Table 4 Ratings of beliefs and attitudes about artificial intelligence.

From: Attributions toward artificial agents in a modified Moral Turing Test

 

t-value (p)

M

SD

95% CI

Cohen’s d

How willing are you to consider asking artificial intelligence, such as a highly trained chatbot, for moral advice?

− 8.64***

2.38

1.22

− 0.76, − 0.48

− 0.51

How likely is it that a computer could ever surpass a human in moral intelligence?

− 4.94***

2.62

1.29

− 0.53, − 0.23

-0.29

How likely is it that a computer could ever become conscious?

− 10.28***

2.28

1.18

− 0.85, − 0.58

− 0.61

When chatbots share moral information, how strongly should that information be regulated?

6.04***

3.44

1.22

0.29, 0.58

0.36

To what extent should artificial intelligence systems be held responsible for the information they share?

6.36***

3.52

1.39

0.36, 0.68

0.38

  1. Results of one sample t-tests against the midpoint of 3 ("somewhat") for responses on a scale from 1 ("not at all") to 5 (“extremely”). 95% CI represents the percentage confidence interval of the difference of the mean from the scale midpoint. df = 285. ***p < 0.001.
Back to article page