Focus on AI in the second workshop on the Sustainable Development Goals
The second workshop linked to the Strategic Research Areas’ work with the Sustainable Development Goals, focused on Future-oriented methodologies, especially artificial intelligence (AI) and machine learning (ML).
Around 50 researchers, doctoral students and administrators participated in the workshop, which began with a presentation on different uses for AI in relation to sustainable development, by Sonja Aits, member of the coordinator group AI Lund. Afterwards, Kajsa M Paulsson, coordinator for research infrastructure at the Faculty of Medicine, gave a presentation on existing collaborations and the importance of getting involved in the subject in the future. After the presentations, the participants were divided into three different groups to discuss: 1) how to foster new cross-discipline collaborations and projects around AI/machine learning efforts, 2) downsides of machine learning and 3) how to implement collaborative work in practice: communication channels, tools, and frameworks.
The discussions highlighted both challenges and concrete proposals on how to achieve meaningful collaborations with a focus on sustainability. It is difficult to get an overview of all the interesting ongoing projects across the university, but one possibility would be some type of forum where researchers can call for or register interest in collaborations. Better utilization of LUCRIS was highlighted in this context. All agreed that it is important to show that they are interested in collaborations in order to be able to more easily connect AI/machine learning expertise with more problem-focused knowledge from, for example, experts in sustainability. Existing infrastructure such as centres and SRA´s should be used to a greater extent to promote interdisciplinary collaborations, for example by arranging thematic workshops or interactive forums to meet calls.
Once a collaboration has been initiated, it is important that both parties deliver in accordance with the agreement, otherwise the funding will not be forthcoming. In order to take the first step and to make the collaboration more efficient, it is thus best, according to the participants, if you know each other from before or at least work at the same university. The advantage of making better use of internal collaborations was also highlighted by the fact that new funding opportunities can be discovered. The project ideas can be developed more efficiently within a core group of partners, and the projects are then supplemented with external partners to achieve specific requirements from the funder.
In the discussion about possible disadvantages of machine learning and how AI, humans and the environment affect each other, lost expertise and fewer jobs appeared to be an obvious risk with increased machine learning. Another disadvantage is that you may place too much faith in AI, which means that you miss other solutions. Many questions were asked, such as "who is responsible when something goes wrong?" and "what happens if countries do not agree on how to measure AI?", but the regulation that exists today does not provide clear answers.
The participants believe that there is thus a need for stricter regulations for what AI can and should be used for. However, regulations must not stand in the way of the "good" AI systems that may be launched and/or used. Finally, it was stated that incorrectly deployed or used AI systems can have catastrophic consequences for the earth and humans. It is therefore crucial to agree globally and evaluate how machine learning should be used. Another important part of the development going forward is to develop quality assurance systems together with standardized documentation.
In the third discussion group, workshop participants talked about what a collaboration oriented towards sustainability should look like in practice. The participants agreed that there are several inspiring examples at the university already, for example, the Sustainability Forum is a channel for reaching out to other environments. At the regional level, SLU Alnarp was mentioned, with whom LU researchers have had many and good interdisciplinary collaborations for several years. Among the main challenges, the workshop participants highlighted the availability of material, above all the permission to reuse data and material that any other researcher has collected. There are unpublished and negative data that could be used for other purposes and in other research, where AI is a possible tool for analysis and exploration. Researchers must therefore become better at sharing their information, for example through interdisciplinary collaborations. Hopefully, successful local and national collaborations will open for international collaborations. Finally, the need for a common platform to share visions and calls for collaborations was reiterated. LUCRIS is good for projects already underway, but now it's time for the next step in the development.
The workshop ended with a summarising discussion and an invitation to use the Padlet set up for the participants to announce skills and competencies, project ideas or research questions that need to be answered, available infrastructure, expert requests and potential collaborations.
Resources and contacts
AI Lund is an open network for research, education and innovation in AI at Lund University, and an excellent forum for networking and finding partners in AI.
Lund University's Research Portal lists important existing infrastructure within the University where you can find existing methodologies/infrastructures. Various initiatives at national and international level also collaborate in important infrastructure investments, such as SciLife, ESS. Both at central and faculty level within Lund University, there are Research Boards that are also responsible for research infrastructure and planning for future investments in new future-oriented methodologies.
The workshop was arranged in collaboration with the Strategic Research Areas eSSENCE and LUCC.