AI and Sustainability II - EnviroInfo 2023

AI and Sustainability - Session II

Assessing Sustainable Artificial Intelligence via Societal Impact Analysis: The Case of Earth Observation

Abstract

The pursuit of ecological and social sustainability in the face of climate challenges has brought attention to the potential of artificial intelligence (AI) as a transformative tool. While AI’s ecological applications like data analysis and decision automation have been extensively researched, their societal implications often remain less apparent due to their indirect nature and focus. This paper addresses this gap by proposing a comprehensive societal impact analysis grid and applying it to the context of Earth Observation (EO) driven by satellite data. The suggested societal impact analysis grid not only extends existing technology assessment frameworks to incorporate critical social and data protection theory aspects but also makes it particularly suitable for assessing the societal implications of AI and data-related technologies. Applying this grid to EO reveals both practical benefits and emerging societal issues. EO’s capacity for valuable sustainability insights, such as environmental monitoring, underscores AI’s potential. However, the analysis also unveils challenges like an observer-observed relationship and global data-related inequities. By constructing this impact grid and examining the EO case, this paper contributes to the evolving discourse on sustainable AI, aligning ecological and social considerations.

How do the European Court of Human Rights rulings in environment-related cases affect the future of Environmental Informatics?

Abstract

In recent years, significant developments in environmental and human rights have taken place within international bodies. The United Nations granted the right to live in a healthy environment as a Human Right in 2021. Following this, the Council of Europe updated its instruction on Article 46 of the Convention in 2022. The President of the ECHR, in April 2023, highlighted issues within the Tribunal related to the implementation and execution of judgments, emphasizing the importance of the Rule of Law and the need to address clear issues. This ongoing research delves into the potential of Environmental Informatics and ESG reporting to predict the consequences of the ECHR Great Chamber’s case on prioritizing climate actions and their costs. The study examines the intersection of Fundamental Rights and Environmental Informatics, particularly in the context of climate action. It explores how these elements can be integrated into software and considers the various perspectives of stakeholders, including developers and regulators. The research embraces a multidisciplinary approach to tackle complex questions surrounding automated decision-making, data aggregation, and users’ responsibility, aiming to contribute to discussions on sustainable and ethical software development.

How to facilitate the Creation of Machine Learning predictive Models

Abstract

This work presents a methodology that facilitates the creation of machine learning predictive models. This procedure merges several mathematical and statistical tools which we call Predictive Factor Association (PFA). PFA is applied to predicting pollution emergencies. This knowledge in used to determine the impact pollution has in population health as health risk factors for different types of population are developed. Artificial intelligence is used to enhance chemical and physical air pollution modelling as input for determining population health risks.

Physics-informed Machine Learning for Green Applications

Abstract

Modern deep learning methods require a lot of computational resources for training. The topic of energy efficiency will therefore be of crucial importance for the development of new algorithms in the future. One way to develop more energy-efficient methods is to intrinsically consider physical laws and constraints and thus, for example, significantly limit the data space for training. In my talk, I will give an overview of Physics-informed Machine Learning and show possible application areas within our current projects in the environmental domain.