Table 3. Phase 1 and 2 outcomes from LIWC-22 analysis, reported by percentile.
| Phase | Words analyzed, n | Authenticity, % | Emotional tone, % |
|---|---|---|---|
| Phase 1 | |||
| Bard | 606 | 25.67 | 76.54 |
| Bing Chat | 406 | 38.44 | 37.99 |
| ChatGPT 3.5 | 500 | 27.35 | 64.36 |
| Phase 2a | |||
| ChatGPT 3.5 | 2198 | 60.03 | 63.21 |
| Claude | 1126 | 46.76 | 57.74 |
| Google Gemini | 461 | 24.49 | 79.99 |
| Microsoft CoPilot | 990 | 32.29 | 32.09 |
The number of words analyzed in phase 2 included the additional questions.