Posted: March 2025
Contact: Max Callaghan, Jan Minx
Climate policies need to be informed by the best and most recent evidence. However, it is a
common problem at the science-policy interface that this evidence is often not available
when decision-makers need it. “Living evidence” defined as continually updated systematic
review evidence has been proposed to overcome this issue. However, the approach
becomes much more feasible, if parts of the systematic review process can be automated. In
this project, we will explore how Large Language Models can help to automate the
systematic review process using the example of carbon pricing. We will use an existing
dataset from a study by Döbbeling-Hildebrandt to explore to what extent and when the
process of study selection and data extraction can be safely automated in an effort to
minimise human supervision in living evidence systems
Required skills: Data science, policy evaluation
