Understanding User Needs in AI-Assisted News Reading

Context 🧭

As generative AI tools like LLM-powered chatbots become more integrated into everyday life, they are starting to reshape how people access, interpret, and engage with news. This shift has sparked growing interest among media researchers and technology developers in understanding the role of AI in modern news consumption experiences.

For news platforms and AI product teams, this presents a critical business opportunity: to design tools that not only improve information sharing but also build trust, support diverse user needs, and increase engagement and retention.

However, we still know little about users’ actual experiences and needs when it comes to AI-assisted news reading, especially across different user groups.

This study explores the potential of AI chatbots to enhance the news reading experience by focusing on two user groups: local readers and immigrant readers.

Research Questions 🔍

  1. What are the opportunities and challenges of using AI chatbots to support news reading?
  2. Do local and immigrant readers have distinct needs and expectations when engaging with AI-assisted news?

Goal 🎯

This study aimed to uncover diverse users’ preferences, expectations, and concerns when interacting with AI-assisted news content. The findings are intended to inform the inclusive and responsible design of future generative AI tools for media consumption.

Method 🧪

I conducted an online experiment to explore and quantify differences in user needs between local and immigrant news readers.

Current project is an informative study using quantitative method.

Outcome 📝

  • Presented findings at an international academic conference attended by researchers across disciplines and industry practitioners.
  • Published at ACM CHI 2025, the premier conference for Human-Computer Interaction research.
  • Invited to contribute an article to  ACM XRDS Magazine (currently in publication).

Impact 🌍

  • Research paper received over 350 downloads within 6 weeks of publication, indicating strong interest across academic and industry audiences.
  • Provided foundational insights for the development of generative AI-supported news reading tools.
  • Helped inform strategic planning for inclusive newsreader technologies for media practitioners.

Team 🤝

This project was a collaborative effort led by Yongle, with support from a cross-disciplinary team:

  • Two graduate students in Information Science, mentored by Yongle, contributed to data collection and partial data analysis.
  • Three undergraduate research assistants from Computer Science and Information Science supported data cleaning and translation of task materials.
  • A faculty advisor provided high-level guidance and feedback on study design and research direction.

My Role 🧑‍💻

Responsible for guiding the study from concept to execution:

  • Defined and scoped the problem space, framing key research questions and planning the project timeline.
  • Selected the online experiment as the primary method to explore user experiences at scale.
  • Led data collection in collaboration with graduate and undergraduate research assistants.
  • Conducted thematic analysis of qualitative responses and performed statistical analysis to uncover key news reading patterns.
  • Synthesized findings and shared actionable insights with academic and media practitioners.

Study Procedure

Our team recruited 120 participants from immigrant and local newsreaders to perform a one-hour online task.

Screenshot of Task interface.

Participants interacted with an AI chatbot to read and ask questions about news articles. During the session, I collected:

  • Log data capturing the questions participants asked the AI chatbot
  • In-task survey responses measuring their perceptions, goals, and satisfaction

This mixed dataset allowed for both behavioral analysis and quantitative comparison across user groups.

Key Findings and Implications 🔍

Visualization of newsreader questions by category.
  • Developed a typology of reader questions that categorizes user needs into four high-level types. This is the first comprehensive framework that captures the range of needs users express when interacting with AI-supported news tools, which offers a foundational lens for future design.
  • Visualized differences in question types between immigrant and local participants, showing distinct patterns in how each group engaged with the AI chatbot.
  • Statistical analysis confirmed meaningful group-level differences: Immigrant participants were less likely to ask analytical questions They were more likely to seek guidance or support from the AI

These insights suggest a valuable design direction: future AI tools should be tailored to support critical thinking and news quality evaluation among user groups such as immigrants. AI agents should help all news readers properly calibrate their trust toward AI responses when offering suggestions.

Return to the Portfolio Main Page