Interviews with twenty forest sector experts show that, despite the ability of artificial intelligence (AI) to simplify mapping, analysis and everyday tasks, there are concerns about so-called black box accountability and the risks of using biased data.
According to Goldman Sachs data, forest managers around the world use remote sensing technologies, machine learning and drones to manage forest areas.
Artificial intelligence could be a highly significant factor in forestry, but at the same time forest sector representatives warn that opaque algorithms and biased data may jeopardise important land management and policy decisions, concludes a new study published in the journal Forest Policy and Economics and based on 20 in-depth interviews with forestry professionals from academia, government and the private sector in the south-western United States. The interviews were conducted by lecturers from the School of Forestry at Northern Arizona University.
More about the study: https://www.sciencedirect.com/science/article/abs/pii/S1389934125002059?via%3Dihub
“There is plenty of discussion about the potential of artificial intelligence in forestry, yet there is very little research into what forest sector professionals actually think about the situation,” says Professor Alark Saxena (Alark Saxena), who led the above-mentioned study together with researchers Luke Ritter (Luke Ritter) and Derek Uhey (Derek Uhey), adding: “This was our first piece of exploratory work in this field, motivated by the need to understand technological change from a human point of view.”
The researchers found that forest sector professionals are worried about the so-called black box problem, namely that AI “decision-making processes” are hidden from users. “Sector specialists are particularly concerned about the so-called black box problem, where it is impossible to understand the AI decision-making process, creating serious problems in the area of accountability,” Professor A. Saxena stressed.
“Professionals pointed out that the main problem is the ‘training’ of AI, when poor-quality or biased data from certain agencies is used. Another risk is trusting erroneous results on important questions of land and forest management or forestry policy analysis.” Examples include recommendations on where to carry out controlled burning or permit logging.
It must be acknowledged that many respondents view artificial intelligence positively as a tool that helps to address labour shortages and can carry out standard tasks. “Foresters see great AI potential in the analysis of complex data, such as light detection and ranging, provided that AI acts as an assistant that enhances rather than replaces the judgement of an experienced professional,” A. Saxena noted.
Researcher Luke Ritter explained that, before artificial intelligence is more widely introduced in forestry, it is very important to understand the scepticism and recommendations of practitioners in the sector: “It is difficult to put the ethical aspects of AI use into practice if we foresters have incomplete knowledge of how and why AI is being used at all. The study brought some interesting themes to light, but questions about artificial intelligence undoubtedly need to keep being discussed in university lecture halls and in forest sector workplaces. We hope that our research will provide a basis both for policy changes and for further studies as artificial intelligence continues to develop rapidly.”
Knowledge, attitudes, and practices of forestry professionals towards artificial intelligence (AI), Forest Policy and Economics https://doi.org/10.1016/j.forpol.2025.103626