< Projects

Title

Challenges and Social Consequences of Artificial Intelligence in Swedish Forests

About the project

 Most of Sweden is covered by forest, we have a long history of active forest management, and forest products still form an important base of Swedish economy. This role is becoming even more critical, given the growing demand on multiple use forestry to provide both timber and ecosystem services, as well as recreational values. Moreover, forests and how they are managed have important implications for reaching international and national policy objectives, such as the United Nation’s Sustainable Development Goals (SDGs) and the Swedish National Environmental Objectives. Relatively recently there has been a shift in the handling of Swedish forests, from local level management involving many people in the planning and decision-making, to a more centralized process where a small group of people manage forests with help of large machines and sophisticated autonomous systems.

Artificial intelligence (AI) has become increasingly important in this process and is now implemented at multiple levels of Swedish forest management. AI combined with satellite remote sensing and geographic information systems can be used to evaluate forest resources, monitor forest development and organize management, a technological advancement that is revolutionizing how forestry is conducted in Sweden and around the world. A key area in forestry is spatial planning, where AI can be used for precision forestry management by creating detailed maps of biodiversity, tree species and wet areas. Managers and decision makers are gradually relying more and more on these AI tools and since their decisions impact forest production and forest ecosystems, they also fundamentally affect the planning of cultural and recreational values in forest landscapes. One such example is the Swedish Forest Agency’s (SFA) suggested use of AI to map key habitats or biodiversity and use this to prevent landowners from harvesting certain parts of their properties. Similarly, AI-based wet area maps can designate parts of a forest property as a riparian zone, which should be left unharvested to protect a nearby stream.

Forests have multiple uses and provide a myriad of services. How forests are managed has consequences for the achievement of national and international objectives, which often are conflicting and require cautious deliberation and balancing. Examples include the national environmental objectives Sustainable Forests and Rich Diversity and Animal Life, as well as SDGs such as Climate Action and Life on Land. Forest management in Sweden has a long tradition of being softly regulated under so-called “freedom with responsibility”, where the implementation of policies and consideration of different values is highly dependent on the forest sector itself. The SFA has in the past decades developed complex collaborative decision-making processes with forest sector stakeholders, with the aim of enhancing process legitimacy and implementation. This approach has resulted in enhanced trust and collaboration between public agencies and commercial forestry – a development that could be seriously challenged by the increased use of AI in forest management. AI interpretations are not always transparent since managers cannot know exactly how the AI tools operate. AI solutions are always dependent on the choice of models, tuning parameters, and quality of documented data, which if erroneous or inappropriate can negatively affect the performance of AI models. Moreover, AI tools could fail to strike an adequate and socially acceptable balance between all of the different policy objectives and values that needs consideration when managing forests with multiple use, hampering our ability to reach the SDGs. Hence, insufficient AI-based tools could erode the trust between stakeholders and risk jeopardizing the progress made in collaborative forest management. It is therefore important that these tools are developed and implemented with awareness of ethical issues and challenges in meeting conflicting policy goals, so both society and the forest industry are prepared for this technological development and can harness the power of AI in a responsible way. We therefore argue that traditionally based AI researchers need to collaborate interdisciplinary with humanistic and social scientific (HS) researches at an early stage.

To support this need, William Lidberg, Assistant Professor in “Challenges and social consequences of artificial intelligence in Swedish forests” is developing knowledge and support mitigation of negative societal consequences of emerging complex intelligent systems in future forest management. The overall objective of the position is to develop an interdisciplinary research group that unite the natural science and social science aspects of AI in multiple use forestry, to provide ethically and socially acceptable AI. The new position will be integrated into the research environment at the Forestry Faculty at SLU (Umeå), the top university in forest research, hosting researches who are world leaders in the research and development of AI based forest-planning systems (Dept. of Forest Resource Management), as well as researchers who have developed a new national AI based soil moisture map of unprecedented resolution (Dept. of Forest Ecology and Management), working towards the SDGs. While SLU has extensive knowledge on the development of AI models for multiple use forestry, research from HS perspectives has not yet received much attention. Thus, this HS position is highly important, as it will strengthen the ambition of SLU to advance science and education by connecting researches from different departments and universities, and fostering a new generation of researches with insight in both natural science and HS issues. The extended HS perspective will contribute and benefit SLU as a whole, adding new dimensions to many research challenges currently addressed by SLU researchers.

Duration
Project type

Assistant Professor Project

Keywords

Universities and institutes

Swedish University of Agricultural Sciences

Project members

William Lidberg

William Lidberg

Assistant Professor

Swedish University of Agricultural Sciences (SLU)

Mariana Busarello

Mariana Busarello

PhD student

Swedish University of Agricultural Sciences (SLU)

Joakim Wising

Joakim Wising

PhD student

Umeå University