Astrid Carolus

Intelligent voice assistants (IVAs) are widely used, but they also risk influencing users without their knowledge. The anthropomorphic nature of IVAs can tempt users to disclose more personal information and trust IVAs more than intended. Therefore, targeted training is necessary to sensitize users and protect them from the persuasive potential of IVAs. This study developed and evaluated two online training modules designed to promote persuasion literacy among student samples. The modules cover the anthropomorphic characteristics of IVAs, their persuasive power, and potential countermeasures. The results indicate that the training improves understanding of persuasion by IVAs, raises awareness of their persuasive potential, and promotes recognition of persuasion attempts. Additionally, the training modules reduce anthropomorphic perceptions of IVAs and reinforce psychological aspects to ensure privacy and self-determined interaction. The training modules provide an innovative approach to enhancing the competent use of IVAs. As a result, they represent an exciting approach for research and educational institutions to further explore and promote persuasion literacy in the context of IVAs and other AI-related applications. Therefore, this work contributes to enhancing the accessibility of digital educational programs.

A growing number of studies have compared human and AI creative performance. These studies differ in AI chatbots, human populations, creativity tasks, and creativity indicators (e.g., originality, usefulness, elaboration). They mostly neglect psychological research on determinants of creative performance such as instructions or processing time. The present study contributes to the theoretical foundation and replicates a study comparing humans’ and AI’s creative output in the Alternate Uses Task. Building on established knowledge of creativity determinants, we modified the Alternate Uses Task’s instructions (call for quality AND quantity), provided more time for the human participants, and added a second task (Remote Associates Task). The Alternate Uses Task output was scored in two ways: the mean and maximum scores of each Alternate Uses Task item, both in terms of semantic distances and in terms of human rating scores. The result shows that AI’s mean scores were significantly higher in the original and modified Alternate Uses Task condition, maximum scores in the original Alternate Uses Task condition, and in the Remote Associates Task. No significant differences between humans and AI were found for the maximum scores in the modified Alternate Uses Task. We mainly replicated the original studies’ findings. Our study provides initial clues that the evaluation of creative performances depends on creativity indicators and approaches (instructions and time).

Artificial intelligence gives companions in games (a special form of non-player characters) social, adaptive, and interactive characteristics that can influence the gaming experience. Initial design frameworks for these characteristics have been developed in research, but empirical validation and methods for quantifying their impact on game experience are still lacking. To close this gap, the study developed the Companion Design Scale, which operationalizes the design of companion characteristics and measures their impact on the game experience. Structural equation modeling shows that the scale measures companion characteristics reliably and validly. In addition, the study emphasizes the relevance of companion design in general, but also regarding specific design factors for the game experience. First practical analyses provide design recommendations for game design. This study thus offers insights into the design principles of humanoid AI companions and provides a basis for systematic studies on the psychological effects of artificial companions on users.

Even though people imagine different embodiments when asked which AI they would like to work with, most studies investigate trust in AI systems without specific physical appearances. This study aims to close this gap by combining influencing factors of trust to analyze their impact on the perceived trustworthiness, warmth, and competence of an embodied AI. We recruited 68 participants who observed three co-working scenes with an embodied AI, presented as expert/novice (expertise), human/AI (humanness), or congruent/slightly incongruent to the environment (congruence). Our results show that the expertise condition had the largest impact on trust, acceptance, and perceived warmth and competence. When controlled for perceived competence, the humanness of the AI and the congruence of its embodiment to the environment also influence acceptance. The results show that besides expertise and the perceived competence of the AI, other design variables are relevant for successful human-AI interaction, especially when the AI is embodied.