AI and Life Sciences: Guidelines for Sustainable Use
27.03.2026
The main aim of the study "Open and sustainable AI: challenges, opportunities and the road ahead in the life sciences," published in «Nature Methods» and coordinated by Silvio Tosatto from the Department of Biomedical Sciences at the University of Padua, is to make artificial intelligence (AI) reusable, reliable, and environmentally sustainable. This study, developed within the European network ELIXIR and with the participation of experts from 12 countries, highlights the growing role of AI in the life sciences.
Transforming Biology through AI
AI, particularly machine learning, has profoundly transformed biology in recent years, using large amounts of data to make accurate predictions, such as protein structures or health data analysis. However, this growth has also brought new challenges, including the reliability of results and the environmental impact of training models, which requires significant computational resources and energy.
The Role of Elixir
Elixir, a research infrastructure that connects 24 European countries, promotes an ecosystem where data and models can be used transparently, accessibly, and sustainably, through the co-design of databases, software, training resources, and computational infrastructures.
Recommendations for Sustainable AISilvio Tosatto explains that their study offers 9 practical recommendations to develop open and sustainable AI (OSAI) throughout the lifecycle of machine learning models. These recommendations focus on three key dimensions: making AI reusable through the standardisation of data and models, ensuring reproducibility by sharing code, methods, and experimental conditions, and reducing environmental impact by adopting "Green AI" practices.
Available Tools and Resources
The study maps over 300 tools and resources already available, allowing researchers to practically apply these recommendations and bridge the gap between theory and practice.
Impact Beyond the Research World
The impact of the study goes beyond the research world, as AI is already improving health, medicine, and quality of life. Making these technologies more transparent and sustainable means achieving more reliable diagnoses, more effective treatments, and more responsible research. In summary, the study represents a decisive step towards more reliable, shareable, and sustainable AI, with tangible benefits for the scientific community and society.
