12,445 papers · continuously updated · last export: 10 Aug 2026livingmeta.ai
← Browse all papers
AI evidence extraction

Trustworthy AI to conduct literature reviews

Isabelle Walsh · Journal of the Association for Information Systems · 2025

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

9/10
Relevance
1/4
Quality (LMQS)
D
Evidence
0
Citations
0.00
FWCI

Methodology & findings

Study design

Descriptive/demonstration study with comparative benchmarking.

Main result

The paper presents that "Hybrid neuro-symbolic systems, which combine transparently logical reasoning for clustering literature with pattern recognition for synthesizing it" can address limitations in current AI approaches for literature reviews. The focus is on ARTIREV, described as "a hybrid AI tool integrating a bibliometric expert system coupled with fine-tuned generative AI (SOCRATES) offering improved transparency, exhaustivity, and reliability."

Research paradigm

pragmatist

Author conclusions

The authors conclude that hybrid neuro-symbolic systems represent an improved approach to AI-supported literature reviews. They propose that "Hybrid neuro-symbolic systems, which combine transparently logical reasoning for clustering literature with pattern recognition for synthesizing it" offer solutions to current limitations in AI-supported academic literature review workflows.

Risk of bias

Lack of systematic comparison protocol - attendees choose comparison baselines; Potential selection bias in choosing 'any other AI solution of their choice'; No mention of blinding or objective evaluation criteria; Single tool focus (ARTIREV) may introduce promotional bias

Limitations

  • The abstract highlights "strengths and limitations of current AI approaches" but does not explicitly state specific limitations of the proposed ARTIREV tool itself
  • The paper notes it will "critically assess the capabilities of the proposed tool," suggesting limitations exist but are not detailed in the abstract.

Open questions raised

  • The abstract identifies gaps in current AI approaches for literature reviews, noting limitations in system transparency, input data handling, user control, and output presentation. The need for hybrid neuro-symbolic systems that combine logical reasoning with pattern recognition is highlighted as an unmet need.
Data: not_statedCode: not_statedExtracted from: pdfAgreement 71%

Explore related topics

Systematic Review AutomationLiterature DiscoveryBenchmarking

Related papers

← Browse all papers