Table 2. Comparison of APS, BS, and “highest perplexity of a sentence” for human-written versus ChatGPT-generated manuscripts.
Average Perplexity Score | Burstiness Score | Highest perplexity of a sentence | |
Human manuscript | 140 | 306 | 1912 |
ChatGPT response, 11th iteration | 55 | 55 | 265 |
ChatGPT response, 6th iteration | 42 | 40 | 170 |