Artificial Intelligence (AI) systems are increasingly sought as powerful tools to inform on-farm management. Soil quality, animal welfare metrics, yield and market predictions are some examples where AI technologies are rapidly transforming the data collection process by expanding the scope, scale and interpretation of accessible information. These technologies promise to increase the production of primary goods while reducing unnecessary resource use and alleviating the harmful ecological effects of modern agriculture. Optimisation is often cited as a key determinant in the success of these AI tools. It is essential, therefore, to analyse how optimisation is evaluated (in a technical sense, and when gesturing towards broader industry improvements), and consider how these technologies are ‘optimised’ for on-farm values and needs. I argue that the integration of AI needs to be aligned with the good farmer, a role recognised in agriculture based on both practical and moral evaluations of competency. I argue that the pursuit of AI optimisation in agriculture requires the good farmer as an essential on-farm role. These systems require high-quality data to produce accurate and relevant outputs for effective decision-making. Farmers, in their role as stewards and agricultural experts, should contribute to data governance through the verification of appropriate, high-quality data. They can also offer diverse considerations needed in farming and long-term land care that go beyond quantifiable metrics. Now more than ever, it is important to ensure AI systems, if adopted, are representative of the appropriate goals of agriculture.