The integration of artificial intelligence (AI) tools such as ChatGPT has significantly transformed approaches to vocabulary acquisition and autonomous learning, particularly in high-stakes educational contexts. This study investigates how competitive exam aspirants in Visakhapatnam perceive the effectiveness of ChatGPT for vocabulary development and self-directed learning, and how its functionalities align with established English Language Teaching (ELT) theories. The research employed a mixed-methods design, collecting quantitative data from 101 aspirants across various coaching institutes and qualitative insights from 20 language instructors, alongside analysis of video content from official GRE and GMAT guides, to achieve methodological triangulation [1, 2]. The findings suggest that aspirants view ChatGPT as effective in delivering instant explanations, context-specific examples, and adaptive feedback. This supports Krashen’s Input Hypothesis (1982) by providing comprehensible input just beyond learners’ current proficiency (i + 1) and aligns with Vygotsky’s Sociocultural Theory (1978) by functioning as a digital More Knowledgeable Other within the Zone of Proximal Development. ChatGPT’s interactive exchanges resonate with Long’s Interaction Hypothesis (1983), while its ability to simplify dense information reflects Sweller’s Cognitive Load Theory (1988). Moreover, it fosters autonomous learning in accordance with Constructivist Learning Theory [3, 4]. Despite these affordances, respondents noted key limitations: repetitive vocabulary suggestions that may conflict with the i + 1 principle, absence of built-in progress tracking, occasional inaccuracies in responses, and affordability concerns for advanced features [5, 6]. Importantly, all findings are based on participants’ self-reported perceptions, not direct performance assessments. Future research should include vocabulary testing or log-based analytics to validate these claims. Recommendations include integrating gamified learning modules, culturally contextualized content, and hybrid human-AI feedback systems as echoed in Kazu and Kuvvetli (2023) to enhance learning outcomes and improve engagement in AI-driven educational platforms.
