Spoken Humanoid Embodied Conversational Agents in Mobile Serious Games: A Usability Assessment

Overview

• This paper investigates how human-like Embodied Conversational Agents (ECAs) can improve the usability of mobile serious games.
• The study compares two versions of an agent: one with high human-likeness and one with low human-likeness (text-only).
• The goal is to assess whether highly human-like agents can create a stronger "illusion of humanness" and impact the overall user experience.

Plain English Explanation

The researchers wanted to see if using human-like virtual characters, or Embodied Conversational Agents (ECAs), could make mobile games more enjoyable and easier to use. They created two versions of a virtual agent - one that looked and sounded very human-like, and one that was just text.

The researchers then had 90 people try out the two versions and share their preferences. The results showed that most people preferred interacting with the highly human-like agent. They said the agent's human-like qualities made the experience more appealing.

This suggests that designing ECAs to be very realistic and human-like, as described in the ECA design model, could improve how people interact with and enjoy mobile games. This could help make these "serious games" more engaging and effective for learning or training purposes.

Technical Explanation

The experiment investigated two styles of agent presentation: a highly human-like HECA (Humanoid Embodied Conversational Agent) and a low human-likeness text-only agent. The HECA was designed following the ECA design model, which provides guidelines for developing realistic and human-like ECAs.

The researchers conducted an experiment with 90 participants to assess the impact of agent humanlikeness on usability metrics like appeal and perceived humanness. The results showed a statistically significant preference for the HECA version, with a large effect size (d=1.01). Participants justified their preference by citing the human-like characteristics of the HECA as making it more appealing.

This supports the idea that highly human-embodied AI agents can positively impact the user experience in mobile serious games, as described in prior work on building better human-agent teams and social influence of embodied AI.

Critical Analysis

The paper provides valuable insights into the potential benefits of highly human-like ECAs for mobile serious games. However, the research is limited to a single experiment and may not generalize to all game genres or user demographics.

Additionally, the paper does not delve into potential downsides or unintended consequences of highly realistic ECAs, such as the risk of creating false expectations or mistrust. Further research is needed to fully understand the tradeoffs and design considerations for using human-like ECAs in interactive applications.

Conclusion

This research suggests that designing Embodied Conversational Agents to be highly human-like can improve the user experience in mobile serious games. By creating a stronger "illusion of humanness," these agents may make the games more appealing and engaging for players.

While promising, this is an area that requires further exploration to understand the full implications and potential pitfalls. Nonetheless, this study provides valuable insights that can inform the development of future mobile serious games and conversational AI systems aimed at enhancing user experiences.