Informing community of ethical use of AI

Applications of AI technology are essential for addressing the multifold challenges associated with environmental and sustainability issues including climate change, biodiversity, air quality and human activities further adding the challenges. AI technologies to identify solutions are very important to ensure a timely response, but it is equally important to build a better understanding among all stakeholders of the environment itself. There are growing concerns about the lack of transparency in AI, leading to a notion of an AI black box. Often users of AI tools do not know how these tools work — the underlying data or method they use, uncertainty and limitations associated with them, social or algorithmic biases they might perpetuate and harmful behaviour or injustice they reinforce. We need to acknowledge the fragmentation of efforts by different stakeholders, each holding a narrow focus on climate efforts. It is essential to build a shared understanding of data science and AI among all stakeholders including the users of AI technologies, especially by demystifying their advantages and limitations.

As the national centre, the Turing is committed to raising awareness of and building public trust in data science by developing AI and data-led technologies that are explainable, ethical and beneficial to our society. Cross-cutting research programmes such as Tools, Practices and Systems (TPS) and Public Policy Programme (PPP) offer projects that are driven by the interest of research communities and public sectors.

In a recent impact report, Better together: The people-centred approaches driving forward data ethics, we discuss how practices and principles developed through ASG’s cross theme projects are paving way for other Turing projects. Training efforts, including the Data Study Groups and data training for biomedical scientists supported by ASG, also work towards building data skills and public engagement in AI-related conversations.

The Turing Way and Turing Commons are community-oriented learning resources hosted by TPS and PPP respectively, that teach best practices in data science for use in academia, healthcare, industry and government sectors. The Turing Way has also enabled the integration of best practices and supported the replication of its community framework in building the Environmental Data Science book, a project led by researchers from Scivision and IceNet using real-world data, we demonstrate how our actions are impacting our environment, how technological advances can help combat environmental change and how to bring diverse stakeholders together to respond to the climate crisis.

AI in environmental research will help researchers around the globe to access and make more efficient use of the vast volumes of environmental data collected by national research facilities. This work will support broader scientific communities and inform future planning of design, development and implementation of digital infrastructure to monitor our environment efficiently. Furthermore, using AI explainability methods - which provide us with the ability to interpret the predictions - researchers will ’open up the black box of the AI and conclude what it has learned from the data, potentially providing new scientific insights contributing to sustainable solutions.

Through community-oriented and open ways of working, the ultimate goal of these projects is to empower the beneficiaries of AI to build tools, reuse methods and adopt from successful projects in other sectors to tackle common high-level real-world challenges around climate response.