In an effort to share ideas beyond the limited time we have during the annual conference, the Community Building Committee asked Graduate Students attending TPRC54 to take notes of paper sessions to be made available to the community. The following is a summary of the paper session entitled “Digital Inclusion” that was presented on Saturday, September 26, 2026 at 2:00pm. The session was moderated by Erezi Ogbo-Gebhardt, North Carolina Central University. The following is an edited summary based on the notes provided by Joanna Adewunmi, University of Illinois, Urbana-Champaign. Full papers can be accessed through ConfTool.
Paper 1: Above COPPA, Below Protection: Governing AI-Mediated Health Information for Adolescents Ages 13–17
Karen V. Jenkins, University of Illinois Urbana-Champaign
Children under 13 have specific protections, and adults over 18 have others, but adolescents fall into a gap between these age groups. As a result, teens may turn to AI chatbots for health information in systems that were not designed for them. For example, a 15-year-old asking an AI chatbot about menstruation may receive information that increases health anxiety rather than providing appropriate guidance. For millions of teens, AI is becoming an unregulated substitute for primary care.
Current legislation does not adequately address the intersection of adolescent health and AI. AI models are often designed to be agreeable and reinforce users’ perspectives. While this may be inconvenient for adults, it can create an echo chamber for adolescents, rewarding anxiety with continued engagement. Digital equity further shapes who is most vulnerable to these harms.
The presentation argues that avoiding online information should not be treated as an individual failure or a lack of digital literacy, but as a potential consequence of inadequate policy and governance. Digital literacy alone cannot protect the most vulnerable teens, and transparency labels serve as notices rather than safeguards against harmful AI behavior. With 33% of teens reportedly preferring to discuss serious matters with an AI companion, social trust in these systems may outweigh their clinical credibility.
To better govern AI-mediated health information for adolescents, the presentation proposes testing AI models for sycophancy before deployment, evaluating their effectiveness specifically for adolescents, and incorporating age-appropriate design and equity considerations. As the presenter emphasized, “We cannot afford for governance to arrive a generation late.”
Paper 2: Under-connectedness and its implications
Grant Blank, University of Oxford; Bibi Reisdorf, UNC Charlotte
The digital divide has largely been overlooked as internet access has become more widespread. This paper provides an empirical assessment of how internet access shapes online activities, particularly for lower-income households that may experience unreliable connections, low connectivity, and other technical problems.
Using data from approximately 36,000 cases, the study measured under-connectedness through five yes-or-no questions. Respondents who met at least two of the five criteria were classified as under-connected. The findings showed that 9.2% of respondents were under-connected. These respondents were more likely to be Hispanic or Indigenous, unemployed, and living in households with children. They were less likely to be married and had substantially lower household incomes.
The study also examined how under-connectedness affects internet use. Under-connected respondents used the internet less for commercial activities, such as banking and online services; medical activities, such as accessing health records and telehealth; entertainment, including streaming and gaming; and communication through email, texting, and social media. They also reported lower use of smart devices.
The findings show that under-connectedness persists even in a context of near-universal internet access. The quality and reliability of access shape meaningful engagement, and under-connectedness remains strongly associated with socioeconomic disadvantage. The research supports the argument that basic access inequalities still matter and that the quality of connectivity affects both the variety and types of activities people can do online.
The policy implications extend beyond simply providing devices or internet access. Interventions must also address reliability, speed, and maintenance, while smartphone-only access may remain insufficient for full digital participation. The paper raises an important question: Are we reaching the limits of digital engagement when access alone is no longer enough? Future research could explore the long-term impacts of under-connectedness.
Paper 3: Digital Welfare in Practice: What Claimant Communities Reveal About Universal Credit
Grant Blank, University of Oxford; Eglė Karpauskaite, University of Oxford
The British welfare system is increasingly moving toward digital services, with Universal Credit serving as the UK’s flagship welfare reform and largest digital platform. But when the system fails or claimants struggle to navigate it, people often turn to strangers on Facebook for help. Unofficial Universal Credit support groups have hundreds of thousands of members between them, raising questions about what their existence reveals about the system people are trying to navigate.
This research examined the problems, informational needs, and emotional expressions found in Universal Credit Facebook groups. It explored three main areas: the prevalence of administrative burdens, the psychological costs of navigating the system, and how online communities mobilize support.
The findings showed that learning and compliance dominate discussions, with eligibility representing the single largest category of questions. Learning how the system works and understanding its requirements are major challenges for claimants. Confusion and uncertainty were the most common emotions, appearing in 84.3% of posts, followed by anxiety and fear at 42.7%. Posts describing psychological distress and hardship carried the most negative tone, while distress and hardship also attracted the most replies.
These findings suggest that much of the burden is informational: confusion signals where the system is failing to communicate clearly, while Facebook groups have become an important form of informal support infrastructure. For the UK Department for Work and Pensions, the volume of eligibility questions may indicate a measurable communication failure. For group administrators, clearer legal information and pathways to formal advice could strengthen the support volunteers already provide without pay.
The study also highlighted limitations. Because the research focuses on people who use Facebook and have found these groups, the most digitally excluded claimants may be missing. The cross-sectional design cannot follow individuals throughout the claims process, and assigning one label to each post may simplify discussions that span multiple categories. Facebook’s platform design also shapes which posts and replies are visible.
Future research could involve working with Facebook group administrators, examining other platforms such as Reddit, and expanding the analysis to other benefits systems, countries, and government agencies. Overall, the study shows how unofficial online communities can reveal the informational and emotional burdens of digital welfare—and how the questions people ask may point to opportunities for improving the systems meant to support them.