Algorithm aversion in agricultural decision-making: Trust dynamics, barriers, and fertiliser-related decision support

Algorithm-based fertiliser recommendations offer substantial potential to improve Nitrogen Use Efficiency (NUE) and support economic and environmental sustainability. However, adoption among farmers in the United Kingdom (UK) remains limited, partly due to algorithm aversion, i.e., the tendency to distrust or avoid algorithmic-generated recommendations, even when they provide benefits. This study examines algorithm aversion in fertiliser-related decision-making among UK farmers and agronomists. Aiming to identify key barriers to adopting decision-support tools (DSTs), improving understanding of stakeholder trust dynamics, and exploring strategies to improve uptake. An online survey of 50 farmers and 26 agronomists assessed confidence in algorithmic recommendations versus human advice, understanding of NUE, perceived adoption barriers, and openness to non-traditional fertiliser recommendations. A follow-up workshop with 10 participants in DSTs trials provided qualitative insights into trust and usability. Farmers reported significantly greater trust in human advice compared to algorithmic recommendations (median 8 vs. 6, p < .001), whereas agronomists showed the reverse pattern (median 8 vs. 7.0, p < .001). Perceived barriers included cost concerns, poor system integration, complexity, and confusion over metrics. Whilst some farmers showed low levels of NUE literacy, agronomists demonstrated higher NUE literacy. Farmers relied on advice grounded in social trust and shared beliefs, while agronomists viewed algorithmic outputs as complements to technical expertise. Workshop participants found DST dashboards informative but often overwhelming. Addressing algorithm aversion through improved interface design, transparency, and tailored education, particularly via trusted advisors, may bridge the trust gap and facilitate digital tool adoption. • Farmers show significant algorithm aversion in fertiliser-related decision-making. • Agronomists are more receptive to algorithmic tools than farmers. • Perceived barriers to use include cost, poor usability, low trust, and unfamiliar metrics. • Trusted advisors may play a role in supporting farmer adoption of digital decision-support tools. • Addressing algorithm aversion has potential to improve existing on-farm practices.