Farmers embrace AI when it delivers real value

Artificial intelligence (AI) is seen as a promising technology for the future of farming, but farmers are not adopting it simply because policymakers say they should. Across agricultural regions, uptake remains inconsistent, revealing a gap between technological ambition and on-the-ground reality. So, what truly determines whether farmers trust, value, and ultimately adopt AI systems in their daily operations.

In a new study, Driving factors of agricultural artificial intelligence adoption intention: an empirical study in Shandong province based on innovation characteristics, technology commitment, and individual heterogeneity, published in Frontiers in Artificial Intelligence, researchers break down the psychological, technological, and demographic forces shaping AI adoption. Using one Chinese province as a case example, the research delivers broader lessons for the global push toward smart agriculture.

Perceived value and usability drive adoption

The researchers surveyed 359 agricultural practitioners and applied an extended technology acceptance framework to examine how innovation characteristics, personal attitudes, and demographic differences influence adoption intentions. Their analysis confirms a fundamental principle of technology uptake: farmers are most likely to adopt AI tools when they believe the technology is both useful and easy to use.

Perceived usefulness emerged as the strongest direct driver of adoption intention. Farmers who believed AI systems could increase yields, reduce input costs, optimize decision-making, or improve efficiency were significantly more inclined to integrate these tools into their operations. In agricultural settings, where profit margins are often narrow and risks high, economic performance remains the central benchmark for technology evaluation.

Perceived ease of use also played a critical role. When farmers felt that AI tools were intuitive, straightforward, and compatible with their daily workflows, they were more willing to adopt them. Ease of use did not only have a direct impact on adoption intention; it also indirectly strengthened adoption by enhancing perceived usefulness. In practice, if a system feels simple to operate, farmers are more likely to recognize its performance benefits.

Among the technological features examined, mobility stood out as a key factor. Mobility refers to the flexibility of AI systems that can function across different farming environments and be accessed via portable devices. For many producers, especially those managing dispersed plots or mixed production systems, mobile accessibility lowers practical barriers and increases engagement.

Technological interest, defined as a farmer’s intrinsic curiosity and enthusiasm toward new technologies, emerged as the most influential psychological factor. Farmers who are naturally inclined to explore digital tools showed stronger perceptions of ease of use and usefulness. This suggests that motivation and mindset are just as important as hardware availability in shaping digital transformation.

Technological control belief, or the sense that one can adjust and manage AI tools according to local conditions, also improved perceived ease of use. Farmers who feel they retain control over system settings experience less uncertainty and frustration.

However, not all innovation features translated into perceived value.

Autonomy, which refers to AI systems operating independently with minimal human intervention, improved perceived ease of use but did not significantly increase perceived usefulness. This finding highlights a critical challenge in agricultural AI deployment. Many autonomous systems are designed for standardized, large-scale operations. In regions where farming is fragmented or diversified, such as certain rural areas used in the study as examples, fully autonomous systems may not align smoothly with local practices. Farmers may recognize reduced manual effort but remain unconvinced about broader productivity gains.

Similarly, technological competence belief, or confidence in one’s digital skills, did not significantly enhance perceptions of usefulness or ease of use. The researchers suggest that this may reflect a gap between perceived digital ability and the practical complexity of AI tools. Without hands-on, field-based training, confidence alone does not guarantee smooth integration.

These findings underline a central lesson for digital agriculture worldwide: innovation must be aligned with real production environments. Sophisticated features alone are insufficient if they do not fit farmers’ operational realities.

Education, experience, and usage patterns matter

The study draws focus to the way individual differences shape adoption pathways. Farmers are not a uniform group, and background variables can significantly moderate how AI is perceived.

Educational background played a nuanced role. Farmers with higher education levels were more capable of translating perceived ease of use into adoption intention. Their stronger learning capacity allows them to more effectively integrate user-friendly technologies into practice. At the same time, higher education appeared to moderate the influence of perceived usefulness. More educated farmers apply broader evaluation criteria, considering long-term adaptability and potential risks before committing to adoption.

Work experience also influenced adoption dynamics. Farmers with extensive field experience were better able to convert technological competence into perceptions of ease of use. Practical knowledge serves as a bridge between abstract digital skills and real-world application. Conversely, less experienced farmers may struggle to connect digital functions with specific production scenarios.

Agricultural type further differentiated adoption patterns. Larger-scale operations were more responsive to autonomous features, recognizing their potential to streamline complex management tasks. In contrast, small-scale or diversified producers were less influenced by autonomy, likely due to the limited applicability of standardized automation systems in fragmented production contexts.

Usage frequency introduced another dividing line. Farmers who used AI tools infrequently placed greater value on mobility and convenience. For occasional users, the flexibility of accessing AI systems through portable devices significantly enhanced perceived usefulness. Among high-frequency users, however, mobility’s influence diminished. Experienced users focus more on system reliability, data depth, and long-term integration rather than basic accessibility.

These moderating effects demonstrate that agricultural AI adoption is not a single, linear process. Instead, it reflects layered interactions between technology design, psychological readiness, and contextual conditions.

Global implications for smart agriculture

Although the research uses one region in China as an illustrative example, the broader impacts extend well beyond national borders. Governments worldwide are investing in digital agriculture to address food security, labor shortages, and climate variability. Yet technology diffusion depends on human decision-making at the farm level.

The study expands traditional technology acceptance theory by integrating innovation characteristics and individual heterogeneity into the agricultural context. By doing so, it provides a more refined framework for understanding why digital transformation advances unevenly across farming communities.

For technology developers, the findings highlight the need for modular and scenario-specific solutions. AI systems must be adaptable to diverse farm sizes and production models. Flexibility, rather than uniform autonomy, may determine long-term success.

For policymakers and extension services, the results highlight the importance of differentiated training strategies. Practical, field-based instruction that connects digital tools with real production challenges is essential. Farmers must be able to see tangible links between AI features and operational benefits.

Promotion strategies should also recognize variation among users. Lower-education farmers may require simplified interfaces and clearer guidance. Large-scale producers may benefit from advanced automation support. Occasional users may respond strongly to convenience features, while experienced users demand deeper system integration.

The study also identifies areas for future research. Its cross-sectional design captures adoption intention at a single point in time and focuses on one geographical context as an example. Expanding research across different countries, production systems, and longitudinal timeframes could provide further insight into how intention translates into sustained behavior.