Prompt Engineering and Soft Metacognitive Competencies: A Framework for Ethical Academic Writing with Generative AI

The emergence of generative artificial intelligence tools such as ChatGPT has revolutionized academic writing practices; however, a lack of effective prompt engineering skills and adequate metacognitive competencies poses risks to the integrity and quality of scholarly work. This systematic literature review employed the PRISMA protocol to analyze 54 selected articles published between 2022 and 2024, sourced exclusively from the Scopus database. Article searches used keywords related to prompt engineering, generative AI, academic writing, and metacognition. The selection process involved identification, screening, and inclusion phases, ensuring that only relevant articles with sufficient methodological quality were further analyzed. Findings show that clear, specific, and context-rich prompts significantly enhance the coherence and analytical depth of AI-generated academic texts. This study proposes a comprehensive soft metacognitive competencies framework, encompassing dialogic intelligence, structured thinking, AI ethics literacy, and self-aware AI user capacities. These competencies equip users to uphold academic and ethical standards in AI-assisted writing. Compared to previous models, the proposed framework offers a more structured and reflective approach, representing a theoretical contribution that enriches contemporary literature in digital literacy and educational technology. Practical recommendations are provided to integrate this framework into higher education curricula, including AI literacy modules, prompt engineering training, and institutional policies establishing ethical boundaries for AI use. This study underscores the urgency of fostering soft metacognitive competencies to ensure generative AI serves as a reflective partner—not a replacement—in producing ethical, original, and high-quality academic work.