Vol. 4 No. 1 (2024)

Dauerhafte URI für die Sammlung

Listen

Neueste Veröffentlichungen

Gerade angezeigt 1 - 5 von 7
  • Item
    Editorial: Sustainable Artificial Intelligence – Critical and Constructive Reflections on Promises and Solutions, Amplifications and Contradictions
    (Weizenbaum Institute, 2024-07-16) Ullrich, André; Rehak, Rainer; Hamm, Andrea; Mühlhoff, Rainer
    The developments in the field of artificial intelligence (AI) have seen many ups and downs since AI’s infancy. Recently, however, surprisingly powerful AI systems have been developed and are widely considered as silver bullets for any kind of social, ecological, political, scientific, or economic problem. However, the critical consideration of AI developments – especially their implications for society and the environment – has not been cultivated to the same extent. This imbalance leaves plenty of room for unreflective belief in technological progress and accompanying “techno-solutionism.” In order to inform and advance the debate regarding sustainability-oriented AI and the sustainability of AI itself, we compiled this thematic issue with reflections on the promises and solutions, amplifications and contradictions created by introducing AI into the sustainability endeavor and introducing sustainability-related application cases into AI development.
  • Item
    AI Literacy for the Common Good
    (Weizenbaum Institute, 2024-07-16) Ullrich, Stefan; Messerschmidt, Reinhard
    Artificial Intelligence (AI) does not provide solutions to pressing social questions, such as those pertaining to a peaceful, sustainable, and socially acceptable world. However, when employed in a purposeful and critically reflective manner, it can assist in formulating more effective inquiries that can enable a better understanding of the terms “AI” and “common good.” Through implementation in response to sustainability issues and given its potential as an inclusive technology, AI could be a powerful and useful tool for the common good. Despite the possibility of useful machine learning applications in terms of a positive cost-benefit calculation for its life cycle energy and resources, the majority of AI is far too energy-hungry for model training and to scale inferences. Despite the considerable variation observed in terms of certain aspects, it is evident that AI is currently neither sustainable in itself nor primarily used for sustainability purposes to address the grand challenges of global society in a world characterized by rapid acceleration. This demands a critical understanding of how AI systems work to enable society to decide upon the areas in which we should, can, or even definitely must not use AI. Based on the UNESCO Framework for AI Competency and the Dagstuhl Declaration of the German Informatics Society, we advocate for a type of critical AI literacy that can be best taught through practical use, that is, “learning by making.” This approach leads to a concise overview of existing options that facilitate a more reflective approach to using and understanding AI, including its potential and limitations. We conclude with a practical example.
  • Item
    Unlocking AI’s Potential: Human Collaboration as the Catalyst
    (Weizenbaum Institute, 2024-05-27) Buxmann, Peter; Ellenrieder, Sara
    Rapid advances in artificial intelligence (AI) have fueled high expectations for the technology’s potential to fundamentally transform our economy and society through automation. However, given the inscrutability and, sometimes, susceptibility to error of AI systems, we argue that the focus should shift towards fostering effective human-AI collaboration rather than pursuing automation alone. In this context, system decisions must be made available to decision-makers in an explainable and understandable manner, as further required by the EU’s recently passed AI Act. Research shows that there is potential for humans to learn from explainable AI systems and improve their own performance over time. Meanwhile, in addition to enabling humans to benefit from working with AI systems on various everyday tasks, such collaboration ensures the safe and reliable use of AI systems, especially in high-risk areas such as medicine, where human oversight remains paramount.
  • Item
    Can Sustainable Shopping Recommendations in Online Retail Help Reduce Global Warming?
    (Weizenbaum Institute, 2024-04-25) Hoffmann, Marja Lena; Nanevski, Ivana; Gossen, Maike; Bergener, Jens; Flick, Alexander; Santarius, Tilman; Biessmann, Felix
    Two dominant and contradictory narratives describe the apparent contribution of information and communication technology (ICT) to climate change. On the one hand, ICT can reduce global greenhouse gas (GHG) emissions by, for example, supporting energy efficiency or promoting sustainable consumption. On the other hand, the increased energy demands of emerging software components leveraging artificial intelligence or machine learning can be directly and indirectly responsible for GHG emissions. This makes it critical to assess whether ICT mitigates or exacerbates net climate impacts and the contributing factors. The impacts of software have received relatively little attention and require the development of new approaches to conduct such assessments. In particular, the net effect of complex real-world applications is frequently not measured. In this study, we provide a detailed step-by-step assessment to quantify the net global warming potential of an online shopping recommendation system that encourages users to make sustainable consumption decisions. We consider the energy consumed and associated GHG emissions in the development and use of the software and compare these to the potentially avoided GHG emissions associated with more sustainable recommended options. The results demonstrate that the software has the potential to indirectly avoid more emissions than it causes and that changes at different steps of the software can amplify this.
  • Item
    Potentials and Limitations of Active Learning: For the Reduction of Energy Consumption During Model Training
    (Weizenbaum Institute, 2024-04-08) Nenno, Sami
    This article investigates the potential and limitations of using Active Learning (AL) to reduce AI’s carbon footprint and increase the accessibility of machine learning to low-resource projects. First, this paper reviews the recent literature on sustainable AI. The core of the article concerns AL as an emissions reduction technique. Because AL reduces the required data for model training, one can hypothesize that energy consumption  and, accordingly, carbon emissions – also decreases. This paper tests this assumption. The leading questions concern whether AL is more efficient than traditional data sampling strategies and how we can optimize AL for sustainability. The experiments show that the benefit of AL strongly depends on its parameter settings and the data set size. Only in limited scenarios does the size reduction outweigh the computational costs for AL. For projects with more resources for annotations, AL is beneficial from an ecological perspective and should ideally be paired with model compression techniques. For smaller projects, however, AL can even have a negative impact on machine learning’s carbon footprint.