The high-quality progress of various information technologies has brought convenience to people’s daily life, which also makes the high-quality development speed of social economy gradually accelerate. To some extent, it can be said that the development and adoption of Information Technology (IT) in the field of modern scientific research has become the engine to promote the progress of social economy. On the other hand, this information reform in all walks of life can also bring considerable benefits to all types of enterprises. Among them, the most intuitive is the improvement of enterprise’s market competitiveness and operation efficiency. At this time, the graphic design field is also seeking a method of information transformation to further improve the competitiveness of different enterprises in the graphic design field. The new mode of products or services created for market operation has also made great changes in many industries. The current Visual Design (VD) of the plane image of commodity packaging needs to be based on the function of transmitting information to realize the implicit publicity of products. Among many current information technologies, Artificial Intelligence (AI) can complete this work through relevant algorithms and derivative technologies. With the powerful data processing ability of AI technology, firstly, the needs of the product packaging to be designed are analyzed, and then a usable plane VD of product packaging is created by learning a large number of sample data. At first, this paper deeply analyzed the workflow of the current VD mode of product packaging plane image, and then confirmed the feasibility and reliability of the adoption of AI computer model in the VD of product packaging plane image. Finally, a VD mode of product packaging plane image oriented to AI computer model was proposed. Through simulation experiments, the performance difference between the AI computer model-oriented product packaging plane image VD mode and the current plane image VD mode on multiple evaluation indicators was analyzed, and the performance of this new design mode on multiple evaluation indicators was determined to be improved by about 26.8% on average.
