Generative artificial intelligence technology is profoundly affecting the field of landscape planning and design, bringing new opportunities and challenges to knowledge production and planning and design practice. The research is based on the difference between landscape planning and design practice before and after the intervention of artificial intelligence. Firstly, the development process of artificial intelligence in three key application scenarios of simulation prediction, scheme generation, and design evaluation is deeply analyzed. 1) Artificial intelligence in simulation and prediction scenarios is mainly used in dynamic simulation of landscape pattern, prediction of spatial pattern evolution, scene simulation, etc., to help predict the future development trend of landscape, which is usually in the current situation and problem analysis stage in the early stage of planning and design. From the initial empirical prediction of the urban dynamics model to the machine prediction stage of the analytical artificial intelligence stage. 2) In the context of scheme generation, artificial intelligence is mainly used in the rapid generation of spatial layout schemes, such as landscape plane generation and landscape rendering, which is usually in the mid-term scheme proposal stage of planning and design. With the evolution of artificial intelligence from non-intelligent tools to generative artificial intelligence, scheme generation has also achieved a leap from auxiliary drawing to generative design. 3) In the context of design evaluation, artificial intelligence is mainly used in landscape performance evaluation, landscape visual quality evaluation, landscape perception evaluation, etc., to help designers evaluate and improve planning and design schemes. It is usually in the early stage of planning and design, and the later stage of performance evaluation. Artificial intelligence tools have evolved from the earliest expert system to data-driven evaluation. Secondly, based on the time period theory and the subject-object dichotomy theory, an analytical framework is proposed to dialectically view the three roles of artificial intelligence of knowledge production in landscape planning and design practice: from the short-term perspective of technological and product innovation, human beings as the main body of planning and design practice, and generative artificial intelligence can be regarded as "servants"; from the medium-term perspective of industry change, generative artificial intelligence, as the main body of planning and design practice, has become an "opponent" under the negative impact of technology; from the long-term perspective of knowledge production and social communication change, when both are subjects, generative artificial intelligence can be regarded as a "partner". The importance of building a new model of human-machine collaboration is further emphasized, and a knowledge production model of landscape planning and design based on human-machine collaboration is proposed. Finally, the future development and challenges of artificial intelligence in the production of practical knowledge of landscape planning and design are prospectively discussed. It is still a big challenge for generative artificial intelligence to participate in the interdisciplinary and local knowledge production of landscape planning and design, and the knowledge production in the teaching and research of landscape planning and design. In the face of the new trend of knowledge production mode change brought by artificial intelligence, we should understand it from the perspective of knowledge production mode change, ideology and methodology evolution, and examine landscape planning and design practice and subject knowledge system with a new problem domain. Further thinking about the role and responsibility of artificial intelligence and planning designers in landscape planning and design practice, and how to better integrate the collaboration between humans and artificial intelligence to create more value.