Fig 1.
Expectation and perception of the virtual assistant’s capability.
Unexpected observations update the user’s expectation of the assistant’s capability.
Fig 2.
a. Hypothesis on uncertainty. A supposedly difficult task “intersects” with the large uncertainty prior distribution at higher probability than when uncertainty is small. When the expectation is low, it is more rational to expect an assistant exhibiting large uncertainty to perform better than it usually does. b. Hypothesis on uncertainty. A supposedly easy task “intersects” with the large uncertainty prior distribution at higher probability than when uncertainty is small. When the expectation is high, it is more rational to expect that an assistant exhibiting large uncertainty may fail to cope with easy tasks.
Fig 3.
Smart speakers used for experiment.
Amazon Echo Plus (left) and Amazon Echo Dot (right).
Fig 4.
Overhead view of experimental setting.
The participant sits 50 centimeters (19.69 inches) away from the table. Echo Dot is placed on a supporting box so that both speakers are at the same altitude.
Fig 5.
Overview of experimental procedures.
Uncertainty was controlled between groups, whereas expectation was controlled within each subject.
Fig 6.
Self-reported intrinsic motivation.
Fig 7.
Self-reported smartness.
Fig 8.
Self-reported comprehensibility.
Fig 9.
Self-reported trust/relatedness.
Fig 10.
Self-reported human-likeness.
Fig 11.
Accumulated numbers of interactions by uncertainty and virtual assistant in charge.
Note that the participants could choose freely to interact with either assistant or to avoid any interaction.
Fig 12.
Numbers of interactions for small-uncertainty group, sorted by motivation type.
Fig 13.
Numbers of interactions for large-uncertainty group, sorted by motivation type.
Fig 14.
Number of interactions with the high-expectation virtual assistant.
Sorted by uncertainty group and motivation type.
Fig 15.
Number of interactions with the low-expectation virtual assistant.
Sorted by uncertainty group and motivation type.
Fig 16.
The original model to explain action selection, from active inference theory [12].
In the presence of a precise goal, action selection is dominated by epistemic value (when uncertainty is great) or extrinsic value (when uncertainty cannot be further reduced).
Fig 17.
Revised model of action selection.
Even in the absence of precise goals, uncertainty perceived during previous interactions should affect action selection.
Fig 18.
Accumulated number of interactions by motivation type.
Fig 19.
Average number of interactions by uncertainty and motivation type.
Fig 20.
Self-reported intrinsic motivation, smartness, and comprehensibility.
Sorted by uncertainty group.