The traditional English writing evaluation is an artificial way to judge, which is a relatively subjective evaluation method. Also, the artificial method has the defects of low timeliness and high error rate, which limits the interest of college students in English learning. If the content of college students’ English writing cannot be judged, it also limits their understanding of the correctness of the English language, because English writing can reflect problems such as English grammar, sentence structure, and emotional expression. Artificial intelligence (AI) technology has developed rapidly, and it can efficiently process the data of research objects. If artificial intelligence technology is combined with English writing evaluation, it can improve the timeliness and error rate of college students’ English evaluation. This is also an innovative way of judging college students’ English writing. Combining the characteristics of great learning English writing and judging criteria, this research uses artificial intelligence technology to design an efficient judging platform. It needs to use a convolutional neural network (CNN) and long-short-term memory (LSTM) neural network to extract features of grammar, sentence patterns, and emotional expressions of college students’ English writing. The research results show that CNN and LSTM methods have high feasibility and accuracy in extracting grammar, sentence patterns, and emotional expressions of college students’ English writing. The prediction error of the college students’ English writing evaluation system is also within a reasonable range. CNN and LSTM methods have high accuracy in predicting English rectangle, grammar, and emotional expression, and the largest prediction error is only 2.91%. Also, the three prediction errors are distributed within 3%.