CIDDL Community Chat: AI for Reviewing Literature
Author: Teddy Kim; info@ciddl.org
On a recent episode of the CIDDL AI Community Office Hours on December 10th, 2024, Dr. Katie Tuck, an associate professor at the University of Kansas, shared her expertise in utilizing AI tools for literature reviews. The session covered innovative AI tools that simplify locating, analyzing, and synthesizing academic literature. This post synthesizes key insights from the event.
Why Access Support for Reviewing Literature?
Conducting a thorough literature review can be time-consuming and complex due to several challenges. Locating relevant articles and identifying appropriate search terms often requires significant time and prior domain knowledge. Additionally, effective searches demand familiarity with domain-specific terminology, which can be a barrier for researchers new to a field. Access to research literature also depends on institutional subscriptions, which vary by university or library, and even with identical databases and keywords, results can differ significantly. AI tools help bypass these barriers by efficiently locating and synthesizing relevant literature, making the review process more streamlined and accessible.
AI to Support Finding Literature
Dr. Tuck highlighted several AI tools that can assist researchers in locating, analyzing, and synthesizing relevant literature. Let’s explore their unique features and benefits.
Consensus is an academic search engine that simplifies finding literature by summarizing key points from research articles. Users enter search terms to receive summaries and a curated list of relevant articles. It's useful for those without access to subscription databases, providing a quick overview of topics and identifying quality literature. Key features include generating summaries for quick comprehension and easy article discovery. However, free access has limited features, and a subscription is needed for advanced features.
Connected Papers visualizes relationships between academic articles, revealing citation networks and thematic connections. Users can get a generated visual map of related papers based on citations and content similarities by inputting an article title. This tool helps users explore broader topics, identify influential works, and find related research. Key features include visualizing article connections and highlighting citation networks. However, exporting mapped data requires a subscription, and free accounts have limited access to advanced features.
Covidence is a tool for managing systematic reviews that helps apply inclusion and exclusion criteria consistently. Based on predefined criteria, users upload references and screen articles in two stages: title/abstract and full text. The tool streamlines the process, supports collaboration, and ensures consistency. It also features automated agreement calculations, detailed note-taking, and progress tracking. However, modifying criteria mid-review requires restarting the process, which can be time-consuming.
Elicit aids in summarizing and synthesizing academic articles for literature reviews. Users can input articles or search within the tool, which organizes direct quotations, summaries, methodologies, and outcomes into a spreadsheet. Key features include exporting data in Excel or CSV format, customizable columns for research needs, and the ability to search for specific findings. However, access is expensive, and users must paraphrase or cite quotes to prevent plagiarism.
Q&A Highlights
Systematic Review Guidance: Transparency and Clarity
Dr. Tuck emphasized the importance of transparency and clarity when using AI tools in systematic reviews. The process should be well-documented to ensure replication and credibility. New Prisma 2020 guidelines now include a column for alternative data collection methods, legitimizing AI-assisted literature searches.
Cost and Accessibility
Many tools, such as Covidence, require subscriptions. Dr. Tuck shared strategies for overcoming these costs, such as utilizing institutional funds or professional development budgets, writing tool subscriptions into research grants, and creating shared accounts to maximize access within a research group.
Ethical Considerations
Concerns about data privacy and proprietary rights were also addressed. To mitigate ethical dilemmas, users should understand user agreements and data policies, be cautious when uploading data to AI tools, and advocate for institutional trials or subscriptions.
Opportunities and Challenges
AI tools are transforming how researchers approach literature reviews. These tools offer innovative solutions to longstanding challenges. However, ethical considerations and cost barriers remain critical factors to navigate. As Dr. Tuck pointed out, transparency and clarity are key to leveraging these technologies effectively.
Keep the Conversation Going!
How can AI tools reshape your literature review process? Share your experiences and insights in our CIDDL community!