They Think AI Can Do More Than It Actually Can: Practices, Challenges, & Opportunities of AI-Supported Reporting In Local Journalism

Paper Title

They Think AI Can Do More Than It Actually Can: Practices, Challenges, & Opportunities of AI-Supported Reporting In Local Journalism

Publication Info

• Topic area: AI-supported reporting in local journalism and its socio-technical implications.
• Keywords: AI-supported reporting, local journalism, data literacy, automation, socio-technical systems, human-AI collaboration, newsrooms, data challenges, journalistic values, HCI.

Background and Problem

• Problem / challenge: Local journalists face challenges in working with digital data and AI due to limited technical expertise, lack of confidence, and resource constraints. Existing studies overlook the specific needs of non-technical local journalists in Germany.
• Significance: Local journalism is critical for democracy, providing unique coverage of underreported issues. However, declining revenues and shrinking newsrooms threaten its sustainability. AI could enhance efficiency and support data-driven reporting, but its adoption remains limited.
• Motivation and related work: Prior research has explored AI in journalism broadly but has not focused on local journalists’ interactions with data and AI. Studies highlight low data literacy among journalists and a lack of tools tailored to their needs. This paper addresses these gaps by examining how AI can support local journalism in Germany.

Solution

• Proposed approach: The study investigates AI-supported reporting through semi-structured interviews with 21 local journalists in Germany, using research prototypes to explore automation in data workflows.
• Novelty:
  1. Analysis of local journalists’ current use of data and AI, revealing limited engagement and challenges.
  2. Identification of socio-technical opportunities for AI-supported reporting, including automation of repetitive tasks and enhanced data workflows.
  3. Recommendations for designing AI tools that align with journalistic values and local news contexts.
• Procedure and key techniques:
  • Conducted 21 semi-structured interviews with local journalists.
  • Presented two research prototypes: one using "automating-through-demonstrations" and the other "automating-through-words."
  • Analyzed qualitative data using reflexive and axial coding to identify themes related to data practices, challenges, and opportunities.

Results

• Concrete findings:
  • Journalists primarily use AI for text-based tasks like summarization, brainstorming, and SEO but rarely for data-intensive workflows.
  • Key challenges include lack of data skills, low confidence, difficulty accessing reliable data, and mistrust of AI outputs.
  • Journalists envision AI as a tool for automating repetitive tasks, fact-checking, and enhancing investigative reporting.
• Advantage over baselines:
  • The study highlights the potential of AI to make data workflows accessible to non-technical journalists, enabling faster and more accurate reporting.
  • AI-supported tools can reduce the cognitive load and time required for data processing, allowing journalists to focus on in-depth reporting.
• Experiments / evaluation:
  • Prototypes demonstrated automation of online data collection, analysis, and visualization, eliciting positive feedback on their potential utility.
  • Participants expressed interest in tools that integrate local context and provide transparent reasoning for AI outputs.
• Limitations and future work:
  • Limited to local journalists in Germany; findings may not generalize globally.
  • Focused on non-technical journalists; future studies could explore perspectives of data journalists.
  • Calls for longitudinal studies and participatory design approaches to refine AI tools for local newsrooms.

Summary

This study examines how AI can support local journalists in Germany by addressing their challenges with data and automation. Findings reveal limited use of AI for data workflows, with journalists primarily relying on it for text-based tasks. Challenges include low data literacy, lack of confidence, and mistrust of AI outputs. The study proposes AI-supported tools to automate repetitive tasks, enhance data processing, and support investigative reporting, emphasizing the need for human oversight and local context. Recommendations include designing transparent, context-aware tools that align with journalistic values, promoting data literacy, and fostering collaboration in newsrooms. These insights aim to sustain local journalism’s role in democratic societies amidst growing digital challenges.