Summary
The episode works through four questions that students often find difficult when they use AI tools to search and read the research literature. It uses made-up examples to separate the precision of an AI search tool's ranked top set from its low and unknown recall, and to show how a known-item test exposes a NOT line that silently removes relevant records from an AI-drafted PubMed search. It then works out why an extraction accuracy of about 96 percent still calls for checking every value, showing that a ten-value spot check would come back clean about two times in three from a table holding about four errors, and it corrects a summary that reported a share of completers as a share of everyone referred. In a citation audit, Sarah mistakes a secondary citation for an accurate one before tracing the finding to an original study without a comparison group; because no source supports the claim as the report worded it, the hosts classify the citation as inaccurate, rewrite the claim and cite the original. The hosts end by debating whether a student should begin a new topic with a deep research report and agree on a rule for using such reports as a source of vocabulary and leads.