Skip to main content
. 2024 Nov 5;10:20552076241297127. doi: 10.1177/20552076241297127

Figure 3.

Figure 3.

Reasoning (A) and errors (B) of GPT-3.5 and GPT-4. Neurological reasoning for correctness is a five-point Likert scale (1 = poor, 5 = excellent).