Evaluating generative AI for qualitative data extraction in community-based fisheries management literature
Bibliographic record
Abstract
Uptake of AI tools in knowledge production processes is rapidly growing. In this pilot study, we explore the ability of generative AI tools to reliably extract qualitative data from a limited sample of peer-reviewed documents in the context of community-based fisheries management (CBFM) literature. Specifically, we evaluate the capacity of multiple AI tools to analyse 33 CBFM papers and extract relevant information for a systematic literature review, comparing the results to those of human reviewers. We address how well AI tools can discern the presence of relevant contextual data, whether the outputs of AI tools are comparable to human extractions, and whether the difficulty of question influences the performance of the extraction. While the AI tools we tested (GPT4-Turbo and Elicit) were not reliable in discerning the presence or absence of contextual data, at least one of the AI tools consistently returned responses that were on par with human reviewers. These results highlight the potential utility of AI tools in the extraction phase of evidence synthesis for supporting human-led reviews, while underscoring the ongoing need for human oversight. This exploratory investigation provides initial insights into the current capabilities and limitations of AI in qualitative data extraction within the specific domain of CBFM, laying groundwork for future, more comprehensive evaluations across diverse fields and larger datasets.
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.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.360 | 0.736 |
| Meta-epidemiology (narrow) | 0.002 | 0.002 |
| Meta-epidemiology (broad) | 0.002 | 0.004 |
| Bibliometrics | 0.028 | 0.019 |
| Science and technology studies | 0.003 | 0.004 |
| Scholarly communication | 0.012 | 0.009 |
| Open science | 0.005 | 0.011 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.007 | 0.001 |
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 itClassification
machine, unvalidatedMachine predicted; the direct Gemma label and the distilled Codex classifier agree on what is shown here.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".