Skip to main content

Table 2 Comparison of response characteristics between GPT-4o and Claude-3

From: The future of AI clinicians: assessing the modern standard of chatbots and their approach to diagnostic uncertainty

 

GPT-4o

Claude-3

P Value

Characteristics

   
 

Response time (sec), mean (95% CI)

12.4 (9.3–15.3)

24.0 (21.0-32.5)

< 0.01

 

Response length (characters), mean (95% CI)

1,596 (1,395.0–1,705.0)

2,001 (1,845-2,212)

< 0.01

 

Rationale for other answer options, N (%)

71 (78.9)

78 (86.7)

0.17

Reason for error, N (%)

n= 48

n= 38

 
 

Logical error

30 (62.5)

17 (44.7)

0.02

 

Statistical error

9 (18.8)

6 (15.8)

0.72

 

Information error

9 (18.8)

15 (31.6)

0.04