The study, recently published in the journal Behaviour & Information Technology, investigates how different forms of interactivity during a sign-up process impact users’ vigilance regarding privacy concerns. The researchers took a novel approach by examining two distinct types of interactivity: “message interactivity,” which involves conversational exchanges with the app that build upon previous user inputs, and “modality interactivity,” which includes interface elements like clicking and zooming on images. Their goal was to understand how these interactive elements influence users’ perceptions of playfulness and, more importantly, their readiness to disclose personal information.

To explore this dynamic, an online experiment was conducted involving 216 participants who were asked to complete the sign-up process for a simulated fitness app. Participants were randomly assigned versions of the app that varied in levels of message and modality interactivity. They then rated their experience using seven-point scales that measured perceived fun, engagement, and privacy concerns. The detailed dataset provided compelling evidence that greater interactivity heightened the playfulness of the app, which in turn decreased individuals’ alertness to privacy risks.

One of the study’s most surprising findings relates to message interactivity. While intuitive assumptions would predict that a conversational, back-and-forth interaction would encourage users to think more critically about the personal data they are sharing, the opposite was true. In fact, message interactivity had a distracting effect, drawing users into a playful mindset and diminishing their caution. This insight challenges long-held beliefs in the design community regarding AI chatbots and conversational systems, highlighting that immersive dialogue can create a false sense of security and promote inadvertent data disclosure.

The lead author Jiaqi Agnes Bao from the University of South Dakota, who completed this work during her doctoral studies at Penn State, emphasizes the critical need for interface designs that foster user awareness. Bao suggests that while interactivity can enhance user engagement—an important factor for app success—it must be balanced carefully against privacy considerations. One promising design solution proposed involves coupling message interactivity with modality interactivity, such as periodically inserting pop-up prompts during conversations to encourage users to pause, reflect, and reassess the information they are submitting.

Senior author S. Shyam Sundar, a distinguished professor at Penn State and director of the Center for Socially Responsible Artificial Intelligence, elaborated on the implications of these findings. According to Sundar, current generative AI models primarily rely on message interactivity, creating highly engaging, conversational user experiences. This study cautions that such engagement can become a double-edged sword: while captivating users, it may inadvertently lower their awareness of privacy risks. Sundar advocates for integrating subtle interruption mechanisms within AI interactions as a way to “jerk users into awareness,” preventing unchecked data oversharing.

The broader context of this research is particularly relevant as generative AI technologies proliferate across diverse sectors—from healthcare and finance to social media and entertainment. The playful nature of these systems, often celebrated for enhancing accessibility and user satisfaction, now faces scrutiny for potentially masking users’ vulnerability to privacy breaches. The researchers argue that developers and designers bear a significant ethical responsibility to embed features that not only inform but actively guide users toward more conscious data sharing.

In addition to theoretical contributions, the research introduces practical guidelines for striking a synergy between playfulness and privacy protection. The combination of message and modality interactivity, for example, can induce users to intermittently evaluate their information disclosure without detracting significantly from the overall user experience. This design strategy points toward a future where interactive systems are both engaging and trustworthy—a crucial advancement as digital ecosystems grow increasingly complex.

Furthermore, the study highlights the importance of moving beyond simplistic user notifications about data sharing. According to co-author Yongnam Jung, a doctoral candidate at Penn State, truly building trust requires platforms to facilitate informed decision-making processes rather than relying on passive acknowledgment. This shift towards user-centric privacy empowerment is fundamental to raising digital literacy and fostering sustainable interactions in AI-driven applications.

This latest investigation builds on a foundation of prior research by the team, which similarly revealed that interactivity, while beneficial for engagement, tends to draw attention away from potential risks. Taken together, these studies underscore a critical trade-off that designers, policymakers, and users must grapple with: enhanced interactivity enriches the user experience but simultaneously complicates privacy management.

The study’s timing coincides with a rapidly evolving landscape in generative AI development. As advanced models generate increasingly natural and compelling conversations, there is a growing urgency to address the unintended consequences of such “playful” interactions. The researchers implore industry leaders to consider methods beyond traditional interfaces, advocating for intelligently embedded modality interruptions that prompt privacy awareness at critical moments during the user journey.

Ultimately, this research serves as a cautionary tale about the seductive power of interactivity in digital environments. It exposes how the very features designed to increase app attractiveness can paradoxically sedate user vigilance, allowing hidden privacy vulnerabilities to flourish. By bringing these insights to light, the Penn State team has opened a vital dialogue about responsible AI design that prioritizes both fun and fundamental data protections in this unprecedented technological era.

Subject of Research: Effects of interactivity on users’ privacy disclosure behavior in mobile apps and AI chatbots
Article Title: Are you fooled by interactivity? The effects of interactivity on privacy disclosure
News Publication Date: 24-Aug-2025
Web References:

• DOI:10.1080/0144929X.2025.2545312
• Penn State Center for Socially Responsible Artificial Intelligence
  References: Behaviour & Information Technology Journal
  Keywords: Generative AI, Artificial intelligence, Communications, Mass media, Social media, Smartphones, Behavioral psychology, Risk aversion

The intricate relationships between syllables in birdsong often reflect the same contextual dependencies found in human language. For instance, just like how the meaning of a phrase like “flies like” oscillates based on subsequent words, birds too demonstrate context sensitivity in their sequences of notes. The research posits that understanding these patterns could provide deeper insights into the cognitive and neural mechanisms that underlie the complexities of language as a whole.

Dezhe Jin, the lead author of the study and an associate professor of physics at Penn State, emphasizes the importance of studying birdsong as a model for language exploration. The research team focused on Bengalese finches because their songs consist of a finite number of syllables arranged in various combinations, which makes them ideal subjects to investigate the structural properties of communication. The team recorded the songs of six finches, each of which demonstrated unique contextual dependencies in their vocalizations.

Employing advanced statistical techniques, the researchers sought to create models that closely reflect the singing patterns specific to individual birds. Unlike traditional approaches that may yield generalized outcomes, this new methodology, called the Partially Observable Markov Model, incorporates context-dependence, thus enhancing the accuracy of the models. This innovative approach reveals how birds adapt their songs based on previously sung syllables, illustrating a sophisticated level of cognitive processing.

To delve deeper into the mechanisms at play, the scientists also studied finches that had not experienced auditory feedback due to hearing impairments. The results were striking: these birds exhibited a significant reduction in context-dependent syllable transitions, suggesting that auditory input is crucial for developing complex song patterns. This finding points toward the fundamentally interactive nature of learning in avian species, where listening to self-generated songs plays a critical role in forming a cohesive vocal repertoire.

The implications of this research extend beyond ornithology. The modeling technique utilized for analyzing birdsong has parallels with language processing in humans, raising intriguing questions about the universality of cognitive mechanisms for communication across species. The researchers were able to apply their models to human language as well, producing constructs that resemble grammatical sentences within the English language. This crossover underlines potential parallels between the neural frameworks governing birdsong and human language.

The notion that birdsong and human communication share underlying neural processes invites reconsideration of how we view the uniqueness of human language. If the mechanisms enabling avian vocalizations can be understood as fundamentally similar to those that facilitate human language, it challenges the traditional narrative of the exceptionalism of human communicative abilities. This perspective paves the way for future studies aimed at mapping the neural underpinnings of both birdsong and human speech.

In addition to revealing the complexities of avian communication, this research also serves as a template for investigating other animal vocalizations. The application of these advanced modeling techniques could translate to a broader understanding of how various species communicate and adapt their vocal behaviors. Such insights are crucial not only for the fields of neurobiology and linguistics but also for conservation efforts, as understanding communication patterns can inform species management and preservation strategies.

The collaborative nature of this research further highlights the multidisciplinary approach necessary for unraveling the mysteries of communication across species. The diverse backgrounds of the research team members—combining physics, neuroscience, and behavioral studies—demonstrate the importance of integrating various scientific perspectives to tackle complex biological questions. As this research continues to evolve, it stands as a testament to the value of collaborative inquiry in advancing our understanding of animal behavior and cognitive processes.

The future direction of this research promises to uncover even more layers of complexity in the interplay between auditory feedback, vocal learning, and neurobiological mechanisms in birds. The researchers express a desire to map specific neuron states to syllable production, which could illuminate the intricacies of how avian brains process and generate song sequences. By elucidating these connections, scientists hope to bridge gaps in our knowledge about the evolution of communication and the cognitive capacities required for its development.

Ultimately, the findings of this study underscore the necessity of continual exploration in the realms of behavioral science and neurobiology. The parallels drawn between birdsong and human language not only enhance our grasp of communication as a biological phenomenon but also evoke broader philosophical inquiries regarding the essence of language and what it means to communicate. As researchers pursue these questions, it is likely that further discoveries will redefine our understanding of the cognitive and biological foundations of language and social interaction.

Subject of Research: Animals
Article Title: Partially observable Markov models inferred using statistical tests reveal context-dependent syllable transitions in Bengalese finch songs
News Publication Date: 8-Jan-2025
Web References: Journal of Neuroscience
References: Not specified
Image Credits: Credit: Zachary Jin

Keywords: Neural mechanisms, Neural modeling, Animal research, Human brain models, Generative AI, Animal psychology, Birds, Modern birds, Neurolinguistics