Gender Differences in the Adoption of Prompt Engineering in Generative AI.

Abstract

Introduction: Generative artificial intelligence (AI) has rapidly reshaped digital work and learning, and prompt engineering the deliberate design and iterative optimisation of inputs to generative models has emerged as a key, practice-oriented competency. Concurrently, substantial literature on digital skills and AI adoption documents persistent gendered disparities in access, participation and confidence. Yet, systematic synthesis focused specifically on gender differences in the adoption of prompt engineering remains limited. This study synthesises existing scholarship, industry reports and platform data to characterise the state of knowledge and to identify levers for more equitable AI literacy.

Objectives:  To (1) review and synthesise literature on the concept and evolution of prompt engineering; (2) examine studies and reports on gender differences in digital technology and AI adoption; (3) analyse secondary datasets and industry surveys to infer patterns in awareness and adoption of prompt engineering across genders; (4) evaluate the applicability of TAM and UTAUT theoretical frameworks to prompt-engineering adoption; (5) identify critical factors and barriers that shape gender disparities; and (6) propose evidence-based strategies for promoting gender-inclusive AI literacy.

Methods: A systematic secondary-research approach was employed. Searches were conducted across academic databases and institutional repositories for literature dated 2015–2025, using targeted keywords (e.g., “prompt engineering,” “generative AI,” “gender AND AI adoption,” “TAM,” “UTAUT”). Inclusion criteria prioritised peer-reviewed studies, reputable industry reports and policy documents. Extracted materials were coded into thematic categories and analysed through qualitative content analysis, thematic synthesis and comparative evaluation, with interpretive mapping onto TAM/UTAUT constructs.

Results:  The synthesis shows that prompt engineering is best characterised as a socio-technical, transdisciplinary skill combining domain knowledge, communicative framing and iterative model-testing. Secondary evidence indicates gendered patterns: women are under-represented in many GenAI technical courses and AI engineering roles (platform analytics and workforce reports suggest female shares near one-third in many GenAI enrollments), report higher AI-related anxiety and lower self-efficacy, and have less exposure to developer communities that foster tacit learning. Triangulated inferences identify (a) an exposure differential favoring men; (b) motivational divergence—men respond strongly to performance expectancy while women are more sensitive to effort expectancy and social facilitation; and (c) design sensitivity scaffolded curricula, no-code tools, cohort mentorship and domain-relevant modules disproportionately increase female participation. Major barriers include access/infrastructure gaps, educational pipeline deficits, time poverty, mentorship shortfalls and ethics/trust concerns. Recommended enablers include scaffolded entry points, contextualised micro-credentials, mentorship/cohort models, institutional supports (devices, time-flexible training, gender-disaggregated monitoring) and embedded ethics training. The review also identifies key gaps: objective, gender-disaggregated measures of prompt-engineering competence and longitudinal evaluations of training-to-outcome pathways are scarce.

Conclusions:  Prompt engineering is a teachable and high-value competency whose equitable diffusion depends on addressing structural, pedagogical and perceptual constraints. Interventions that reduce effort expectancy, strengthen social facilitation and increase facilitating conditions are theoretically and empirically justified to narrow gender gaps. To move from inference to evidence, future research should develop validated prompt-quality metrics, collect representative gender-disaggregated outcome data (including non-binary categories), and evaluate interventions via longitudinal and mixed-method designs.