Zhilin Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Zhilin Zhang has a total of 20 co-authors. Here are the top co-authors by number of shared publications:
- Ge Wang
Siebel School of Computing and Data Science, University of Illinois at Urbana-Champaign, Champaign, Illinois, USA
4 shared publications - Jun Zhao
University of Oxford
4 shared publications - Max Van Kleek
University of Oxford, Oxford, United Kingdom
4 shared publications - Nigel Shadbolt
University of Oxford
4 shared publications - Samantha-Kaye Johnston
The University of Oxford, United Kingdom
3 shared publications - Diyi Liu
Department of Communication University of Copenhagen Denmark
2 shared publications
Latest Publications
- Open Access
- 13 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-17
Attitudes, Imagined Roles, and Governance Boundaries for AI in Decentralized Social Media
Decentralised social media (DSM) platforms such as Mastodon offer community-governed alternatives to corporate social networks but place substantial governance burdens on volunteer operators. As interest grows in applying artificial intelligence (AI) to support this work, little is known about whether DSM operators want AI, what roles they consider appropriate, and what governance boundaries they require. We conducted semi-structured interviews with 20 operators across Mastodon, Pixelfed, PeerTube, Lemmy, Pleroma, and Funkwhale, using generative feature probes and speculative scenarios to explore their perceptions of AI. Operators rejected AI as an autonomous actor, instead envisioning it as governance infrastructure that provides contextual intelligence, supports cross-instance coordination, and sustains community and moderator well-being. They also articulated strict boundaries rooted in DSM values, including human accountability, reversibility, transparency, community-centred configuration, and strong data-governance constraints. We contribute empirical insights and design implications for AI compatible with decentralised, federated social media.
- 13 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-17
Open AccessAttitudes, Imagined Roles, and Governance Boundaries for AI in Decentralized Social Media
Decentralised social media (DSM) platforms such as Mastodon offer community-governed alternatives to corporate social networks but place substantial governance burdens on volunteer operators. As interest grows in applying artificial intelligence (AI) to support this work, little is known about whether DSM operators want AI, what roles they consider appropriate, and what governance boundaries they require. We conducted semi-structured interviews with 20 operators across Mastodon, Pixelfed, PeerTube, Lemmy, Pleroma, and Funkwhale, using generative feature probes and speculative scenarios to explore their perceptions of AI. Operators rejected AI as an autonomous actor, instead envisioning it as governance infrastructure that provides contextual intelligence, supports cross-instance coordination, and sustains community and moderator well-being. They also articulated strict boundaries rooted in DSM values, including human accountability, reversibility, transparency, community-centred configuration, and strong data-governance constraints. We contribute empirical insights and design implications for AI compatible with decentralised, federated social media.
- Open Access
- 7 November 2024
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Human-Computer Interaction
- Vol. 8 (CSCW2) , 1-24
Trouble in Paradise? Understanding Mastodon Admin's Motivations, Experiences, and Challenges Running Decentralised Social Media
Decentralised social media platforms are increasingly being recognised as viable alternatives to their centralised counterparts. Among these, Mastodon stands out as a popular alternative, offering a citizen-powered option distinct from larger and centralised platforms like Twitter/X. However, the future path of Mastodon remains uncertain, particularly in terms of its challenges and the long-term viability of a more citizen-powered internet. In this paper, following a pre-study survey, we conducted semi-structured interviews with 16 Mastodon instance administrators, including those who host instances to support marginalised and stigmatised communities, to understand their motivations and lived experiences of running decentralised social media. Our research indicates that while decentralised social media offers significant potential in supporting the safety, identity and privacy needs of marginalised and stigmatised communities, they also face considerable challenges in content moderation, community building and governance. We emphasise the importance of considering the community's values and diversity when designing future support mechanisms.
- 7 November 2024
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Human-Computer Interaction
- Vol. 8 (CSCW2) , 1-24
Open AccessTrouble in Paradise? Understanding Mastodon Admin's Motivations, Experiences, and Challenges Running Decentralised Social Media
Decentralised social media platforms are increasingly being recognised as viable alternatives to their centralised counterparts. Among these, Mastodon stands out as a popular alternative, offering a citizen-powered option distinct from larger and centralised platforms like Twitter/X. However, the future path of Mastodon remains uncertain, particularly in terms of its challenges and the long-term viability of a more citizen-powered internet. In this paper, following a pre-study survey, we conducted semi-structured interviews with 16 Mastodon instance administrators, including those who host instances to support marginalised and stigmatised communities, to understand their motivations and lived experiences of running decentralised social media. Our research indicates that while decentralised social media offers significant potential in supporting the safety, identity and privacy needs of marginalised and stigmatised communities, they also face considerable challenges in content moderation, community building and governance. We emphasise the importance of considering the community's values and diversity when designing future support mechanisms.
- Open Access
- 11 May 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-18
KOALA Hero Toolkit: A New Approach to Inform Families of Mobile Datafication Risks
Children today are deeply immersed in the online world, where their activities are routinely tracked, analysed, and monetised. This exposes them to various datafication risks, including harmful profiling, micro-targeting and behavioural engineering. Most existing measures focus on immediate online threats, rather than informing children about these implicit risks. In this paper, we present The KOALA Hero Toolkit, a hybrid toolkit designed to help children and parents jointly understand the datafication risks posed by their mobile apps. Through user studies involving 17 families we evaluate how the toolkit influenced families’ thought processes, perceptions and decision-making regarding mobile datafication risks. Our findings show that KOALA Hero supports families’ critical thinking and promotes family engagement. We identify future design recommendations for family support, featuring ideas such as integrating triggering moments and bonding moments in toolkit designs. This work provides timely inputs on global efforts aimed at addressing datafication risks and underscores the importance of strengthening legislative and policy enforcement of ethical data governance.
- 11 May 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-18
Open AccessKOALA Hero Toolkit: A New Approach to Inform Families of Mobile Datafication Risks
Children today are deeply immersed in the online world, where their activities are routinely tracked, analysed, and monetised. This exposes them to various datafication risks, including harmful profiling, micro-targeting and behavioural engineering. Most existing measures focus on immediate online threats, rather than informing children about these implicit risks. In this paper, we present The KOALA Hero Toolkit, a hybrid toolkit designed to help children and parents jointly understand the datafication risks posed by their mobile apps. Through user studies involving 17 families we evaluate how the toolkit influenced families’ thought processes, perceptions and decision-making regarding mobile datafication risks. Our findings show that KOALA Hero supports families’ critical thinking and promotes family engagement. We identify future design recommendations for family support, featuring ideas such as integrating triggering moments and bonding moments in toolkit designs. This work provides timely inputs on global efforts aimed at addressing datafication risks and underscores the importance of strengthening legislative and policy enforcement of ethical data governance.
- Open Access
- 11 May 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-19
CHAITok: A Proof-of-Concept System Supporting Children's Sense of Data Autonomy on Social Media
- 11 May 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-19
Open AccessCHAITok: A Proof-of-Concept System Supporting Children's Sense of Data Autonomy on Social Media
- Open Access
- 7 August 2023
- Published by Association for Computing Machinery (ACM)
- p. 159-176
Am I Wrong, or Is the Autograder Wrong? Effects of AI Grading Mistakes on Learning
Errors in AI grading and feedback often have an intractable set of causes and are, by their nature, difficult to completely avoid. Since inaccurate feedback potentially harms learning, there is a need for designs and workflows that mitigate these harms. To better understand the mechanisms by which erroneous AI feedback impacts students’ learning, we conducted surveys and interviews that recorded students’ interactions with a short-answer AI autograder for “Explain in Plain English” code reading problems. Using causal modeling, we inferred the learning impacts of wrong answers marked as right (false positives, FPs) and right answers marked as wrong (false negatives, FNs). We further explored explanations for the learning impacts, including errors influencing participants’ engagement with feedback and assessments of their answers’ correctness, and participants’ prior performance in the class. FPs harmed learning in large part due to participants’ failures to detect the errors. This was due to participants not paying attention to the feedback after being marked as right, and an apparent bias against admitting one’s answer was wrong once marked right. On the other hand, FNs harmed learning only for survey participants, suggesting that interviewees’ greater behavioral and cognitive engagement protected them from learning harms. Based on these findings, we propose ways to help learners detect FPs and encourage deeper reflection on FNs to mitigate the learning harms of AI errors.
- 7 August 2023
- Published by Association for Computing Machinery (ACM)
- p. 159-176
Open AccessAm I Wrong, or Is the Autograder Wrong? Effects of AI Grading Mistakes on Learning
Errors in AI grading and feedback often have an intractable set of causes and are, by their nature, difficult to completely avoid. Since inaccurate feedback potentially harms learning, there is a need for designs and workflows that mitigate these harms. To better understand the mechanisms by which erroneous AI feedback impacts students’ learning, we conducted surveys and interviews that recorded students’ interactions with a short-answer AI autograder for “Explain in Plain English” code reading problems. Using causal modeling, we inferred the learning impacts of wrong answers marked as right (false positives, FPs) and right answers marked as wrong (false negatives, FNs). We further explored explanations for the learning impacts, including errors influencing participants’ engagement with feedback and assessments of their answers’ correctness, and participants’ prior performance in the class. FPs harmed learning in large part due to participants’ failures to detect the errors. This was due to participants not paying attention to the feedback after being marked as right, and an apparent bias against admitting one’s answer was wrong once marked right. On the other hand, FNs harmed learning only for survey participants, suggesting that interviewees’ greater behavioral and cognitive engagement protected them from learning harms. Based on these findings, we propose ways to help learners detect FPs and encourage deeper reflection on FNs to mitigate the learning harms of AI errors.