Evaluation of Using AI in Analysing the Discourse of English Movies Critically

Abstract

This study is an attempt to evaluate how artificial intelligence (large language models, LLMs) analyses the discourse of movies critically. The LLMs are asked to analyze a movie called Hidden Figures critically using van Dijk’s sociocognitive approach, specifically the ideological square and the micro-macro analysis. The LLMs are Chat GPT based on GPT-4, Deep Seek 3, Perplexity Assistant, Gemini (Google AI) and Gemini (Google AI) Flash 2 . The study uses a qualitative method for analyzing the data. The data is from an American movie called Hidden Figures (2016) which is about three African American female mathematicians and their resistance to the systemic racism and sexism at NASA during the space race (1960s). The study aims at revealing the power dynamics of English movies, especially racism against black females. The main expected conclusion of this study can be that racism can be found in American movies. The significance of this study lies in using AI for discourse analysis. Since all the LLMs are recently developed, there are few studies in which AI is used to analyze discourse. The analysis demonstrates that the LLMs analyse the movie similarly except perplexity which analyses the in-group and out-group of the ideological square in a reverse way. Chat GPT, Deep Seek and Geminie flash 2 refer to the racial laws that were valid in the 1960s.