2026-10-27 –, LUMC06
It’s impossible to escape generative AI (genAI) in the modern research landscape; in fact, some of the very abstracts in the program may have been written by it! But for every time-saving benefit it has brought us, it has also exacerbated existing problems in publishing, like overburdening reproducibility mechanisms and boosting paper mills and fraud, as well as introducing new problems like publishing hallucinations and misinformation because authors might simply not have read what AI has produced before submitting.
Thankfully, a good solution framework already exists: persistent identifiers! Come join us, the Knowledge Exchange PIDs4AI working group – a collaboration of six European research infrastructure providers and research supporting organisations – and learn how PIDs could address the proliferation of genAI outputs and their opaque provenance with a value based approach that focuses on: transparency over detection, infrastructure over policy, normalisation over stigma, evolution over perfection and coordination over fragmentation.
Generative AI (genAI) has entered the broader research landscape with a bang, but thanks to PIDs, it does not need to be a death knell.
In the last couple of years, the developments of genAI have offered researchers some real-time saving benefits but at a significant cost. For example, traditional reproducibility mechanisms have become overwhelmed by AI-generated submissions, degrading trust. This also extends as genAI models cannot distinguish between good and bad research methods and may incorporate "paper mill" and other low-quality research articles into answers with equal or even higher weight due to their prevalence (i.e. the abundance of low quality articles) and as a result it may turbocharge the citation problems of the past. Additionally, as new genAI models are trained on ever more massive datasets, new problems may arise when the genAI model is: (1) using hallucinated works; (2) mixing original works with hallucination or mixing original works with each other; (3) using original works without permission; and (4) original authors being unaware their work is being used. This could lead to a scenario where there is research stagnation, propagation of misinformation, and increased distrust both within the research community and society at large.
However, this does not need to be the case! With PIDs, the research community can take a well-established and trusted framework and adapt it to include both the inputs and outputs of genAI as well as the models themselves, providing a stabilising force in these uncertain times. You will see a vision of how PIDs can address the proliferation of genAI outputs and their opaque provenance with a value-based approach that focuses on: transparency over detection, infrastructure over policy, normalisation over stigma, evolution over perfection, and coordination over fragmentation. You will be presented with an outline of how the existing technology and PID infrastructures can already address reproducibility and providence as well as be extended. We will also address the adoption pathways that exist for different stakeholders.
Come join us, get up to speed and help pave the way to a better research landscape with genAI instead of a fractured wasteland of slop and distrust!
Knowledge Exchange focuses on the cooperation between experts and institutions and the formation of coalitions of pioneers in the field of open science. Knowledge Exchange consists of CSC (Finland), CNRS (France), DeiC (Denmark), DFG(Germany), Jisc (UK) and SURF (The Netherlands)