PEMETAAN TEMATIK PENGEMBANGAN CHATBOT AI MULTIBAHASA: TINJAUAN LITERATUR SISTEMATIS TERHADAP TREN DAN TANTANGAN MASA DEPAN

Abstract

The rapid development of artificial intelligence (AI), particularly through the emergence of Large Language Models (LLMs), has significantly accelerated the advancement of multilingual chatbots capable of understanding and generating multiple languages in a more natural manner. In the context of globalization and linguistic diversity, this technology plays a crucial role in improving access to information and services across sectors such as education, healthcare, governance, and social communication. This study conducts a Systematic Literature Review (SLR) based on a Grounded Theory Literature Review approach, analyzing 100 articles published between 2018 and 2025 to map research developments, methodological trends, and application focuses in multilingual chatbot research. The analysis follows the stages of open coding, axial coding, and selective coding to identify recurring patterns and core themes in prior studies. The findings reveal a substantial surge in research after 2023, coinciding with increased accessibility to chatbot development enabled by LLMs. This shift has redirected research emphasis from technical engineering toward implementation, evaluation, and user experience. System development methods (62%) and experimental approaches (26%) emerge as the most dominant methodologies, while the most frequently explored domains include education, healthcare, and public services. Multilingual chatbots represent the most extensively studied type, reflecting a transition toward a multilingual-by-default standard. Nevertheless, challenges such as English language bias, underrepresentation of minority languages, and limited cultural sensitivity remain central issues. These findings underscore the importance of strengthening the quality of multilingual corpora and adopting more socio-culturally responsive approaches to develop inclusive and linguistically equitable chatbot systems.

Ähnliche Arbeiten