Generative AI applications present unique design challenges. As generative AI technologies are increasingly being incorporated into mainstream applications, there is an urgent need for guidance on how to design user experiences that foster effective and safe use. We present six principles for the design of generative AI applications that address unique characteristics of generative AI UX and offer new interpretations and extensions of known issues in the design of AI applications. Each principle is coupled with a set of design strategies for implementing that principle via UX capabilities or through the design process. The principles and strategies were developed through an iterative process involving literature review, feedback from design practitioners, validation against real-world generative AI applications, and incorporation into the design process of two generative AI applications. We anticipate the principles to usefully inform the design of generative AI applications by driving actionable design recommendations.

As social media continues to grow as a space for emotional self-disclosure, it is important to understand whether self-disclosure acts as a causal factor impacting positive outcomes for users. Thus we developed Chirp, an anonymous social media sandbox space designed to explore the underlying effects of disclosure within online spaces. Users in Chirp are prompted to self-disclose moods and emotions using emojis. Through a between-subjects study among a cohort of first-year undergraduate student users on Chirp, we evaluate the effect of self-disclosure within semi-private online spaces on social support. While Chirp use does not show a significant increase in measured feelings of social support, user responses suggest that self-disclosure in Chirp may provide more social support than typical social media use or simple mood tracking over a two-week period. Our findings indicate that even in pseudo-anonymous, low-bandwidth communication platforms, self-disclosure may cause increased feelings of social support. This work highlights the impact of communication in semi-private online spaces on perceived social support.

The severe impact of COVID-19 in the United States has forced many students to replace in-person socialization with online digital contact. In this study, we investigate the mental health impacts associated with this shift by examining properties of online interactions that may affect loneliness and perceived social support. Students were surveyed (N=827) across 97 universities across the US during their first full semester impacted by the COVID-19 pandemic (Fall 2020). Private online interactions (messaging, phone call, video call) were found to have a comparable correlation to social support as face-to-face interactions, but public online interactions (social media) were associated with more negative outcomes. Among private platforms, messaging had the strongest correlation with social support; and daily self-disclosure over messaging yielded social support levels that were 1.21x higher than rarely or never disclosing over this platform. We speculate that factors such as the level of privacy and peoples' feelings of control contributed to disclosure and perceived social support in online platforms.

Messaging is a common mode of communication, with conversations written informally between individuals. Interpreting emotional affect from messaging data can lead to a powerful form of reflection or act as a support for clinical therapy. Existing analysis techniques for social media commonly use LIWC and VADER for automated sentiment estimation. We correlate LIWC, VADER, and ratings from human reviewers with affect scores from 25 participants. We explore differences in how and when each technique is successful. Results show that human review does better than VADER, the best automated technique, when humans are judging positive affect ($r_s=0.45$ correlation when confident, $r_s=0.30$ overall). Surprisingly, human reviewers only do slightly better than VADER when judging negative affect ($r_s=0.38$ correlation when confident, $r_s=0.29$ overall). Compared to prior literature, VADER correlates more closely with PANAS scores for private messaging than public social media. Our results indicate that while any technique that serves as a proxy for PANAS scores has moderate correlation at best, there are some areas to improve the automated techniques by better considering context and timing in conversations.

Social Virtual Reality based Learning Environments (VRLEs) such as vSocial render instructional content in a three-dimensional immersive computer experience for training youth with learning impediments. There are limited prior works that explored attack vulnerability in VR technology, and hence there is a need for systematic frameworks to quantify risks corresponding to security, privacy, and safety (SPS) threats. The SPS threats can adversely impact the educational user experience and hinder delivery of VRLE content. In this paper, we propose a novel risk assessment framework that utilizes attack trees to calculate a risk score for varied VRLE threats with rate and duration of threats as inputs. We compare the impact of a well-constructed attack tree with an adhoc attack tree to study the trade-offs between overheads in managing attack trees, and the cost of risk mitigation when vulnerabilities are identified. We use a vSocial VRLE testbed in a case study to showcase the effectiveness of our framework and demonstrate how a suitable attack tree formalism can result in a more safer, privacy-preserving and secure VRLE system.