Artificial intelligence behind the scenes: PubMed’s Best Match algorithm
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
This article focuses on PubMed's Best Match sorting algorithm, presenting a simplified explanation of how it operates and highlighting how artificial intelligence affects search results in ways that are not seen by users. We further discuss user search behaviors and the ethical implications of algorithms, specifically for health care practitioners. PubMed recently began using artificial intelligence to improve the sorting of search results using a Best Match option. In 2020, PubMed deployed this algorithm as the default search method, necessitating serious discussion around the ethics of this and similar algorithms, as users do not always know when an algorithm uses artificial intelligence, what artificial intelligence is, and how it may impact their everyday tasks. These implications resonate strongly in health care, in which the speed and relevancy of search results is crucial but does not negate the importance of a lack of bias in how those search results are selected or presented to the user. As a health care provider will not often venture past the first few results in search of a clinical decision, will Best Match help them find the answers they need more quickly? Or will the algorithm bias their results, leading to the potential suppression of more recent or relevant results?
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.017 | 0.038 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.000 | 0.007 |
| Insufficient payload (model declined to judge) | 0.003 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it