[Paper Review] Generative AI for Requirements Engineering: A Systematic Literature Review

Haowei Cheng, Jati H. Husen+5 more|arXiv (Cornell University)|Sep 10, 2024

Big Data and Business Intelligence5 citations

TL;DR

An SLR analyzing 27 primary studies on Generative AI in Requirements Engineering, highlighting focus on elicitation/analysis, GPT-era dominance, and ongoing challenges in domain specificity, interpretability, and ethics.

ABSTRACT

Introduction: Requirements engineering faces challenges due to the handling of increasingly complex software systems. These challenges can be addressed using generative AI. Given that GenAI based RE has not been systematically analyzed in detail, this review examines related research, focusing on trends, methodologies, challenges, and future directions. Methods: A systematic methodology for paper selection, data extraction, and feature analysis is used to comprehensively review 238 articles published from 2019 to 2025 and available from major academic databases. Results: Generative pretrained transformer models dominate current applications (67.3%), but research remains unevenly distributed across RE phases, with analysis (30.0%) and elicitation (22.1%) receiving the most attention, and management (6.8%) underexplored. Three core challenges: reproducibility (66.8%), hallucinations (63.4%), and interpretability (57.1%) form a tightly interlinked triad affecting trust and consistency. Strong correlations (35% cooccurrence) indicate these challenges must be addressed holistically. Industrial adoption remains nascent, with over 90% of studies corresponding to early stage development and only 1.3% reaching production level integration. Conclusions: Evaluation practices show maturity gaps, limited tool and dataset availability, and fragmented benchmarking approaches. Despite the transformative potential of GenAI based RE, several barriers hinder practical adoption. The strong correlations among core challenges demand specialized architectures targeting interdependencies rather than isolated solutions. The limited deployment reflects systemic bottlenecks in generalizability, data quality, and scalable evaluation methods. Successful adoption requires coordinated development across technical robustness, methodological maturity, and governance integration.

Motivation & Objective

• Assess the current state of GenAI applications in Requirements Engineering (RE).
• Identify research trends, venues, and geographic distribution of GenAI-for-RE studies.
• Catalog predominant methods, models, and techniques used in GenAI-for-RE.
• Evaluate the quality and limitations of existing GenAI-for-RE research.
• Propose future research directions and responsible AI integration in RE practices.

Proposed method

• Systematic literature review following ISO/IEC/IEEE 29148:2018 alignment for RE processes.
• Scopus as primary search source, extended with ArXiv and Google Scholar.
• Inclusion criteria: peer-reviewed English publications (2019–2024) related to GenAI in RE; Exclusion: non-peer-reviewed, irrelevant, duplicates, gray literature.
• Screened 42 papers; 27 primary studies included, with independent screening by three researchers.
• Data extraction and synthesis focused on RE phases, GenAI models/techniques, and implementation/adoption challenges.
• Discussed future directions, ethical considerations, and human–AI collaboration in GenAI-for-RE.

Experimental results

Research questions

• RQ1 RQ1: What are the current research trends in applying GenAI to RE (publication venues, timelines, geographic distribution)?
• RQ2 RQ2: What are the predominant approaches and techniques used in GenAI for RE (models, prompt engineering, fine-tuning, workflows)?
• RQ3 RQ3: How is the quality of current GenAI-for-RE research evaluated (methodology, goals, rigor)?
• RQ4 RQ4: What are the main challenges and future directions for applying GenAI to RE (technical, ethical, practical dimensions)?

Key findings

• There is a predominant focus on early RE stages, especially elicitation and analysis.
• Large language models, particularly GPT-series, dominate the GenAI-for-RE landscape.
• Challenges persist in domain-specific application, interpretability, and the reliability of AI-generated outputs.
• Ethical, security, privacy, and bias concerns are recurrent challenges for GenAI in RE.
• There is a call for comprehensive evaluation frameworks and improved human–AI collaboration models.