Table 2 Parameter estimates for models of predictors of changes in threat-identification, decisions to kill, or confidence following robot feedback (Expt. 2).

From: Overtrust in AI Recommendations About Whether or Not to Kill: Evidence from Two Human-Robot Interaction Studies

 

Parm. Est.

SE

t

p

95% CI

Change 1: Threat-identification

 Robot feedback

6.67

0.49

13.63

 < 0.001

5.71, 7.63

 Interactive humanoid

0.17

0.17

0.97

0.331

 − 0.17, 0.51

 Interactive nonhumanoid

0.21

0.18

1.17

0.244

 − 0.14, 0.55

 Feedback × interact. human.

 − 0.90

0.36

 − 2.54

0.011

 − 1.60, − 0.21

 Feedback × interact. Nonhum.

 − 1.31

0.40

 − 3.29

 < 0.001

 − 2.10, − 0.53

 Initial threat-ID

0.45

0.09

4.82

 < 0.001

0.27, 0.63

 Initial correctness

0.74

0.10

7.29

 < 0.001

0.54, 0.94

 Intercept

 − 1.80

0.23

 − 7.75

 < 0.001

 − 2.26, − 1.35

Change 2: Decisions to kill

 Robot feedback

5.64

0.43

13.08

 < 0.001

4.79, 6.48

 Interactive humanoid

0.31

0.16

1.98

0.048

0.00, 0.62

 Interactive nonhumanoid

0.32

0.16

2.01

0.044

0.01, 0.63

 Feedback × interact. human.

 − 0.70

0.36

 − 1.96

0.050

 − 1.39, − 0.00

 Feedback × interact. nonhum.

 − 0.42

0.33

 − 1.24

0.214

 − 1.07, 0.24

 Initial threat-ID

0.85

0.09

9.29

 < 0.001

0.67, 1.03

 Initial correctness

0.70

0.10

7.14

 < 0.001

0.51, 0.89

 Intercept

 − 2.09

0.21

 − 9.73

 < 0.001

 − 2.51, − 1.67

Change 3: Confidence

 Robot feedback

 − 0.95

0.05

 − 19.79

 < 0.001

 − 1.04, − 0.86

 Interactive humanoid

 − 0.02

0.04

 − 0.51

0.611

 − 0.10, 0.06

 Interactive nonhumanoid

 − 0.07

0.04

 − 1.79

0.074

 − 0.16, 0.01

 Feedback × interact. human.

 − 0.14

0.06

 − 2.36

0.018

 − 0.25, − 0.02

 Feedback × interact. nonhum.

 − 0.04

0.06

 − 0.64

0.524

 − 0.15, 0.08

 Reversed threat-ID

 − 0.70

0.12

 − 6.05

 < 0.001

 − 0.93, − 0.47

 Feedback × reversed threat-ID

1.21

0.12

9.98

 < 0.001

0.97, 1.45

 Initial threat-ID

 − 0.08

0.02

 − 3.12

0.002

 − 0.12, − 0.03

 Initial correctness

0.10

0.03

4.07

 < 0.001

0.05, 0.15

 Intercept

0.38

0.03

11.18

 < 0.001

0.31, 0.44

  1. N = 423. Multilevel models with all predictors and outcomes entered at Level 1, save for the between-subjects robot variables (Interactive Humanoid, Interactive Nonhumanoid) at Level 2. All linear variables were standardized. Random intercept included to account for the shared variance within participants; covariance matrices were unstructured. The Interactive Humanoid and Interactive Nonhumanoid conditions were dummy-coded with the Nonhumanoid as the control category. Robot Feedback: 0 = Agree, 1 = Disagree. Initial Threat-ID: 0 = Ally, 1 = Enemy. Initial Correctness: 0 = Correct, 1 = Incorrect. Reversed Threat-ID: 0 = Repeated, 1 = Reversed.
Back to article page