Table 1 Parameter estimates for models of predictors of changes in threat-identification, decisions to kill, or confidence following robot feedback (Expt. 1).

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

5.29

0.48

11.04

 < 0.001

4.35, 6.23

 Embodiment

0.22

0.26

0.82

0.414

 − 0.30, 0.73

 Feedback × embodiment

 − 0.12

0.67

 − 0.18

0.860

 − 1.44, 1.20

 Initial threat-ID

0.49

0.17

2.95

0.003

0.16, 0.81

 Initial correctness

0.83

0.18

4.56

 < 0.001

0.48, 1.19

 Intercept

 − 1.33

0.25

 − 5.40

 < 0.001

 − 1.81, − 0.85

Change 2: Decisions to kill

 Robot feedback

4.95

0.41

12.16

 < 0.001

4.15, 5.75

 Embodiment

0.15

0.25

0.62

0.535

 − 0.33, 0.64

 Feedback × embodiment

 − 0.00

0.58

 − 0.01

0.996

 − 1.13, 1.13

 Initial threat-ID

0.75

0.16

4.60

 < 0.001

0.43, 1.07

 Initial correctness

0.54

0.18

3.02

0.003

0.19, 0.88

 Intercept

 − 1.38

0.24

 − 5.82

 < 0.001

 − 1.84, − 0.91

Change 3: Confidence

 Robot feedback

 − 0.95

0.15

 − 6.46

 < 0.001

 − 1.23, − 0.66

 Embodiment

0.12

0.06

1.86

0.064

 − 0.01, 0.25

 Feedback × embodiment

 − 0.08

0.09

 − 0.90

0.367

 − 0.26, 0.10

 Reversed threat-ID

 − 0.79

0.29

 − 2.73

0.006

 − 1.36, − 0.22

 Feedback × reversed threat-ID

1.20

0.30

4.05

 < 0.001

0.62, 1.79

 Initial threat-ID

0.01

0.05

0.16

0.873

 − 0.08, 0.10

 Initial correctness

0.12

0.05

2.30

0.022

0.02, 0.22

 Intercept

0.21

0.10

1.98

0.048

0.00, 0.41

  1. N = 135. Multilevel models with all predictors and outcomes entered at Level 1, save for the between-subjects robot Embodiment variable at Level 2. All linear variables were standardized. Random intercept included to account for shared variance within participants; covariance matrices were unstructured. Robot Feedback: 0 = Agree, 1 = Disagree. Embodiment: 0 = Disembodied, 1 = Embodied. Initial Threat-ID: 0 = Ally, 1 = Enemy. Initial Correctness: 0 = Correct, 1 = Incorrect. Reversed Threat-ID: 0 = Repeated, 1 = Reversed.
Back to article page