scieee AI-readable full text Open interactive document viewer

Openness in the age of AI – A Conversation with Norway's National Centres for Artificial Intelligence

Goncharenko, Igor

Publisher
Zenodo
DOI
10.5281/zenodo.17652120
Repository
zenodo.org
Language
en

Abstract

Artificial intelligence is making its way into the research landscape, and in this OA week event we will explore whether it is possible to maintain open science ideals, such as openness, transparency, collaboration, and reproducibility alongside AI? We will ask the question: How open can AI research be? Voices from several of Norway’s newly established national AI research centres will join us to discuss how they work with open science ideals. Attached are the presentations from the event and below is more detailed information about the speakers and the AI research centres. Speakers Alexander Refsum Jensenius is Professor of Music Technology at the University of Oslo, where he is Director of the RITMO Centre for Interdisciplinary Studies in Rhythm, Time and Motion, and the fourMs Lab. He is also the Director of the MishMash Centre for AI and Creativity. MishMash is a major Norwegian center dedicated to exploring the intersection of AI and creativity. Its primary aim is to create, explore, and reflect on AI for, though, and within creative practices. The center investigates AI’s impact on creative processes, develops co-creative AI systems, and examines the ethical, cultural, and societal consequences of AI for creative professions. Michail (Michalis) Giannakos is Professor of Interaction Design and Learning Technologies at the Department of Computer Science at NTNU, Head of the Learner-Computer Interaction Lab, and a co-director of the AI Centre for Empowerment of Human Learning (AI-LEARN) — one of six national AI centres in Norway. His work focuses on developing new ways for humans to interact with intelligent learning systems. The AI Centre for the Empowerment of Human Learning (AI LEARN) is an interdisciplinary and intersectoral effort dedicated to understanding and shaping the human dimensions of AI interaction and adoption for sustainable, inclusive & responsible integration across both public and private sectors and for all citizens. AI-LEARN focuses on the interaction between humans and AI, with an emphasis on developing human-centred infrastructures designed to enhance and empower human learning. Anna Smajdor is Professor of Practical Philosophy at the University of Oslo and member of the leadership group at the Norwegian Centre for Trustworthy AI. She studied philosophy at the University of Edinburgh and earned her PhD from Imperial College London, focusing on the ethical and legal implications of artificial gametes. Her research explores ethical challenges in medicine, innovation and the life sciences, with particular emphasis on the intersection of bioethics and artificial intelligence. The Norwegian Centre for Trustworthy AI (TRUST) aims to make artificial intelligence fair, safe, and responsible. TRUST’s mission is to develop an interdisciplinary knowledge base for the development of safe, fair, and responsible AI. The center conducts research on technology, societal impact, and innovation in close collaboration with academia, research institutes, industry, the public sector, and civil society. Karin Rydving (chair) is the Assistant Director of the University of Oslo Library, overseeing the Medical and Science Library. She holds a Master's in Management of Library and Information Institutions as well as a Master of Arts in Language Education. With over a decade of experience in university libraries, she has led various departments, focusing on organisational development and change management. Karin is currently part of the LIBER Taskforce on Artificial Intelligence and has represented her institution on numerous committees related to Research Data Management and Library Systems. Event organisers Elin Stangeland (coordinator), UiS Karin Rydving, UiO Sondre Strandskog Arnesen, HVL Therese Skarås Skagen, HVL Igor Goncharenco, UiA Research Data Alliance NO - Training and outreach working group

Full text

TRUST The Norwegian Centre for Trustworthy AI The centre's consortium is committed to enabling future AI systems to be accurate, interpretable, inclusive, fair safe, sustainable and well governed The aim of TRUST is to provide ground-breaking transdisciplinary research results that make AI accurate, interpretable, aligned and inclusive, safe, sustainable, well-governed, and with the capacity to reach all this at scale. and thus, be trustworthy. Solutions will be tested in the real world with partners in academia, industry, government and civil society. •Assumption: •Openness leads to trust •Openness is necessary/sufficient for trust •Question: •Does it? •Necessary and/or sufficient or neither? •Grounds for uncertainty • EG O’Neil on autonomy and trust •Reification of ethical ideals when formalised Openness and Trust •Assumption: •There is one right ethical answer or principle •We just have to find it •Reality: •There are many competing ethical values •Each ethical decision is a trade off •Maximising one dimension is rarely best •Teleology •Requires us to define goals •Ethics is not free floating, it is contextual •Openness is not an ethical panacea Ethics in tension; ethics’ intention The key challenge How to translate the abstract normative ideals of TRUST into concrete actualities •What work needs to be done? •When should this work happen? •By whom should it be done? Is it already too late….? Relevant documents •Consortium Avtale •Data Management Plan •Responsible Research and Innovation (RRI) Consortium Agreement •Access rights to project results: royalty free for consortium participants involved (conditions involved if it is for commercial utilisation) •Project results disseminated quickly, and made publicly available except if countervailing terms and conditions, or national legislation •PhDs must be published openly unless for confidentiality reasons •(Lengthy confidentiality stipulations) •Ethical requirements: consortium participants must ‘maintain the highest ethical standards’ and make sure their employees and subcontractors do the same Data Management Plan •Many different forms of data collected/created •Reliance on ‘national or international data storage infrastructures, e.g. (NIRD) •Alignment with FAIR principles: findable, accessible, interoperable and reusable •Separate extra secure data storage for highly sensitive data under GDPR, eg medical data •Use of Creative Commons licences for open data, but for commercial collaborations, NDA may be used •Robust informed consent procedures where human participants involved •Compliance with ethical standards, institutional policies etc.