Skip to main content
. 2024 Jun 3;2:51. doi: 10.1038/s44271-024-00091-8

Fig. 7. Test Retest reliability of ChatGPT and GPT4 and open source models.

Fig. 7

a CRT results (i.e., correct and intuitive response rate) for ChatGPT and GPT-4 at two different time pints (March – June 2023). b Linda/Bill results (correct response rate for the Linda and Bill items) for ChatGPT and GPT-4 at two different time pints (March–June 2023). c CRT results for four open access models. d Linda/Bill results for four open access models. For visual reference, dotted colored lines represent the performance of DV3 model.