Mapping cohesiveness and innovation outcomes in generative artificial intelligence academic research through a systematic literature review

Generative artificial intelligence (AI) is increasingly reshaping academic research by supporting idea generation, literature synthesis, identifying research gaps, writing, data interpretation, collaboration, and methodological decision-making. This systematic literature review examines how generative AI interventions may contribute to more cohesive and innovative academic research in the world of academia. Using the Scopus database, this study reviewed 38 peer-reviewed publications published between 2018 and 2025. The review followed a structured screening process guided by inclusion and exclusion criteria, supported by bibliometric and content analysis. The findings show a sharp increase in publications after 2021, reflecting the growing influence of advanced large language models (LLMs) such as ChatGPT, GPT-4, and related AI tools. The results indicate that generative AI is most frequently applied in writing and editing support, literature review automation, idea generation, collaborative research, and data interpretation. These tools are associated with enhanced research cohesiveness by improving clarity, logical structure, thematic consistency, and manuscript organization. They also support innovation by helping researchers identify research gaps, generate novel hypotheses, and develop interdisciplinary connections. However, the review also highlights major ethical and practical concerns, including hallucinated outputs, authorship ambiguity, bias, lack of transparency, data privacy risks, and possible erosion of researchers’ critical thinking skills. The study further reveals uneven adoption across disciplines and regions. Overall, the review concludes that generative AI has strong potential to improve academic research quality, but its responsible integration requires ethical guidelines, institutional policies, AI literacy training, and human oversight.