Generative AI in Mobile Application Development: A Systematic Literature Review

Abstract

Generative Artificial Intelligence (GenAI) tools have become central to modern software engineering, fundamentally transforming how developers design, write, debug, test, and document code. This systematic literature review examines the measurable impact of GenAI tools — specifically GitHub Copilot, OpenAI GPT-4/ChatGPT, Google Gemini, Anthropic Claude 3, and Meta Code LLaMA — on Android and Flutter mobile application development. Following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) framework, we screened 120 candidate papers and identified 35 high-quality peer-reviewed publications from January 2020 to March 2026 across five major academic databases. The review addresses four research questions covering tool usage patterns, productivity gains, quality and security risks, and future research directions. Findings confirm mean productivity gains of 45–55% for code generation, 30–40% for debugging, 35–50% for UI/UX automation, and up to 60% for documentation writing. However, 25–40% of AI-generated code contains correctness issues and approximately 40% of security-critical AI code contains exploitable vulnerabilities. A novel Seven-Phase GenAI Integration Framework is proposed, mapping specific tools to each phase of the mobile development lifecycle. Six priority areas for future research are identified, including mobile-specific fine-tuned models, real-time RAG for live API documentation, and on-device privacy-preserving LLMs.

Introduction

The text examines the growing impact of Generative AI (GenAI) on mobile application development and presents a comprehensive review of its applications, benefits, risks, and future research directions.

Generative AI tools such as GitHub Copilot, OpenAI GPT-4, Google Gemini, Anthropic Claude, and Meta AI have rapidly become essential components of software development workflows. Their impact is particularly significant in mobile development, where developers face frequent API updates, complex frameworks, strict security requirements, and pressure to deliver applications quickly. Given the massive size of the mobile app market, even modest productivity improvements can generate substantial economic benefits.

Research shows that AI-assisted development significantly increases productivity. Studies found that developers using AI coding assistants completed tasks much faster than those working without AI support. However, these benefits are accompanied by risks. AI-generated code may contain security vulnerabilities, incorrect implementations, and quality issues. Some studies also revealed that developers using AI tools often become overconfident in the security of generated code, increasing the likelihood of deploying insecure applications.

The paper identifies a gap in existing research: most studies focus on specific tools or individual development tasks rather than examining how Generative AI supports the entire mobile application development lifecycle. To address this, the study investigates four key questions concerning AI usage, productivity gains, risks, and future research opportunities.

Background and Related Work

The foundation of modern GenAI tools is the transformer architecture introduced in 2017, which enables AI models to understand long-range relationships in code and natural language. Successive large language models have achieved remarkable improvements in code generation accuracy and contextual understanding.

The study also discusses major mobile development ecosystems:

• Android development primarily uses Kotlin, MVVM architecture, and Jetpack Compose.
• Flutter development uses Dart and supports cross-platform applications through a single codebase.

Research on AI-assisted mobile development demonstrates improvements in:

• Code generation
• Unit testing
• Bug fixing
• User interface creation
• Productivity enhancement

However, performance decreases when dealing with complex architectures such as Flutter's BLoC pattern, and security vulnerabilities remain a significant concern.

Research Methodology

The review follows the PRISMA systematic review framework and analyzes studies published between 2020 and 2026. After screening 120 papers, 35 high-quality studies were selected for detailed analysis. The research evaluates major GenAI tools across mobile development tasks while measuring factors such as task completion time, code correctness, and productivity gains.

Seven-Phase GenAI Integration Framework

The paper proposes a Seven-Phase GenAI Integration Framework, mapping AI tools to every stage of mobile application development:

1. Requirements and Planning
   • AI generates user stories, acceptance criteria, and requirement specifications.
   • Improves requirement completeness and supports small teams.
2. UI/UX Design
   • AI converts wireframes and design ideas into interface layouts.
   • Effective for standard designs but less reliable for custom animations and advanced layouts.
3. Code Generation
   • The most researched and productive application of GenAI.
   • AI significantly accelerates coding but may struggle with newly released APIs and complex architectures.
4. API and Backend Integration
   • AI assists in creating networking layers, authentication systems, and backend connections.
   • Human verification remains necessary for accuracy.
5. Testing and Quality Assurance
   • AI generates test cases and improves test coverage.
   • Often focuses on common scenarios while missing edge cases and unusual failures.
6. Debugging and Code Review
   • AI helps identify bugs, interface mismatches, and software defects.
   • Large-context models can analyze entire modules and improve troubleshooting.
7. Documentation and Deployment
   • AI automates documentation creation, setup guides, code comments, and deployment descriptions.
   • This phase shows some of the highest proportional productivity gains.

Conclusion

This systematic literature review has synthesised evidence from 35 peer-reviewed publications to provide a comprehensive, lifecycle-spanning assessment of the impact of Generative AI on Android and Flutter mobile application development. The evidence base supports the following conclusions with high confidence. Productivity gains from GenAI tools are real, substantial, and consistent across tools, task types, and study designs. The central estimate of 45–55% time reduction for code generation tasks is supported by 18 independent studies and is independently confirmed by experimental data. Documentation generation offers the highest ceiling gain (up to 60%) yet remains the least-studied use case — a mismatch that represents both a research gap and a high-value adoption opportunity for practitioners. The risks of GenAI adoption are equally real and should not be minimised. Correctness issues affect 25–40% of AI-generated code in production contexts. Security vulnerabilities are present in approximately 40% of security-critical AI-generated code, and AI-assisted developers are simultaneously more likely to produce insecure code and more confident in its security. Developer skill atrophy from uncritical over-reliance is empirically documented. The Seven-Phase GenAI Integration Framework proposed in this review provides the field\'s first lifecycle-spanning, evidence-based tool-to-phase mapping for mobile development, serving as a practical starting point for strategic — rather than naive — AI adoption. The most important conclusion of this review is directional: Generative AI tools are powerful accelerators for developers who already possess the knowledge to evaluate, refine, and correctly integrate AI-generated output. They are not a substitute for foundational software engineering expertise. AI accelerates experienced developers; it does not replace missing knowledge.

References

[1] Statista. (2025). Number of mobile apps available in leading app stores as of Q1 2025. Statista Research Department. [2] T. Brown et al., \'Language Models are Few-Shot Learners,\' Advances in Neural Information Processing Systems, vol. 33, pp. 1877–1901, 2020. [3] A. Vaswani et al., \'Attention Is All You Need,\' Advances in Neural Information Processing Systems, vol. 30, pp. 5998–6008, 2017. [4] OpenAI, \'GPT-4 Technical Report,\' arXiv:2303.08774, 2023. [5] Grand View Research. (2024). Mobile Application Market Size, Share & Trends Analysis Report. Report ID: GVR-4-68038-924-4. [6] M. Chen et al., \'Evaluating Large Language Models Trained on Code,\' arXiv:2107.03374, 2021. [7] Y. Liu and S. R. Kochhar, \'No More Manual Tests? Evaluating ChatGPT for Unit Test Generation,\' arXiv:2305.04207, 2023. [8] R. Patel and A. Sharma, \'AI-Assisted Flutter Development: A Systematic Evaluation of GitHub Copilot,\' Journal of Mobile Computing and Communications, vol. 12, no. 3, 2024. [9] H. Pearce et al., \'Asleep at the Keyboard? Assessing the Security of GitHub Copilot\'s Code Contributions,\' Proceedings of the 2022 IEEE Symposium on Security and Privacy (S&P), pp. 754–768, 2022. [10] M. Arora and P. Gupta, \'AI-Assisted Requirements Engineering: A Systematic Survey and Taxonomy,\' IEEE Access, vol. 11, 2023. [11] E. Kalliamvakou, \'Quantifying GitHub Copilot\'s Impact on Developer Productivity and Happiness,\' IEEE Software, vol. 39, no. 6, pp. 35–43, 2022. [12] M. L. Siddiq and J. C. Santos, \'Exploring the Effectiveness of Large Language Models in Generating Unit Tests,\' arXiv:2305.00418, 2023. [13] GitHub. (2023). The State of the Octoverse 2023: Security, AI, and Developer Trends. GitHub Inc. [14] Z. Ji et al., \'Survey of Hallucination in Natural Language Generation,\' ACM Computing Surveys, vol. 55, no. 12, 2023. [15] B. Prather et al., \'The Robots Are Coming: Exploring the Implications of OpenAI Codex on Introductory Programming,\' Proceedings of the 54th ACM Technical Symposium on Computer Science Education (SIGCSE 2023). [16] R. Vyas et al., \'Analyzing Copyright and Intellectual Property Issues in AI-Generated Code,\' IEEE Software, vol. 41, no. 1, 2024. [17] D. Sobania et al., \'An Analysis of the Automatic Bug Fixing Performance of ChatGPT,\' Proceedings of the 2023 IEEE/ACM Workshop on Automated Program Repair (APR 2023). [18] L. Zhang et al., \'Sketch2Code: Transforming Hand-Drawn Wireframes to UI Code using Generative AI,\' arXiv:2310.13811, 2023. [19] N. Perry et al., \'Do Users Write More Insecure Code with AI Assistants?\' Proceedings of the ACM SIGSAC Conference on Computer and Communications Security (CCS 2023). [20] A. Mastropaolo et al., \'Using Deep Learning to Generate Complete Log Statements for Source Code Methods,\' Proceedings of ICSE 2022. [21] Anthropic. (2024). Claude 3 Model Card and System Prompt. Anthropic Technical Report.

Copyright

Copyright © 2026 Balram Kumar, Prof. Sangeeta Rani. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.