Table 4.
Descriptive statistics of tasks (N=394) attempted.
| Parameter | Time per task (s), median (IQRa) | Attempts, median (IQR) |
Time per attempt (s), median (IQR) |
Task failure, n (%) |
Potential resulting harm, n (%) |
Potential resulting death, n (%) |
|
| Overall | 74.5 (44.8-126.3) | 5.0 (3.0-7.0) | 11.0 (8.0-17.0) | 226 (57.4) | 49 (12.4) | 27 (6.9) | |
| Task type | |||||||
|
|
Medication | 77.5 (47.3-138.0) | 5.0 (3.0-7.8) | 11.0 (8.0-18.0) | 153 (56.9) | 39 (14.5) | 18 (6.7) |
|
|
Emergency | 67.0 (39.8-107.0) | 4.0 (2.0-7.0) | 11.0 (8.0-17.0) | 73 (58.4) | 10 (8.0) | 9 (7.2) |
| System | |||||||
|
|
Alexa | 63.0 (41.3-106.5) | 6.0 (4.0-8.0) | 10.0 (8.0-13.0) | 125 (91.9)b | 2 (1.4)b | 2 (1.4)b |
|
|
Siri | 88.0 (45.0-158.0) | 3.0 (2.0-5.0) | 17.0 (10.0-38.0) | 29 (22.4)b | 27 (20.9)b | 18 (14)b |
|
|
Google Assistant | 79.0 (49.0-116.0) | 6.0 (4.0-8.0) | 12.0 (9.0-18.0) | 72 (55.8)b | 20 (15.5)b | 7 (5.4)b |
aIQR: interquartile range.
bThese data were used in statistical tests of differences between conversational assistants.