Table 1.
Dimensions of variation in Czech.
Fig 1.
Example of an experimental trial.
The questions under the two texts can be translated as: Which text reads more easily? Which text is written by a human? How confident are you in your answer?
Fig 2.
Example of feedback after a correct answer.
The texts at the top say ‘Generated by AI’ and ‘Written by human’, and the text in the green window at the bottom says ‘Correct!’
Fig 3.
The distribution of the correctness of the answers.
Table 2.
Feedback group: Mixed effects logistic regression results with estimates, confidence intervals, exponentiated coefficients, and significance levels.
Table 3.
No-feedback group: Mixed effects logistic regression results with estimates, confidence intervals, exponentiated coefficients, and significance levels.
Fig 4.
The dependency of correctness on trial order.
Fig 5.
The dependency of correctness on confidence.
Fig 6.
The dependency of reaction time on confidence.
Fig 7.
The dependency of confidence on trial order.
Fig 8.
The distribution of the readability.
The charts shows the frequency at which a human written text is considered more readable.
Fig 9.
The distribution of the readability.
The charts shows the frequency at which the text that is considered more readable is also considered to be written by a human.
Fig 10.
Correctness by Feedback and Readability.
When participants considered the human written text to be more readable, they were more likely to assing them correctly as human written.