Still Not a Remedy for Academics: The Use of Generative AI-Powered Tools in Bibliometric Analysis

Abstract

Artificial intelligence (AI) based tools hold great promise for automating time-consuming and labor-intensive tasks where humans are prone to making mistakes. So far, AI solutions are actively used in healthcare, IT, and large data processing. Yet, their ongoing adoption across all areas of everyday lives – by both professionals and individuals – raises ethical concerns, before all in education, where a risk of misjudgments or incorrect data manipulations remains high. However, the up-to-date literature barely addresses the problems of AI integration in real-world settings for academic and teaching purposes, focusing on generalized literature reviews. At the same time, some findings report the problem of large language models (LLM), which are known for persistent hallucinations. In this preliminary paper, we portrait the existing AI-powered platforms for bibliometric analysis – both traditional academic search engines with generative extensions (Scopus AI, WoS Research Assistant, etc.) and commonly used LLM chatbots. Our major outcome is that no one of the solutions on the market is able to substitute researchers in text-mining, but rather a combination of different tools may simplify the routine tasks. We also admit that more research is needed to compare existing tools in different settings, with different data sources and bibliometric analysis tasks, in order to develop evidence-based guidelines and pedagogical safeguards to support responsible integration of AI tools into scholarly and educational contexts.

Ähnliche Arbeiten