AI-Assisted Literature Review

The U.S. Geological Survey’s National Climate Adaptation Science Center (NCASC) is seeking a student contractor to test the use of AI tools for science synthesis projects focused on climate change impacts to fish and wildlife.

Research Project: Systematic literature reviews are a useful way to understand the state of knowledge on topics of interest. To remain useful, they need to be updated regularly as new information becomes available. The rapid pace of new publications makes it increasingly difficult to review all relevant papers and produce timely updates. Advances in AI/ML may be able to speed up reviews by screening out irrelevant articles and/or by extracting data from relevant papers potentially improving the efficiency and sustainability of USGS synthesis products. We will build on past USGS efforts to develop a transferable workflow using AI/ML to dynamically update reviews, using a systematic review on species range shifts as a case study (https://doi.org/10.1186/s13750-023-00296-0). From this review, we have a list of included and excluded articles, as well as the data we extracted from original research papers. We will look at what kinds of data AI can extract well, where it does not work well, and compare error rates to humans. Results will be shared via publications, open-source workflows, and presentations. Lessons learned will help USGS researchers apply AI/ML to diverse synthesis efforts to improve efficiency.

Types of Services Required:

Services may include one or more of the following activities: 

1. Create training and testing data from an existing range shift database;
2. Use AI tools to extract data from the articles included in the original review;
3. Compare AI results against human decisions, including performance metrics such as error rates, precision, and recall;
4. Document methods, tools, and evaluation results;
5. Contribute to publications, presentations, and workflow documentation as appropriate.

Required Expertise/Skills:

Enrolled in a bachelor’s degree in biology, ecology, environmental science, computer science, or related field. Knowledge of scientific disciplines related to ecology and climate change. Basic knowledge of statistics. Skill in conducting literature searches, reading and interpreting scientific papers, and using AI. Experience writing scientific papers. Knowledge of basic computer operations and Microsoft Office.

Compensation:

$21.49 per hour for a student who has completed a 2+ years of college. (Note compensation rate depends on location and may vary if students are not based in the Washington D.C. region).

Hours: 

20 hours per week during the fall semester.

Application materials:

• Letter of interest (describe interest in the program; detail any relevant experience) 
• Current resume
• (Un)official transcripts

Deadline:

Please send application materials to Sarah Weiskopf (sweiskopf@usgs.gov) or on Handshake by Friday, August 21.