Table 2 Analysis of conversations with the chatbot impersonating a patient
From: The feasibility of using generative artificial intelligence for history taking in virtual patients
No | Word Counts (Sentences) | Instances of implausible answers by chatbot (words / uttterances, %) * | ||||
|---|---|---|---|---|---|---|
Questions by Student Doctor | Answers by Chatbot | Inarticulate words | Halluci-nations | Missing important information | Total | |
1 | 301 (24) | 234 (23) | 1 | 5 | 0 | 6 |
2 | 244 (21) | 389 (56) | 4 | 0 | 1 | 5 |
3 | 185 (17) | 387 (50) | 11 | 0 | 1 | 12 |
4 | 211 (18) | 132 (19) | 7 | 0 | 1 | 8 |
5 | 242 (16) | 183 (19) | 0 | 2 | 1 | 3 |
Total | 1,183 (96) | 1,325 (177) | 23 (1.7%) | 7 (0.5%) | 4 (0.3%) | 34 (2.6%) |