82. AI in Practice
発表担当
目次
説明
People frequently face preference-based planning tasks requiring balancing goals with nuanced constraints, yet even advanced LLMs demand considerable effort to produce and adjust plans reflecting complex user preferences. We present MAVIS (Multi-Agent Virtual Interactive Synergy), a multi-agent system within a virtual workspace that introduces an incremental collaboration mechanism. This mechanism automatically decomposes tasks into guidelines and sequentially introduces expert agents. Each agent proactively engages users in focused dialog to uncover implicit preferences, while successive agents add perspectives and transparently negotiate trade-offs. To mitigate textual overload, MAVIS employs spatial visualizations that externalize agents' reasoning through step-linked summaries and context-aware boards, with embodied avatars supporting natural interaction. Across studies, Study 1 showed our collaboration mechanism doubled expressed preferences and improved planning quality by 60.3% over a conventional LLM baseline. Study 2 affirmed visualization's benefits over a non-spatial baseline, while Study 3 confirmed its versatility across VR and desktop modalities and diverse tasks.
日本語まとめ
説明
In the mid-2010s, media artists began developing practices using machine learning (ML) as an artistic medium. Since 2022, the rise of large generative models, the mainstreaming of AI as consumer products, and intensifying ethical disputes have reconfigured the conditions of their artistic practice. This paper aims to understand how artists working with ML over the past decade respond to these shifts, shedding light on how practices, tools, and culture co-evolve. We address this question through thematic analysis of semi-structured interviews with 30 artists active before 2020. Our findings show how artists experience narrowing aesthetics and reduced malleability of post-2020 ML systems, have diverging views on where to locate moral responsibility with large AI models, and face shifting cultural reception that challenges the legibility of their work. We map how artists envision their practice going forward and discuss those orientations with respect to HCI conversations on design and creativity.
日本語まとめ
説明
Human–AI co-creativity is often understood through a tool-use or collaboration frame, obscuring the relational dynamics of artistic practice with machine learning (ML) systems. To advance the field, we recast such artistic practice through Material Engagement Theory (MET), emphasizing how creative and aesthetic processes emerge in sustained relations between artist and system. We examined 18 contemporary artists’ engagements with ML from 54 documents using a qualitative, interpretive, and framework-guided approach. Our findings show how artists and ML systems co-constitute creative and aesthetic properties by mutually discovering and attuning to each other’s affordances, where creativity unfolds along a control-chance continuum driven by creative and aesthetic thinging, enabling a mutual aesthetic becoming through sustained interaction. Key factors include developing artistic intuition for ML, internalisation, and the evolving role of the meta-artist. Herewith, we contribute MET as a framework for enriching our understanding of human–AI co-creativity as emergent and relational.
日本語まとめ
説明
Control is a critical yet underexplored concept in human-AI co-creativity and more broadly human-AI collaboration, where AI systems are expected to act as collaborative partners with creative autonomy. Existing frameworks for characterizing control remain limited and often fall short in capturing the tensions and complexities of co-creation dynamics. In this paper, we examine how experts conceptualize control and expect human-AI control dynamics by leveraging a recent framework on characterizing control as our theoretical probe. We conduct a semi-structured focus group study with nine experts in HCI, co-creativity, and AI. Our findings reveal that control is widely viewed as a dynamic, context-dependent construct that should adapt across different phases of co-creation, domains, and levels of trust in AI. Drawing on our findings, we propose a conceptualization of control along with actionable design implications for designing such AI systems. This work contributes to the literature on Human-AI collaboration, Computational Creativity, and HCI, advancing our understanding of control in co-creative human-AI partnerships.
日本語まとめ
説明
AI coding assistants are changing how software engineers engage in coding work. This shift raises a key question: does the changing of coding work also alter how software engineers evaluate and demonstrate coding expertise? We explore this question through a simulated live coding interview involving two software engineers, one as evaluator and the other as candidate, with AI tools allowed. Participants continued to rely on familiar criteria but adjusted the evidence they sought, as AI assistants both introduced new forms of demonstrating expertise and obscured some established workflows. The importance of these evolving enactions varied with evaluators’ emphasis on implementation versus planning. Lacking a clear link to expertise, heightened productivity expectations created additional tensions around these evolving enactions. We conclude by discussing how extended enactions can be supported through AI-focused tools and training, and how tensions between diminished enactions and productivity call for collaborative attention.
日本語まとめ
説明
Large language models (LLMs) are being introduced into the public sector – for example, to assist caseworkers in making decisions on citizens’ cases. However, there is limited knowledge of how LLM tools can be used effectively in this complex task, including legal and cultural variables. This qualitative study foregrounds the perspectives of caseworkers from a Finnish public institution to dismantle their decision-making process and to build nuanced understanding on which sub-tasks of the process could benefit from the use of LLMs and how. To suggest meaningful uses for LLMs in the public sector, decision-making needs to be understood as a process that consists of several parts and that varies considerably in different contexts. We contribute to the fields of human–computer interaction and public administration by detailing the decision-making process of caseworkers and their perspectives on technological assistance, to suggest practical integration possibilities for LLM tools.
日本語まとめ
説明
Large Language Models (LLMs) are becoming increasingly ubiquitous in daily life, impacting decision-making across various domains. A substantial body of prior work has shown that individuals tend to evaluate positive predictions more favorably than negative ones---a phenomenon often referred to as the personal validation effect---across various non-AI prediction sources. Building on this foundation, we extend this well-established psychological effect to the context of LLM-based predictions, examining how prediction valence influences users’ perceptions when the source is an AI system. We investigate how positive AI-generated responses affect perceived validity, personalization, reliability, and usefulness of chatbot predictions, even when those predictions are fictitious and pre-scripted. In a study of 238 participants, positive predictions were perceived as significantly more valid (36% increase), personalized (42% increase), reliable (27% increase), and useful (22% increase) than negative predictions. These findings demonstrate that the personal validation effect persists in interactions with LLMs and underscore the substantial role of prediction valence in shaping user perceptions, with important implications for the design and deployment of AI systems across diverse applications.