The Application of Prompt Engineering in AI-Assisted Writing: A Cross-Domain Biliometric Literature Review (2020-July 2025)

Abstract

Prompt engineering has rapidly emerged as the pivotal interface between generative artificial intelligence and human writing, yet the contours, drivers, and blind spots of this burgeoning field remain unmapped. We present the first large-scale bibliometric synthesis of scholarly work on prompt engineering for AI-assisted writing, analysing 1,049 peer-reviewed articles and reviews published between 1 January 2020 and 22 July 2025. Leveraging the Web of Science Core Collection and PRISMA-guided screening, we chart publication growth, geographical and institutional patterns, venue landscapes, and thematic foci. Results reveal an explosive 125-fold increase in annual output, led by a Sino-US duopoly of institutions and dominated by Chinese- and Korean-surnamed authors. Publication venues are overwhelmingly engineering- or health-oriented, with negligible representation from linguistics, education, or socio-technical outlets. Thematically, the corpus clusters around algorithmic optimisation and downstream applications in education and healthcare, while exhibiting striking absences of ethics, bias, multilingualism, and non-English contexts. These patterns signal a widening gap between technical capability and socio-ethical grounding. We conclude by proposing an integrated research agenda that reframes prompt engineering as a three-tier socio-technical design challenge—encompassing micro-level token optimisation, meso-level task–AI fit, and macro-level people–AI fit—so that future systems amplify human creativity and learning without entrenching new digital divides.

Ähnliche Arbeiten