Research on Optimization Algorithm of Generative AI in Design of Personalized Learning Path for English

This paper proposes a personalized learning path optimization algorithm based on GAN (GAN-LP). The algorithm automatically generates customized learning paths for each learner through adversarial training of the generator and the discriminator, and dynamically adjusts the learning content and sequence according to personalized characteristics such as learning progress, learner interests, and abilities, thereby optimizing the learning effect. The GAN-LP algorithm generates learning paths that meet the learner's personalized needs through the generator, while the discriminator evaluates whether the path is effective, and then continuously optimizes the performance of the generator through adversarial training. Using a real English learning dataset (including learners' historical scores, behavioral data, and learning progress), this paper compares GAN-LP with traditional rule-based learning path generation methods and algorithms based on recommendation systems. Experimental results show that GAN-LP is superior to other methods in terms of learning effect, learning efficiency, and learner satisfaction. Specifically, GAN-LP can improve learners' performance in terms of test scores and task completion time by about 20%, and improve learners' satisfaction by more than 15%. This study provides innovative ideas for the application of generative AI in the design of personalized learning paths, making full use of the generation and optimization capabilities of generative adversarial networks to improve the flexibility and adaptability of learning paths. This algorithm not only has broad application potential in the field of English learning, but also provides a feasible technical framework for the design of personalized learning paths in other fields.