arXiv subjects

Ned Cooper

Publications and source records attributed to Ned Cooper.

14 recordsLinked to original sources

arXiv2026

"Death by a thousand taxonomies?": AI Risk Classification In Practice

The harms in which AI is implicated range in nature and scope from unsafe user interactions through to the societal-wide consequences of AI adoption. Classification of the diverse risks of AI is foundational to AI governance: regulators, technology firms, and policymakers need structured accounts of risk upon which to act. Researchers and practitioners have accordingly developed many Sociotechnical Outcome Taxonomies (SOT). This paper presents an empirical study of SOT development and use, drawing on 25 interviews with researchers and practitioners across industry, academia, civil society, and government. We find SOT are weakly integrated into AI governance processes, and identify two features of SOT design and use that explain why. First, the design choices through which SOT produce structured representations of the complex problem space of AI risks tend to be invisible to downstream taxonomy users. Those users treat the resulting categories as exhaustive accounts of risk rather than as interpretive aids. Second, SOT typically enumerate harms without linking them to decision points or actors implicated in their occurrence, leaving accountability difficult to assign. We close with design recommendations for SOT developers and users, and argue realising the potential of SOT requires governance infrastructure that does not yet exist.

cs.CY↗

arXiv2026

Large Language Models Persuade Without Planning Theory of Mind

A growing body of work attempts to evaluate the theory of mind (ToM) abilities of humans and large language models (LLMs) using static, non-interactive question-and-answer benchmarks. However, theoretical work in the field suggests that first-personal interaction is a crucial part of ToM and that such predictive, spectatorial tasks may fail to evaluate it. We address this gap with a novel ToM task that requires an agent to persuade a target to choose one of three policy proposals by strategically revealing information. Success depends on a persuader's sensitivity to a given target's knowledge states (what the target knows about the policies) and motivational states (how much the target values different outcomes). We varied whether these states were Revealed to persuaders or Hidden, in which case persuaders had to inquire about or infer them. In Experiment 1, participants persuaded a bot programmed to make only rational inferences. LLMs excelled in the Revealed condition but performed below chance in the Hidden condition, suggesting difficulty with the multi-step planning required to elicit and use mental state information. Humans performed moderately well in both conditions, indicating an ability to engage such planning. In Experiment 2, where a human target role-played the bot, and in Experiment 3, where we measured whether human targets' real beliefs changed, LLMs outperformed human persuaders across all conditions. These results suggest that effective persuasion can occur without explicit ToM reasoning (e.g., through rhetorical strategies) and that LLMs excel at this form of persuasion. Overall, our results caution against attributing human-like ToM to LLMs while highlighting LLMs' potential to influence people's beliefs and behavior.

cs.CL↗

arXiv2026

Framing Responsible Design of AI for Mental Well-Being: AI as Primary Care, Nutritional Supplement, or Yoga Instructor?

Millions of people now use non-clinical Large Language Model (LLM) tools like ChatGPT for mental well-being support. This paper investigates what it means to design such tools responsibly, and how to operationalize that responsibility in their design and evaluation. By interviewing experts and analyzing related regulations, we found that designing an LLM tool responsibly involves: (1) Articulating the specific benefits it guarantees and for whom. Does it guarantee specific, proven relief, like an over-the-counter drug, or offer minimal guarantees, like a nutritional supplement? (2) Specifying the LLM tool's "active ingredients" for improving well-being and whether it guarantees their effective delivery (like a primary care provider) or not (like a yoga instructor). These specifications outline an LLM tool's pertinent risks, appropriate evaluation metrics, and the respective responsibilities of LLM developers, tool designers, and users. These analogies - LLM tools as supplements, drugs, yoga instructors, and primary care providers - can scaffold further conversations about their responsible design.

cs.HC↗

arXiv2025

Do Large Language Models Have a Planning Theory of Mind? Evidence from MindGames: a Multi-Step Persuasion Task

Recent evidence suggests Large Language Models (LLMs) display Theory of Mind (ToM) abilities. Most ToM experiments place participants in a spectatorial role, wherein they predict and interpret other agents' behavior. However, human ToM also contributes to dynamically planning action and strategically intervening on others' mental states. We present MindGames: a novel `planning theory of mind' (PToM) task which requires agents to infer an interlocutor's beliefs and desires to persuade them to alter their behavior. Unlike previous evaluations, we explicitly evaluate use cases of ToM. We find that humans significantly outperform o1-preview (an LLM) at our PToM task (11% higher; $p=0.006$). We hypothesize this is because humans have an implicit causal model of other agents (e.g., they know, as our task requires, to ask about people's preferences). In contrast, o1-preview outperforms humans in a baseline condition which requires a similar amount of planning but minimal mental state inferences (e.g., o1-preview is better than humans at planning when already given someone's preferences). These results suggest a significant gap between human-like social reasoning and LLM abilities.

cs.CL↗

arXiv2025

Designing Speech Technologies for Australian Aboriginal English: Opportunities, Risks and Participation

In Australia, post-contact language varieties, including creoles and local varieties of international languages, emerged as a result of forced contact between Indigenous communities and English speakers. These contact varieties are widely used, yet are poorly supported by language technologies. This gap presents barriers to participation in civil and economic society for Indigenous communities using these varieties, and reproduces minoritisation of contemporary Indigenous sociolinguistic identities. This paper concerns three questions regarding this context. First, can speech technologies support speakers of Australian Aboriginal English, a local indigenised variety of English? Second, what risks are inherent in such a project? Third, what technology development practices are appropriate for this context, and how can researchers integrate meaningful community participation in order to mitigate risks? We argue that opportunities do exist -- as well as risks -- and demonstrate this through a case study exploring design practices in a real-world project aiming to improve speech technologies for Australian Aboriginal English. We discuss how we integrated culturally appropriate and participatory processes throughout the project. We call for increased support for languages used by Indigenous communities, including contact varieties, which provide practical economic and socio-cultural benefits, provided that participatory and culturally safe practices are enacted.

cs.CL↗

arXiv2024

Troubling Taxonomies in GenAI Evaluation

To evaluate the societal impacts of GenAI requires a model of how social harms emerge from interactions between GenAI, people, and societal structures. Yet a model is rarely explicitly defined in societal impact evaluations, or in the taxonomies of societal impacts that support them. In this provocation, we argue that societal impacts should be conceptualised as application- and context-specific, incommensurable, and shaped by questions of social power. Doing so leads us to conclude that societal impact evaluations using existing taxonomies are inherently limited, in terms of their potential to reveal how GenAI systems may interact with people when introduced into specific social contexts. We therefore propose a governance-first approach to managing societal harms attended by GenAI technologies.

cs.HC↗

arXiv2024

Constraining Participation: Affordances of Feedback Features in Interfaces to Large Language Models

Large language models (LLMs) are now accessible to anyone with a computer, a web browser, and an internet connection via browser-based interfaces, shifting the dynamics of participation in AI development. This article examines how interactive feedback features in ChatGPT's interface afford user participation in LLM iteration. Drawing on a survey of early ChatGPT users and applying the mechanisms and conditions framework of affordances, we analyse how these features shape user input. Our analysis indicates that these features encourage simple, frequent, and performance-focused feedback while discouraging collective input and discussions among users. Drawing on participatory design literature, we argue such constraints, if replicated across broader user bases, risk reinforcing power imbalances between users, the public, and companies developing LLMs. Our analysis contributes to the growing literature on participatory AI by critically examining the limitations of existing feedback processes and proposing directions for redesign. Rather than focusing solely on aligning model outputs with specific user preferences, we advocate for creating infrastructure that supports sustained dialogue about the purpose and applications of LLMs. This approach requires attention to the ongoing work of "infrastructuring" - creating and sustaining the social, technical, and institutional structures necessary to address matters of concern to stakeholders impacted by LLM development and deployment.

cs.HC↗

arXiv2024

Understanding and Shaping Human-Technology Assemblages in the Age of Generative AI

Generative AI capabilities are rapidly transforming how we perceive, interact with, and relate to machines. This one-day workshop invites HCI researchers, designers, and practitioners to imaginatively inhabit and explore the possible futures that might emerge from humans combining generative AI capabilities into everyday technologies at massive scale. Workshop participants will craft stories, visualisations, and prototypes through scenario-based design to investigate these possible futures, resulting in the production of an open-annotated scenario library and a journal or interactions article to disseminate the findings. We aim to gather the DIS community knowledge to explore, understand and shape the relations this new interaction paradigm is forging between humans, their technologies and the environment in safe, sustainable, enriching, and responsible ways.

cs.HC↗

arXiv2024

From Fitting Participation to Forging Relationships: The Art of Participatory ML

Participatory machine learning (ML) encourages the inclusion of end users and people affected by ML systems in design and development processes. We interviewed 18 participation brokers -- individuals who facilitate such inclusion and transform the products of participants' labour into inputs for an ML artefact or system -- across a range of organisational settings and project locations. Our findings demonstrate the inherent challenges of integrating messy contextual information generated through participation with the structured data formats required by ML workflows and the uneven power dynamics in project contexts. We advocate for evolution in the role of brokers to more equitably balance value generated in Participatory ML projects for design and development teams with value created for participants. To move beyond `fitting' participation to existing processes and empower participants to envision alternative futures through ML, brokers must become educators and advocates for end users, while attending to frustration and dissent from indirect stakeholders.

cs.HC↗

arXiv2024

"It's how you do things that matters": Attending to Process to Better Serve Indigenous Communities with Language Technologies

Indigenous languages are historically under-served by Natural Language Processing (NLP) technologies, but this is changing for some languages with the recent scaling of large multilingual models and an increased focus by the NLP community on endangered languages. This position paper explores ethical considerations in building NLP technologies for Indigenous languages, based on the premise that such projects should primarily serve Indigenous communities. We report on interviews with 17 researchers working in or with Aboriginal and/or Torres Strait Islander communities on language technology projects in Australia. Drawing on insights from the interviews, we recommend practices for NLP researchers to increase attention to the process of engagements with Indigenous communities, rather than focusing only on decontextualised artefacts.

cs.CL↗

arXiv2023

From human-centered to social-centered artificial intelligence: Assessing ChatGPT's impact through disruptive events

Large language models (LLMs) and dialogue agents represent a significant shift in artificial intelligence (AI) research, particularly with the recent release of the GPT family of models. ChatGPT's generative capabilities and versatility across technical and creative domains led to its widespread adoption, marking a departure from more limited deployments of previous AI systems. While society grapples with the emerging cultural impacts of this new societal-scale technology, critiques of ChatGPT's impact within machine learning research communities have coalesced around its performance or other conventional safety evaluations relating to bias, toxicity, and "hallucination." We argue that these critiques draw heavily on a particular conceptualization of the "human-centered" framework, which tends to cast atomized individuals as the key recipients of technology's benefits and detriments. In this article, we direct attention to another dimension of LLMs and dialogue agents' impact: their effects on social groups, institutions, and accompanying norms and practices. By analyzing ChatGPT's social impact through a social-centered framework, we challenge individualistic approaches in AI development and contribute to ongoing debates around the ethical and responsible deployment of AI systems. We hope this effort will call attention to more comprehensive and longitudinal evaluation tools (e.g., including more ethnographic analyses and participatory approaches) and compel technologists to complement human-centered thinking with social-centered approaches.

cs.CY↗

arXiv2023

Would the Trees Dim the Lights? Adopting the Intentional Stance for More-Than-Human Participatory Design

The 2019/20 Black Summer bushfires in Australia demonstrated the brutal and disastrous consequences of changing the technological world without considering linkages with the biophysical, ecological or human worlds. An emerging more-than-human design philosophy encourages designers to consider such interrelations between humans and non-human entities. Yet, the design research community has focused on situated or embodied experiences for designers, rather than developing processes to legitimate the perspectives of non-human entities through participatory design. This paper explores how adopting the `intentional stance', a concept from philosophy, might provide a heuristic for more-than-human participatory design. Through experimentation with the intentional stance in the context of smart lighting systems, the paper demonstrates that the approach has potential for non-human entities from the ecological world, but less so for the biophysical world. The paper concludes by encouraging critique and evolution of the intentional stance, and of other approaches, to legitimate the perspectives of non-human entities in everyday design.

cs.HC↗

arXiv2023

Bridging Deliberative Democracy and Deployment of Societal-Scale Technology

This position paper encourages the Human-Computer Interaction (HCI) community to focus on designing deliberative processes to inform and coordinate technology and policy design for large language models (LLMs) -- a `societal-scale technology'. First, I propose a definition for societal-scale technology and locate LLMs within this definition. Next, I argue that existing processes to ensure the safety of LLMs are insufficient and do not give the systems democratic legitimacy. Instead, we require processes of deliberation amongst users and other stakeholders on questions about the safety of outputs and deployment contexts. This shift in AI safety research and practice will require the design of corporate and public policies that determine how to enact deliberation and the design of interfaces and technical features to translate the outcomes of deliberation into technical development processes. To conclude, I propose roles for the HCI community to ensure deliberative processes inform technology and policy design for LLMs and other societal-scale technology.

cs.CY↗

arXiv2022

A Systematic Review and Thematic Analysis of Community-Collaborative Approaches to Computing Research

HCI researchers have been gradually shifting attention from individual users to communities when engaging in research, design, and system development. However, our field has yet to establish a cohesive, systematic understanding of the challenges, benefits, and commitments of community-collaborative approaches to research. We conducted a systematic review and thematic analysis of 47 computing research papers discussing participatory research with communities for the development of technological artifacts and systems, published over the last two decades. From this review, we identified seven themes associated with the evolution of a project: from establishing community partnerships to sustaining results. Our findings suggest that several tensions characterize these projects, many of which relate to the power and position of researchers, and the computing research environment, relative to community partners. We discuss the implications of our findings and offer methodological proposals to guide HCI, and computing research more broadly, towards practices that center communities.

cs.HC↗