Table 4. Overall performance metrics from Scenario 1: Original vs. AI-generated abstracts.

Metric	GPTZero (%)	ZeroGPT (%)	DetectGPT (%)
Accuracy	97.22	64.35	54.63
- By author categories
- Native	99.07	64.81	58.33
- Non-Native	97.22	63.89	50.93
- By disciplines
- Technology & Engineering	100.00	61.11	56.94
- Social Sciences	98.61	66.67	63.89
- Interdisciplinary	95.83	65.28	43.06
False Positive Rate (FPR)	0.00	16.67	31.94
- By author categories
- Native	0.00	19.44	27.78
- Non-Native	0.00	13.89	36.11
- By disciplines
- Technology & Engineering	0.00	12.50	41.67
- Social Sciences	0.00	12.50	12.50
- Interdisciplinary	0.00	25.00	41.67
False Negative Rate (FNR)	2.78	45.15	52.08
- By author categories
- Native	2.78	43.06	48.61
- Non-Native	8.33	47.22	55.56
- By disciplines
- Technology & Engineering	2.08	52.08	43.75
- Social Sciences	4.17	43.75	47.92
- Interdisciplinary	2.08	39.58	64.58
False Accusation Rate (FAR)			44.44
- By author categories
- Native			44.44
- Non-Native			44.44
- By disciplines
- Technology & Engineering			45.83
- Social Sciences			25.00
- Interdisciplinary			62.50
Majority False Accusation Rate (MFAR)			4.17
- By author categories
- Native			2.78
- Non-Native			5.56
- By disciplines
- Technology & Engineering			8.33
- Social Sciences			0.00
- Interdisciplinary			4.17