AI evidence extraction

Artificial intelligence use and scientific innovation

Yuanyuan Liu, Yundong Xie, Xiaobei Shen, Dengsheng Wu · Journal of the Association for Information Science and Technology · 2025

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

0.86

FWCI

Top 10%

Impact

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.1002/asi.70043

Methodology & findings

Study design

Metric-based observational study using bibliometric analysis.

Sample

not–reported

Primary method

Bibliometric analysis using AI use score metric (term-based and knowledge-based dimensions); statistical association testing between AI use and innovation outcomes; temporal trend analysis; text similarity analysis. Specific statistical tests and software not detailed in abstract.

Main result

The study found that "AI use is positively and significantly associated with both scientific disruption and novelty, with this relationship being particularly pronounced in STEM disciplines." Additionally, "combining both dimensions of AI use shows the strongest correlation with innovative outcomes," with "term-based AI use is more strongly linked to disruptive innovation, while knowledge-based AI use is more closely associated with scientific novelty."

Reports effect sizes.

Research paradigm

Positivist/Quantitative empiricism

Author conclusions

The authors conclude that "AI use is positively and significantly associated with both scientific disruption and novelty, with this relationship being particularly pronounced in STEM disciplines" and that "combining both dimensions of AI use shows the strongest correlation with innovative outcomes," suggesting that "term-based AI use is more strongly linked to disruptive innovation, while knowledge-based AI use is more closely associated with scientific novelty."

Risk of bias

Selection bias: Study relies on bibliometric data and may not capture all AI use in scientific research; Measurement bias: AI use score based on explicit terms and citations may undercount implicit AI applications; Confounding: Other factors influencing scientific innovation not controlled in the analysis; Selection bias in document selection and what constitutes 'AI-related' terms and literature; Measurement bias in term-based scoring dependent on terminology choices in abstracts/titles; Temporal confounding: cannot establish causality from observational data; Discipline-specific variation in AI terminology usage and documentation practices; Citation practices may not fully capture knowledge-based AI use; Not reported in the abstract. Potential confounders such as research funding, institutional prestige, or disciplinary norms are not addressed in the provided text.

Open questions raised

The authors note that while prior studies have explored AI applications across disciplines, "an understanding of whether AI use contributes to scientific innovation remains limited," suggesting this gap motivated their research.
The abstract indicates that prior understanding of "whether AI use contributes to scientific innovation remains limited," suggesting this gap motivated the current study. Future research directions are not explicitly stated in the abstract.

Data: not_statedCode: not_statedExtracted from: pdfAgreement 67%

Explore related topics

Productivity Effects Research Productivity Empirical Study