A study on the effects of mixed explicit and implicit communications in human-artificial-agent interactions
Organizations: Manchester Centre for Robotics and AI, The University of Manchester, Oxford Rd, Manchester, M13 9PL, UK.
Abstract
Communication between humans and artificial agents is essential for their interaction. This is often inspired by human communication, which uses gestures, facial expressions, gaze direction, and other explicit and implicit means. This work presents interaction experiments where humans and artificial agents interact through explicit and implicit communication to evaluate the effect of mixed explicit-implicit communication against purely explicit communication and the impact of the task difficulty in this evaluation. Results obtained using Bayesian parameter estimation show that the task execution time did not significantly change when mixed explicit and implicit communications were used in neither of our experiments, which varied in the type of artificial agent (virtual agent and humanoid robot) used and task difficulty. The number of errors was affected by the communication only when the human was executing a more difficult task, and an impact on the perceived efficiency of the interaction was only observed in the interaction with the robot, for both easy and difficult tasks. In contrast, acceptance, sociability, and transparency of the artificial agent increased when using mixed communication modalities in both our experiments and task difficulty levels. This suggests that task-related measures, such as time, number of errors, and perceived efficiency of the interaction, as well as the impact of the communication on them, are more sensitive to the type of task and the difficulty level, whereas the combination of explicit and implicit communications more consistently improves human perceptions about artificial agents.
Figures & tables
| Explicit | Implicit | |
|---|---|---|
| ( Breazeal et al., 2005 ; Bauer et al., 2008 ) | Information conveyed deliberately | Information inherent to the behaviour |
| ( Knepper et al., 2017 ; Che et al., 2020 ) | Deliberate and unambiguous | Information incorporated to behavior/action, with context-dependent interpretation |
| Present work | Directly interpreted | Subjective inferences and interpretation |
| Human | Virtual agent | |
|---|---|---|
| Explicit | gestures manual entries | voice sound signals information on screen |
| Implicit | gaze direction location | facial expressions raise of eyebrows gaze direction |
| Acceptance of the virtual agent 1. I found VA intimidating. (R) 2. I found VA friendly. 3. I felt comfortable while interacting with VA. 4. I liked to interact with VA. 5. I found unpleasant to interact with VA. (R) 6. I would like to interact more with VA. |
|---|
| Sociability of the virtual agent 1. I felt like VA understood what I was doing. 2. When interacting with VA, I felt like I was with a real person. 3. Sometimes I felt like VA was really looking at me. 4. Sometimes VA seemed to have real feelings. 5. VA’s behaviour is similar to a person’s. |
| Transparency of the virtual agent 1. I understood VA. 2. I was able to know what VA was “thinking.” 3. I knew when VA was paying attention on me. 4. During the interaction, VA’s intentions were clear to me. |
| Perceived efficiency of the interaction 1. I could count with VA to help me during the task. 2. VA helped me to execute the task. 3. VA got in my way. (R) 4. VA made no difference to my performance. (R) |
| Goal | Experiment 1 | Experiment 2 | ||
|---|---|---|---|---|
| Easy | Difficult | |||
| Time | 7.3% | 8.6% | 8.4% | |
| 2.6% | 4.2% | 3% | ||
| Number of errors | 4.7% | 0.7% | 41% | |
| 2.6% | 0.5% | 41% | ||
| Acceptance | 26.3% | 7.2% | 6.2% | |
Appendix figures & tables1 asset
Supplementary material from the paper’s appendix.
Appendix
| Advantages | Drawbacks | |
| Keep empty levels | Original data | Possible negative probabilities |
| Compress empty levels | No negative probabilities | Changed data |
| Estimation bias | ||
| Restricted prior distributions for thresholds | Original data | Estimation bias |
| No negative probabilities | ||
| Add extra data | No negative probabilities | Changed data |
Human-AI Interaction
Human-AI interaction studies how people and AI systems communicate, collaborate, make decisions, and shape one another over time.
See all 351 papers