Table 5 Classification accuracy scales for human-written and AI-generated texts

Human-written (NEGATIVE) text (docs 01-Hum & 02-MT), and the tool says that it is written by a:

 [100—80%) human

True negative

TN

  [80—60%) human

Partially true negative

PTN

  [60—40%) human

Unclear

UNC

  [40—20%) human

Partially false positive

PFP

  [20—0%] human

False positive

FP

AI-generated (POSITIVE) text (docs 03-AI, 04-AI, 05-ManEd & 06-Para), and the tool says it is written by a:

  [100—80%) human

False negative

FN

  [80—60%) human

Partially false negative

PFN

  [60—40%) human

Unclear

UNC

  [40—20%) human

Partially true positive

PTP

  [20—0%] human

True positive

TP