AI evidence extraction

Should We Collaborate with AI to Conduct Literature Reviews? Changing Epistemic Values in a Flattening World

Ojelanki Ngwenyama, Frantz Rowe · Journal of the Association for Information Systems · 2024

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

9/10

Relevance

0/4

Quality (LMQS)

Evidence

Citations

10.98

FWCI

Top 10%

Impact

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.17705/1jais.00869

Methodology & findings

Study design

Conceptual analysis and narrative review of AI collaboration in literature review processes, examining epistemic values at risk when using machine learning and generative AI tools at different review stages..

Main result

The paper finds that "although AI tools accelerate search and screening tasks, particularly when there are vast amounts of literature involved, they may compromise quality, especially when it comes to transparency and explainability." The authors note that "expert systems are less likely to have a negative impact on these tasks" and emphasize that "any AI method should preserve researchers' ability to critically select, analyze, and interpret the literature."

Research paradigm

Critical/interpretive epistemology examining epistemic values in AI-assisted research

Author conclusions

The authors conclude that collaboration with AI for literature reviews requires careful consideration of epistemic values: "any AI method should preserve researchers' ability to critically select, analyze, and interpret the literature," and they call for reflection on "the epistemic values at risk when using certain types of AI tools based on machine learning or generative AI at different stages of the review process."

Limitations

The paper does not conduct empirical testing of AI tools in actual literature review workflows
The authors acknowledge that "any AI method should preserve researchers' ability to critically select, analyze, and interpret the literature," implying that current tools may not adequately support this critical scholarly function, though specific empirical limitations of particular tools are not detailed in the abstract.

Open questions raised

The paper calls for further reflection on the epistemic values at risk when using AI tools in literature reviews, and identifies the need to understand how different AI approaches (expert systems versus machine learning/generative AI) affect scholarly quality, transparency, and explainability in the review process.
The authors call for further reflection on epistemic values at risk when using AI tools in literature reviews and identify the need for better understanding of how AI impacts transparency and explainability across different review stages.
The authors identify the need for further reflection on epistemic values at risk when using AI tools in literature reviews, and highlight the tension between efficiency gains and maintenance of research quality, transparency, and explainability.

Data: not_statedCode: not_statedExtracted from: pdfAgreement 79%

Explore related topics

Systematic Review Automation Literature Discovery Conceptual