Cyan DeVeaux’s research while affiliated with Stanford University and other places

Generative artificial intelligence has shown promise in prompting virtual worlds into existence, yet little attention has been given to understanding how this process unfolds as social interaction. We present Social Conjurer, a framework for AI-augmented dynamic 3D scene co-creation, where multiple users collaboratively build and modify virtual worlds in real-time. Through an expanded set of interactions, including social and tool-based engagements as well as spatial reasoning, our framework facilitates the creation of rich, diverse virtual environments. Findings from a preliminary user study (N=12) provide insight into the user experience of this approach, how social contexts shape the prompting of spatial environments, and perspective on social applications of prompt-based 3D co-creation. In addition to highlighting the potential of AI-supported multi-user world creation and offering new pathways for AI-augmented creative processes in VR, this article presents a set of implications for designing human-centered interfaces that incorporate AI models into 3D content generation.

Social virtual reality is an emerging medium of communication. In this medium, a user's avatar (virtual representation) is controlled by the tracked motion of the user's headset and hand controllers. This tracked motion is a rich data stream that can leak characteristics of the user or can be effectively matched to previously-identified data to identify a user. To better understand the boundaries of motion data identifiability, we investigate how varying training data duration and train-test delay affects the accuracy at which a machine learning model can correctly classify user motion in a supervised learning task simulating re-identification. The dataset we use has a unique combination of a large number of participants, long duration per session, large number of sessions, and a long time span over which sessions were conducted. We find that training data duration and train-test delay affect identifiability; that minimal train-test delay leads to very high accuracy; and that train-test delay should be controlled in future experiments.

Social virtual reality (VR), by definition, focuses on people, using networked VR systems to bring avatars together. Previous studies have examined how different factors affect social interaction, in small groups such as dyads or triads. However, in a typical social VR scene there tends to be dozens of avatars, even those not directly interacting with a given user. Furthermore, beyond the virtual environment, VR users are also situated in various immediate physical social contexts. In two field experiments, we investigate how the presence of virtual and physical people contextualize and influence nonverbal behaviors. Study 1 examines virtual context and asks how interacting with others in a private or public virtual environment influences nonverbal outcomes during interactions in a social VR platform. Across two sessions, participants (n = 104) met either in a private virtual environment with their group members alone or in a public environment surrounded by four other groups. Results showed that participants moved their avatars slower and stood closer to group members in public versus private environments. Study 2 examines physical context and asks how interacting with virtual others while physically together or alone influences nonverbal behaviors. Participants (n = 61) met in virtual environments while they were in either a shared physical environment or separated physical environments. Results showed that, compared to remote participants, participants who were physically together moved their bodies more slowly, but their avatars faster. Moreover, there was more mutual gaze among remote participants. We discuss implications to theories of social influence in VR.

In networked virtual reality (VR), user behaviors, individual differences, and group dynamics can serve as important signals into future speech behaviors, such as who the next speaker will be and the timing of turn-taking behaviors. The ability to predict and understand these behaviors offers opportunities to provide adaptive and personalized assistance, for example helping users with varying sensory abilities navigate complex social scenes and instantiating virtual moderators with natural behaviors. In this work, we predict turn-taking behaviors using features extracted based on social dynamics literature. We discuss results from a large-scale VR classroom dataset consisting of 77 sessions and 1660 minutes of small-group social interactions collected over four weeks. In our evaluation, gradient boosting classifiers achieved the best performance, with accuracies of 0.71--0.78 AUC (area under the ROC curve) across three tasks concerning the "what", "who", and "when" of turn-taking behaviors. In interpreting these models, we found that group size, listener personality, speech-related behavior (e.g., time elapsed since the listener's last speech event), group gaze (e.g., how much the group looks at the speaker), as well as the listener's and previous speaker's head pitch, head y-axis position, and left hand y-axis position more saliently influenced predictions. Results suggested that these features remain reliable indicators in novel social VR settings, as prediction performance is robust over time and with groups and activities not used in the training dataset. We discuss theoretical and practical implications of the work.

As social virtual reality (VR) continues to grow as a medium for digital communication, sustaining presence among communicators remains one of the main constructs that researchers and practitioners use to assess the quality of user experience. In the present paper, we explore language patterns as a behavioral link to presence. We accomplished this through an exploratory text analysis of over 4,800 min of conversation in social VR, consisting of over 130,000 spoken words from 126 participants. We observed that the use of self-references and collective references positively correlated to social presence and spatial presence. Furthermore, median interpersonal distance between communicators was positively associated with using impersonal pronouns, suggesting that participants who stood farther apart from their interlocutors tended to speak in more impersonal terms. Our work sheds light on the possible psychological mechanisms behind presence and the potential of using speech data to help build systems that enhance user engagement.

Research on physical-world environments has shown that the spatial properties of built worlds are consequential for shaping psychological states and social behavior. However, it has been difficult to empirically test this in natural settings in the physical world. This study uses immersive virtual reality (VR) environments, which have shown to have comparable effects to physical-world environments, to investigate the influence of two spatial dimensions (ceiling height and floor area) on individuals' attitudes and nonverbal behaviors during social interactions. In the present study, groups of three to four physically remote participants wore VR headsets (n = 110) and took part in discussions every week for four weeks in one of four virtual environments that varied in their spatial dimensions (low or high ceilings, small or large floor areas). Results showed that, when in a virtual environment with a high ceiling, participants reported feeling greater perceived restorativeness, awe, and momentary affective well-being, compared to when they were in virtual environments with low ceilings. Participants also paid more social attention (i.e., looked at other group members), when they were in virtual environments with high ceilings. When in a virtual environment with a large floor area, participants reported having a greater sense of awe, compared to environments with small floor areas. Furthermore, when in a large environment with a high ceiling, participants physically moved their heads more slowly and virtually stood further apart from their group members, compared to the other three conditions. We discuss implications for theoretical work on context and behaviors as well as design of social VR environments.