Machine learning is becoming increasingly important in environmental decision-making, particularly in forestry. While forest-owner typologies help in understanding private forest management strategies, they often overlook owners' relationships with technology. This is crucial for ensuring that data-driven advancements in forestry benefit society. Using Swedish forestry policy as a case, we applied Q-methodology to explore forest owners' perceptions of machine learning. We conducted 11 qualitative interviews to generate 33 statements, which were then ranked by 26 participants. Inverted factor analysis identified four ideal-type perceptions of machine learning, interpreted through self-determination theory. The first perception views machine learning as unhelpful and socially disruptive. The second sees it as a complement to forest governance. The third expresses no strong opinions reflecting a relative disengagement from forestry. The fourth considers it essential for decision-making, particularly for absentee forest owners. The extracted perceptions align with existing forest owner typologies when it comes to reliance on others and willingness to take advice. The discussion includes concrete policy recommendations, focusing on privacy concerns, educational initiatives, and strategies for communicating uncertainty.