Large Language Model-based Tools in Language Teaching to Develop Critical Thinking and Sustainable Cognitive Structures

Abstract

Experts assert that Large Language Model (LLM) based tools like ChatGPT are the next generation in the evolution of Artificial Intelligence and will permeate all walks of human life including education. The current narrative is that we need to embed the LLM-based tools into the system taking advantage of their personalised, dynamic, adaptive nature while being mindful of their limitations. One of the greatest limitations so far identified is that these pre-trained transformer-based encoder models fine-tuned on Natural Language Processing (NLP) tasks do not reveal verifiable reasoning ability. As a result, the information generated by these tools is subject to ethical and factual errors that need human oversight. This paper uses the integrative literature review to identify and synthesize Critical Digital Literacy frameworks in language teaching in the light of the essential competencies and learning domains identified by the UNESCO Education for Sustainable Development directives. The Critical AI Literacy framework proposed in this paper would enable language teachers to adopt LLM-based tools to enhance their instructional strategies. The cognitive, affective and conative competencies developed through the new CAIL framework would empower learners to understand the manipulative nature of language and use language to build a sustainable future.