When AI sits between two people — smoothing a reply, translating a message, mediating a team — it does not stay neutral. It changes the language, and it changes the relationship. This project studies that second effect, the social one, which is the harder and less-studied half.
Threads here include trust and unguarded disclosure when the AI gives no cue that it is there; anthropomorphism driven by system performance rather than surface design; disclosure and credit in human–AI teams, where who did what becomes genuinely unclear; and persona AI, including systems that reconstruct deceased people, where attachment and authenticity come apart.
The project also carries the lab's methodological argument: that studying AI-mediated communication well requires a middle space between controlled lab experiments and uncontrolled field studies, and that building that infrastructure is itself a research contribution.
Themes across projects
- Disclosure, trust, and accountability in human–AI teamsAlso in Imagined Futures of GenAI in Education
- How AI capabilities shape what people believe AI to be
- Persona AI — including systems that reconstruct the dead — and authenticity, attachment, trust
- Relationships with AI, and relationships mediated by AIAlso in Platform Care Work
- Philosophical and ethical questions around trustAlso in Bystander Intervention & Online Safety, Misinformation & Community Norms
- How AI changes the way that we “are people”Also in Platform Care Work
- Methodological pluralism — questions shape methods, not dogmaShared by every project
Publications
Published
- 2023
Artificial intelligence in communication impacts language and social relationships
Hohenstein, Kizilcec, DiFranzo, Aghajari, Mieczkowski, Levy, Naaman, Hancock & Jung · Scientific Reports 13(1), 5487
- 2020
Robot-assisted tower construction: a method to study the impact of a robot’s allocation behavior on interpersonal dynamics and collaboration in groups
Jung, DiFranzo, Shen, Stoll, Claure & Lawrence · ACM Transactions on Human-Robot Interaction 10(1)
- 2021
Behavioral contagion on social media: effects of social norms, design interventions, and critical media literacy on self-disclosure
Masur, DiFranzo & Bazarova · PLOS ONE 16(7), e0254670
- 2023
The impact of what others do, approve of, and expect you to do: an in-depth analysis of social norms and self-disclosure on social media
Masur, Bazarova & DiFranzo · Social Media + Society 9(1)
- 2023
Methodological middle spaces: addressing the need for methodological innovation in experimental studies of AI-mediated communication
Aghajari, Baumer, Hohenstein, Jung & DiFranzo · PACM HCI 7(CSCW1)
- 2020
“I just shared your responses”: extending communication privacy management theory to interactions with conversational agents
Sannon, Stoll, DiFranzo, Jung & Bazarova · PACM HCI 4(GROUP)
- 2026
Who gets credit? Operationalizing AI disclosure as epistemic coordination in human–AI teams
Shi & DiFranzo · CHI 2026 Extended Abstracts (Late-Breaking Work)
- 2026
arXiv
Outer Limits: an experimental approach to controlled content manipulation within the Reddit interface
Mao, Shi, Jia, Unguryan, Baumer & DiFranzo · CSS 2026, Santa Fe · camera-ready October 2026
Workshops & extended abstracts
- 2026
arXiv
Human control is the anchor, not the answer: early divergence of oversight in agentic AI communities
Shi & DiFranzo · CHI 2026 Workshop on Human-Agent Collaboration
- 2026
arXiv
Alignment in time: peak-aware orchestration for long-horizon agentic systems
Shi & DiFranzo · CHI 2026 AutomationXP Workshop
- 2025
Building trust in multilingual large language models for cross-cultural collaboration
Shi & DiFranzo · HCII 2025 Workshop on Generative AI for Collaborative Experiences
Preprints & work in progress
- 2026
arXiv
Culturally-grounded governance for multilingual language models
Shi & DiFranzo · arXiv:2602.00497
- 2026
arXiv
When visibility outpaces verification
Shi & DiFranzo · arXiv:2602.11412
- 2026
arXiv
When readability and source retention diverge: an evaluability gap in AI translation
Mao, Shi, Jia, Wegrzyn & DiFranzo · arXiv:2608.19083
- 2026
Beloved afterlives: public accountability for AI resurrection of relational others
Shi & DiFranzo · under review · AIES 2026
- 2026
The quiet anthropomorphism: trust and unguarded disclosure in cue-free AI translation
Mao, Shi, Jia, Wegrzyn & DiFranzo · under review · AIES 2026
- 2026
A hybrid field–lab methodology for experiments on live social media
Mao, Jia, Baumer & DiFranzo · under review · ACM GROUP 2027
- 2026
The underspecified team in human–AI teaming: a reporting audit of team form, authority, and accountability
Shi, Wang, Doherty & DiFranzo · under review · ACM HCOMP 2026
-
Afterglow / Petopia: designing AI-mediated memorials that support remembrance without immersion
Shi & DiFranzo · in preparation · target CHI 2027