Artificial Intelligence-based tools in the context of Open Science

Abstract

This poster presents possibilities and issues concerning the new wave of Artifical Intelligence (AI)-based services and tools entering academia aimed to help researchers and students. Iris.ai, semanticscholar.org, yewno.com, connectedpapers.com, researchrabbit.ai, www.paper-digest.com, openknowled-gemaps.org, and keenious.com, are all examples of services part of a complex landscape of educational and research support resources driven by AI. Unfortunately, researchers and students have very few arenas to learn about all the aspects of using those resources. Trust, ethics, interpretability and reliability, are all topics to be addressed when using those tools (Guidotti et al., 2018). Moreover, the lack of possibility to test and influence how literature is analyzed and new knowledge created by AI-based services, is an emerging concern in various academic libraries (Gasparini & Kautonen, 2016). The increasing production of fake science in AI-based papermills implies new challenges for academic quality control and for the accountability and reliability of research as a whole (Løkeland-Stai, 2022). The poster proposes to use the site PhD-on-track (https://www.phdontrack.net/, Faber et al., 2018), one of the preferred starting points for new PhD candidates and early career researchers, to contextualize AI-based services in researchers’ literature search and analysis. PhD on Track aims to enable PhD candidates from all academic fields to easily access information on different aspects of open science and support academic integrity in their research practices. By addressing AI-based services and tools in an early stage, PhD on Track will contribute to avoid opacity and clarify non-intuitive aspects of the use of technology in research. Furthermore, we argue for the possibility that these technical innovations will change the workflows of researchers and students. Academia needs to react to this development, offer a framework and support a shift of focus on AI-based services from the micro level (users) to a wider institutional one (the university). The poster will present the following issues and questions, with respective possible reactions and solutions: What is the minimum level of competence PhD candidates should have about machine learning and deep learning? How should PhD candidates choose reliable AI-based services? When and where should PhD candidates and researchers gather reliable information about AI-based services? How should academic libraries support access to repositories of Open Access articles? How should universities ensure the correct use of AI-based tools? How can universities and academic libraries address ethical questions which may be raised by the increasing availability and use of AI-based tools?