As the internet thrives on the circulation of easily copied content, ensuring attribution is properly given has been a perennial challenge. Following the rise of synthetic media and generative AI tools, and corresponding technologies which enable detailed media provenance, the picture has become considerably more complicated. We present a design research project to consider the implications of these developments from the perspective of the public, everyday (non-professional) user and ‘mundane content’ creation. Through the design, exhibition, and study of a collaborative storytelling tool, ORAgen Fables, we introduce technologies which enable detailed attribution and media provenance and explore contemporary attitudes and concerns about attribution. Our findings suggest that attribution should be understood as relational and dynamic with users having the right to ongoing management of their attribution. This opens a design space for understanding how technical systems could be deployed to define and ascribe attribution for past and future interactions.
Large Language Models (LLMs) are increasingly being integrated into embodied, tangible, and ambient forms, expanding beyond the established chatbot interaction paradigm. LLMs are inherently general and open-ended. In contrast, design practice typically stabilises artefacts by prescribing their role or function through fixed metaphors. We present Machine Eye, a Research through Design (RtD) exploration of an embodied LLM that resists metaphorical closure. Rather than prescribing a specific role or function, the artefact is deliberately ambiguous, inviting participants to explore new forms of relational engagement with AI. Firstly, we explicate our design process, revealing three key tensions encountered when designing against metaphor for embodied LLMs. Secondly, we present findings from a qualitative study (N=15) investigating how participants interpret and engage with Machine Eye. We find that as participants actively explore new and non-prescriptive modes of embodied interaction, perceived roles can be dynamically contested and renegotiated, allowing for a kind of boundless relationship to emerge.
Around 90% of CEOs see AI as the “most critical technology for ensuring future profitability and competitiveness.” At the same time, up to 95% of AI projects fail. Currently, little is known about the leaders who approve and guide AI initiatives. We call them AI Deciders. This study investigates how AI Deciders reason about AI benefits and risks, and how their knowledge about AI influences their decisions on what and where to innovate. We interviewed AI Deciders across diverse organizations. We found no ideation. AI Deciders just consider one concept at a time. Design and HCI played no role in deciding what to build. Many AI Deciders overestimate AI’s benefits while underestimating risks. Based on these findings, we identified opportunities for design and HCI to support impactful and responsible AI innovation. This should reduce AI project failure.
Theory of Mind (ToM)—the ability to infer transient mental states—is traditionally considered fundamental to human social interactions. This has sparked growing efforts in building and benchmarking AI’s ToM, yet little is known about how it could translate into the design and experience of everyday user-facing AI products and services. We conducted 13 co-design sessions with 26 U.S.-based AI practitioners to envision, reflect, and distill design recommendations for ToM-enabled everyday AI systems that are both future-looking and grounded in the realities of AI design and development practices. Analysis revealed three interrelated design recommendations: ToM-enabled AI should 1) be situated in the social context that shape users' mental states, 2) be responsive to the dynamic nature of mental states, and 3) be attuned to subjective individual differences. We surface design tensions within each recommendation that reveal a broader gap between practitioners' envisioned futures of ToM-enabled AI and the realities of current AI development practices. These findings point toward the need to move beyond static, inference-driven approach to ToM and toward designing ToM as a pervasive capability that supports continuous human-AI interaction loops.
Users often struggle to make rational privacy decisions within the notice and choice framework, primarily due to difficulties in understanding and processing privacy policies. Recent studies suggest that large language model (LLM)-based chatbots can improve privacy policy comprehension and privacy awareness. Meanwhile, the specific expectations and needs that users bring to LLM-based chatbots for rational privacy decision-making, and whether those needs are met, have been underexplored. Employing a technology probe and focus groups, we investigate the roles users expect from the chatbots and the needs that arise during interaction. We further interview three experts to corroborate our findings. Our study reveals a typology of three user-expected roles for LLM-based chatbots—Interpreter, Guardian, and Evaluator—along with satisfied and unmet needs within these roles. Finally, we outline implications that clarify where LLMs are—and are not—viable in privacy decision-making, while highlighting structural limitations of notice and choice.
Artificial intelligence (AI) is deepening human social experiences within online spaces in increasingly layered ways. Amid these shifts, we designed AmongOthers, an online community populated with 800 AI agents, and rethought human–AI social interactions. Eight participants engaged with AmongOthers for four weeks. For the first two weeks, they were told the community was exclusively for immigrants and international students, after which we disclosed that most users were AI agents. Participants shared periodic reflections and later joined interviews. Initially, AmongOthers was described as warm and respectful. However, after disclosure, participants diverged in their attitudes toward AI in online social communities, ranging from embracing and denying to imagining it only as a conditional possibility. We discuss these tensions in human perceptions of AI and highlight the risks of framing AI as failed replicas or preferable proxies. We finally suggest rethinking AI as distinct social entities in their own right.
AI-assisted writing raises concerns about autonomy and ownership when benefiting writers. Personalization has been proposed as an effective solution while also risking writers' reliance on AI and behavior shifting. For better personalization design, existing studies rely on interaction and information solely within the writing phase; however, few studies have examined how reading behaviors can inform personalized writing. This study investigates the effects of integrating reading highlights for personalization on AI-assisted writing. A between-subjects study with 46 participants revealed that the personalization condition encouraged participants to produce more highlights. However, highlighting unexpectedly shifted from a sense-making strategy to an instrumental act of "feeding the AI," leading to significant reliance on AI and declines in writers' sense of autonomy, ownership, and self-credit. These findings indicate personalization risks in AI-assisted writing, emphasize the importance of personalization strategies, and provide design implications.