They found that interactivity enhances perceived playfulness and user intent to engage with apps. This was accompanied by a decrease in privacy concerns. Surprisingly, Sundar said the interactivity of messages that researchers thought would increase user vigilance was distracting instead from thinking about personal information users may be sharing with the system. That is, researchers say how AI chatbots work today – how they build responses based on previous inputs from users are less likely to think about the sensitive information they share.
“Recently, when users are involved with AI agents, there are a lot of interaction conversations. The experience is so engaging that Jiaqi Agnes Bao, professor of strategic communications at the University of South Dakota, who completed his research at the University of Pennsylvania. “We wanted to understand how to better design the interface to ensure that users are aware of information disclosure.”
While user vigilance plays a major role in preventing unintended disclosure of personal information, APP and AI developers can balance playful and privacy concerns through design choices that bring equally favorable situations for individuals and businesses, Bao says.
“We found that if both message interactivity and modality interactivity are designed to work in tandem, users could pause and reflect,” she said. “So when users talk with AI chatbots, a pop-up button that asks users to rate the experience or leave a comment on how to improve customized responses, helping the company provide a better, customized experience to consider the type of information they share with their systems.”
Penn State doctoral candidate Yongnam Jung said the AI platform's responsibility goes beyond simply providing users with the option to share or share personal information.
“It's about not only notifying users, but helping them make informed choices. This is a responsible way to build trust between the platform and users,” she added.
The researchers said the study was based on previous research from the team and revealed a similar pattern. Together, the two studies said they emphasized important trade-offs. Interactiveness enhances the user experience, but highlights the benefits of the app and draws attention from potential privacy risks.
Generated AI is based on the interactivity of messages, which is essentially conversational, mostly in most application domains, Sundar, who is also director of the Center for Socially Responsible Artificial Intelligence (CSRAI) in Pennsylvania. He added that the findings of this study challenge current thinking among designers who make conversation-based tools more cognitively vigilant towards negative aspects such as privacy concerns, unlike click and swipe tools.
“In fact, conversation-based tools have turned out to be playful exercises, and we see them reflected in the larger discourse about generating AI where there are all sorts of stories about people being drawn into conversations to do things that seem illogical,” he said. “They follow the advice of generative AI tools for highly high-stakes decision-making. In some respects, our research is a caveat of this new suite of generative AI tools.
