ESP practitioners’ perspectives on the efficiency of using artificial intelligence ‘ChatGPT’ in needs analysis

This research investigated English for Specific Purposes (ESP) practitioners’ perspectives regarding using ChatGPT (Chat Generative Pre-Trained Transformer), an artificial intelligence tool, for linguistic needs analysis, namely target and language learning needs. It addresses a notable research gap in applying AI tools to ESP needs analysis. This study is particularly significant as it explored the potential of AI tools to enhance data accuracy and streamline the time-consuming needs analysis process. This is a significant challenge for ESP course designers. Following the qualitative approach, the study employed thematic analysis of data collected from fifteen participants familiar with ChatGPT. These participants are divided into two groups: ten are individually interviewed, and five form a focus group. Participants reviewed a summary table of needs analysis results generated by two methods – one following traditional standards and the other based on ChatGPT prompts designed by the researcher. Both sets of results relied on Hutchinson and Waters’ model, which explains that ESP is an approach rather than a product, and it is based on specific learners’ needs depicted through needs analysis. The research aimed to determine ESP practitioners’ perspectives on the utility and efficiency of ChatGPT in needs analysis and its potential to address challenges encountered in traditional or standard needs analysis procedures. The findings indicated a positive reception of ChatGPT for needs analysis among participants. ESP practitioners expressed confidence in the accuracy of data provided by ChatGPT, particularly regarding target situation requirements, which enriched their understanding of learners’ profession-specific needs. Moreover, participants believed that ChatGPT has the potential to mitigate common challenges in standard needs analysis, such as time constraints and data accuracy. However, they had reservations about the tool’s efficacy in identifying learners’ language lacks compared to traditional testing methods.