Figure 3
From: A comparison of human and GPT-4 use of probabilistic phrases in a coordination game
Discordance. We computed discordance, a measure of disagreement among each human observer and the remaining human observers and between GPT-4 and the human observers. See text. The left and right panels are boxplots of discordance values for the Investment Context and for the Medical Context, respectively. The top and bottom of the boxes mark the 75th and 25th percentiles for each context. The discordance for GPT-4 is marked by a solid red diamond in each context. The discordance for GPT-4 is below the median discordance (the solid red line segment) for the human participants in both contexts.
