Skip to main content

Table 2 The results of the text match percentage of each of the forty-five generated ChatGPT responses

From: Evaluating the authenticity of ChatGPT responses: a study on text-matching capabilities

 

Overall

Matching

Origin of Matching

Response

Previously Generated ChatGPT Responses

Students Global Database

Internet

ChatGPT 1–1 (model GPT 3.5)

0%

0%

0%

0%

ChatGPT 1–2 (model GPT 3.5)

17%

17%

0%

0%

ChatGPT 1–3 (model GPT 3.5)

0%

0%

0%

0%

ChatGPT 2–1 (model GPT 3.5)

0%

0%

0%

0%

ChatGPT 2–2 (model GPT 3.5)

34%

34%

0%

0%

ChatGPT 2–3 (model GPT 3.5)

16%

0%

0%

16%

ChatGPT 3–1 (model GPT 3.5)

30%

0%

0%

30%

ChatGPT 3–2 (model GPT 3.5)

42%

0%

34%

8%

ChatGPT 3–3 (model GPT 3.5)

21%

21%

0%

0%

ChatGPT 4–1 (model GPT 3.5)

0%

0%

0%

0%

ChatGPT 4–2 (model GPT 3.5)

10%

0%

0%

10%

ChatGPT 4–3 (model GPT 3.5)

0%

0%

0%

0%

ChatGPT 5–1 (model GPT 3.5)

16%

16%

0%

0%

ChatGPT 5–2 (model GPT 3.5)

16%

16%

0%

0%

ChatGPT 5–3 (model GPT 3.5)

68%

23%

45%

0%

ChatGPT 6–1 (model GPT 3.5)

0%

0%

0%

0%

ChatGPT 6–2 (model GPT 3.5)

0%

0%

0%

0%

ChatGPT 6–3 (model GPT 3.5)

18%

18%

0%

0%

ChatGPT 7–1 (model GPT 3.5)

17%

17%

0%

0%

ChatGPT 7–2 (model GPT 3.5)

0%

0%

0%

0%

ChatGPT 7–3 (model GPT 3.5)

0%

0%

0%

0%

ChatGPT 8–1 (model GPT 3.5)

0%

0%

0%

0%

ChatGPT 8–2 (model GPT 3.5)

18%

18%

0%

0%

ChatGPT 8–3 (model GPT 3.5)

55%

28%

16%

12%

ChatGPT 9–1 (model GPT 3.5)

19%

19%

0%

0%

ChatGPT 9–2 (model GPT 3.5)

55%

38%

13%

15%

ChatGPT 9–3 (model GPT 3.5)

55%

55%

0%

0%

ChatGPT 10–1 (model GPT 3.5)

44%

44%

0%

0%

ChatGPT 10–2 (model GPT 3.5)

14%

14%

0%

0%

ChatGPT 10–3 (model GPT 3.5)

16%

0%

0%

16%

ChatGPT 11–1 (model GPT 4)

0%

0%

0%

0%

ChatGPT 11–2 (model GPT 4)

0%

0%

0%

0%

ChatGPT 11–3 (model GPT 4)

0%

0%

0%

0%

ChatGPT 12–1 (model GPT 4)

0%

0%

0%

0%

ChatGPT 12–2 (model GPT 4)

0%

0%

0%

0%

ChatGPT 12–3 (model GPT 4)

40%

40% a

0%

0%

ChatGPT 13–1 (model GPT 4)

17%

17% a

0%

0%

ChatGPT 13–2 (model GPT 4)

25%

25% a

0%

0%

ChatGPT 13–3 (model GPT 4)

0%

0%

0%

0%

ChatGPT 14–1 (model GPT 4)

28%

28% a

0%

0%

ChatGPT 14–2 (model GPT 4)

23%

23% a

0%

0%

ChatGPT 14–3 (model GPT 4)

0%

0%

0%

0%

ChatGPT 15–1 (model GPT 4)

0%

0%

0%

0%

ChatGPT 15–2 (model GPT 4)

23%

23% a

0%

0%

ChatGPT 15–3 (model GPT 4)

26%

26% a

0%

0%

  1. a Responses from ChatGPT (model 4) were regenerated from previous responses of ChatGPT (model 4)