Abstract: Agricultural knowledge-driven service technology (AKDST) can enable to generate natural language using promising artificial intelligence (AI). Various domains and formats of intelligent and personalized knowledge can be provided from crop production to food processing in the agricultural industry. AKDST can serve as a promising knowledge service to promote the quality and productivity in modern agriculture. This is also the main goal of current agriculture to fully meet the essential requirements of modern society. As such, there is great potential and opportunities for AKDST in the frontier of agricultural research. The whole process of technology development can be covered from the conception design, implementation, evaluation, application, and dissemination. Meanwhile, it is urgently necessary to require sufficient and efficient knowledge services in the agricultural industry at present. The current knowledge service can be improved to realize the short waiting time with low cost, high coverage, and accuracy. AKDST can be expected to translate the great progress in personalized and customized knowledge services. The most relevant and useful knowledge can also be found in the preferred modalities and formats, according to the needs and preferences. Especially, the advanced ChatGPT has been released to provide interactive and participatory knowledge services since November 2022. The large-scale pre-trained models can be potential for agricultural knowledge-intelligent services. The existing knowledge can be easily accessed to share the innovative technology. ChatGPT can serve as the prime example to generate fluent and coherent dialogues with technical support and feedback in the AKDST advancement. This review aims to analyze the current status and trend of AKDST-related technologies, and then prospect the potential of AKDST in the field of agriculture. Future research was also recommended to design and implement the large-scale pre-trained models. The more powerful and versatile AKDST was achieved in the large model, performance, and learning. In addition, the current mode of agricultural knowledge service was updated from the data retrieval, semantic matching, and the passive and static knowledge bases. Furthermore, technical support was combined with the agricultural machinery, information technology, agronomic practices, and communication channels for different components in the agricultural information system. Multimodal service was integrated with the text, image, voice, and video. The human-machine interaction was further enhanced suitable for human behaviors, habits, and cultures, considering human needs, preferences, emotions, human values, rights, and dignity. Technical support was also provided in the intelligence of agriculture, leading to the transformation from the agricultural knowledge service to the generative knowledge-driven mode. New knowledge was created using existing knowledge. Novel and diverse knowledge was output, such as summaries, explanations, suggestions, and evaluations.