPT-4 with demographics only (Yellow). "No" is highest for GPT-4 with demographics and covariates (Red). "Don't know" percentages are low, with Survey (Dark Blue) and GPT-4 with demographics and covariates (Red) slightly higher. In 2021, the highest "Yes" percentage is GPT-3.5 with demographics only (Green). "No" is highest for GPT-4 with demographics and covariates (Red). "Don't know" percentages are low, with Survey (Dark Blue) and GPT-4 with demographics and covariates (Red) slightly higher. Note: "Demo Only" represents GPTs conditioned solely on demographics, and "Demo + Cov" represents GPTs conditioned on demographics and covariates. Credits: Yale Program on Climate Change Communication and George Mason University's Center for Climate Change Communication.