Minisymposium

In recent years the hpc community has increased its efforts in sustainability. But cutting edge performance requires ever increasing resources, above all electrical power, but also semiconductors, raw materials, cooling fluids, land and obviously money to buy all of these. Increases in the supply do of compute do not seem to affect demand much. For all sustainability efforts this means that whatever relative efficiency gains can be implemented, they do not translate to an absolute reduction in ecological footprint. The largest target of sustainability efforts cannot be attacked directly, lest at the cost of performance. The demand for performance is an externality which is partially imposed by society, by science and by politics. But it also partially reflects the narratives of the hpc world. The aim of our talk is two fold. Firstly we will show how relative and absolute sustainability efforts of HLRS have developed historically. Secondly we will discuss how we, as a hpc center, have positioned us with respect to the dilemma of ever increasing absolute resource consumption. We close by asking if there is any prospect for changing the current hpc narrative so that absolute resource reductions can become part of it.

Data mining and artificial intelligence (AI) have become an increasingly used method of medical research in recent years. The primary goal is to identify patterns that can prevent diseases, enhance and accelerate diagnostic as well as therapeutic interventions. This aligns with the United Nations Sustainable Development Goal 3: "Ensure healthy lives and promote well-being for all at all ages". Every medical appointment generates data that can be valuable for data-driven medical research. The quality of this data improves with number of individuals and the data points per individual. Commonly sharing patient data across medical institutions could advance data driven medical research. However, this approach raises concerns about individual autonomy and privacy. Addressing this ethical conflict is crucial and far from being trivial. How can a widely accepted balance between ethical principals be found? Is the current practice of obtaining informed consent from patients sufficient to protect their privacy in the face of AI-driven data analysis?

Open source software has a notorious diversity problem. Women’s participation in OSS development is significantly lower than that of industry (10% compared to 30% of programmers). Multiple studies have highlighted women’s lower participation levels in OSS and their barriers to entry, and more recent studies have begun to expand to other underrepresented groups, such as Black and Hispanic contributors. To understand the state of the literature on diversity, equity, and inclusion in OSS, we have conducted a literature review to coalesce findings and recommendations. We take a broader definition of diversity, and sought out to analyze all articles concerning gender, race/ethnicity, ability, sexuality, and geographic location/national origin in OSS development and communities. We studied almost 200 papers, of which 108 discussed diversity in OSS. Here, we present our coding scheme, as well as future directions for researchers and OSS practitioners, communities, and advocates.

This session will be an open discussion on developments and future directions in ethical and societal considerations for scientific computing, through the lens of the 17 UN Sustainable Development Goals (SDGs). Participants are encouraged to ask questions, challenge perspectives, and contribute their experiences to foster a lively, interdisciplinary conversation. The session will take place in two phases: We will start the discussion with several prepared questions and challenges focused on the intersection of scientific computing and the SDGs, and then open the session to questions and challenges from attendees on ethical and societal considerations for scientific computing.