Paper deep dive
The State of Generative AI in Software Development: Insights from Literature and a Developer Survey
Vincent Gurgul, Robin Gubela, Stefan Lessmann
Intelligence
Status: succeeded | Model: google/gemini-3.1-flash-lite-preview | Prompt: intel-v1 | Confidence: 95%
Last extracted: 3/22/2026, 5:42:38 AM
Summary
This study investigates the impact of Generative AI (GenAI) on the Software Development Lifecycle (SDLC) through a systematic literature review and a survey of 65 software developers. Findings indicate that GenAI significantly boosts productivity in design, implementation, testing, and documentation, though benefits are lower in planning and requirements analysis. While 79% of developers use GenAI daily, the study highlights critical risks including skill erosion, technical debt, and security vulnerabilities, emphasizing the necessity of robust governance and human-in-the-loop mechanisms.
Entities (5)
Relation Signals (4)
Software Developers â uses â Generative AI
confidence 98% · 79 % of survey respondents use GenAI daily
Generative AI â impacts â Design and Implementation
confidence 95% · GenAI exerts its highest impact in design, implementation, testing, and documentation
Generative AI â transforms â Software Development Lifecycle
confidence 95% · Generative Artificial Intelligence (GenAI) rapidly transforms software engineering
Organizations â implements â Governance
confidence 90% · two-thirds of organizations maintaining formal or informal guidelines
Cypher Suggestions (0)
No Cypher suggestions yet.
Abstract
Abstract:Generative Artificial Intelligence (GenAI) rapidly transforms software engineering, yet existing research remains fragmented across individual tasks in the Software Development Lifecycle. This study integrates a systematic literature review with a survey of 65 software developers. The results show that GenAI exerts its highest impact in design, implementation, testing, and documentation, where over 70 % of developers report at least halving the time for boilerplate and documentation tasks. 79 % of survey respondents use GenAI daily, preferring browser-based Large Language Models over alternatives integrated directly in their development environment. Governance is maturing, with two-thirds of organizations maintaining formal or informal guidelines. In contrast, early SDLC phases such as planning and requirements analysis show markedly lower reported benefits. In a nutshell, GenAI shifts value creation from routine coding toward specification quality, architectural reasoning, and oversight, while risks such as uncritical adoption, skill erosion, and technical debt require robust governance and human-in-the-loop mechanisms.
Tags
Links
- Source: https://arxiv.org/abs/2603.16975v1
- Canonical: https://arxiv.org/abs/2603.16975v1
Trouble viewing inline? Open PDF directly â
Full Text
57,309 characters extracted from source content.
Expand or collapse full text
The State of Generative AI in Software Development: Insights from Literature and a Developer Survey A Preprint Vincent Gurgul Chair of Information Systems Humboldt-UniversitĂ€t zu Berlin Unter den Linden 6, 10117 Berlin Robin Gubela Chair of Information Systems Hochschule fĂŒr Technik und Wirtschaft Berlin Treskowallee 8, 10318 Berlin Stefan Lessmann Chair of Information Systems Humboldt-UniversitĂ€t zu Berlin Unter den Linden 6, 10117 Berlin Bucharest University of Economic Studies 6 Piata Romana, 010374, Romania March 19, 2026 Abstract Generative Artificial Intelligence (GenAI) rapidly transforms software engineering, yet existing research remains fragmented across individual tasks in the Software Development Lifecycle. This study integrates a systematic literature review with a survey of 65 software developers. The results show that GenAI exerts its highest impact in design, implementation, testing, and documentation, where over 70 % of developers report at least halving the time for boilerplate and documentation tasks. 79 % of survey respondents use GenAI daily, preferring browser-based Large Language Models over alternatives integrated directly in their development environment. Governance is maturing, with two-thirds of organizations maintaining formal or informal guidelines. In contrast, early SDLC phases such as planning and requirements analysis show markedly lower reported benefits. In a nutshell, GenAI shifts value creation from routine coding toward specification quality, architectural reasoning, and oversight, while risks such as uncritical adoption, skill erosion, and technical debt require robust governance and human-in-the-loop mechanisms. Keywords Generative Artificial Intelligence·Software Engineering·Large Language Models·AI-assisted Coding· IT Governance· Agile Development 1 Introduction The rapid adoption of Generative Artificial Intelligence (GenAI), particularly Large Language Models (LLMs), is substantially transforming the Software Development Life Cycle (SDLC). Chui et al. [2023] identify software engineering as one of the domains with the highest economic impact potential from GenAI. AI-assisted tools such as GitHub Copilot are increasingly embedded in professional software engineering environments, supporting activities that span the entire lifecycleâfrom requirements analysis to architectural design, code generation, testing, debugging, documentation, and maintenance. GenAI refers to a class of machine learning systems capable of producing novel content (such as text, images, code, or sound) by learning patterns from large datasets and generating contextually coherent outputs [Goodfellow et al., 2016]. Prominent examples include LLMs such as ChatGPT, which process and generate natural language. While these systems do not replace human judgment or accountability, they might augment arXiv:2603.16975v1 [cs.SE] 17 Mar 2026 The State of Generative AI in Software Development: Insights from Literature and a Developer Survey developer capabilities, potentially enabling faster iteration cycles, improved efficiency, and higher software quality. At the same time, its adoption raises questions related to maintainability, security, intellectual property, transparency, regulatory compliance, and long-term developer skill development. Despite growing interest, the literature remains fragmented: Studies typically focus on aspects related to isolated SDLC phases, while comprehensive assessments across the full lifecycle with empirical data remain scarce. Research agendas such as Feuerriegel et al. [2024] call for more empirical investigations into how GenAI transforms organizational work practices and, specifically, how such systems affect software development productivity. To address these gaps, we derive the following research questions. RQ1: How are GenAI tools currently used by software developers? RQ2: How is the impact of GenAI perceived across the phases of the SDLC? RQ3: What governance mechanisms are implemented to manage the adoption of GenAI in software development departments? RQ4: What are the perceived risks associated with GenAI in software development? This study addresses these questions by integrating three complementary sources of evidence. First, a structured review of academic literature synthesizes current scholarly knowledge on the potential and limitations of GenAI across the phases of the SDLC and in agile development. Second, industry reports are incorporated to capture emerging practices, technological developments, and practitioner-oriented assessments that often precede academic publication. Third, a cross-sectional survey of 65 software developers building on the Technology Acceptance Model (TAM) [Davis, 1989, Venkatesh and Davis, 2000] provides primary empirical evidence on real-world tool usage, perceived productivity effects, governance maturity, and associated risks. By integrating insights from academic research, industry discourse, and practitioner experience, the study enables a more comprehensive and empirically grounded assessment of how GenAI is reshaping development practices, influencing role profiles, and altering governance requirements. 2 Methodology for Literature Analysis This section outlines the methodology used to identify and select relevant studies for the literature analysis presented in Section 3. Our literature selection process follows established methodological guidelines for conducting systematic literature reviews in Information Systems research [Webster and Watson, 2002, Okoli, 2015]. Following the typology proposed by ParĂ© et al. [2015], who distinguish nine literature review types, ours can be characterized as a scoping review: Its primary goal is to identify and structure the current body of research on GenAI across the SDLC, rather than to aggregate quantitative data or to develop new theoretical propositions. By organizing the literature along the SDLC phases, we follow the concept-centric approach suggested by Webster and Watson [2002], shifting the focus from what individual authors have found to what is collectively known about each stage of the development process. The selection criteria included peer-reviewed journal articles, conference proceedings, preprints, and technical reports published by established consulting and research organizations. Extending the evidence base beyond academic publications is in line with Okoli [2015], who acknowledges non-peer-reviewed sources as part of the searchable evidence base in systematic reviews. Including such literature is relevant for our research in the rapidly evolving research field of GenAI, where much novel knowledge is not yet reflected in peer-reviewed publications. Furthermore, only publications released between 2021 and 2026 were considered, reflecting the period during which large-scale GenAI systems became widely accessible. The primary databases searched were Scopus, IEEE Xplore, the ACM Digital Library, SpringerLink, arXiv, and Google Scholar. These sources were chosen to ensure broad coverage of software engineering, information systems, and applied AI research. Search queries combined general and domain-specific keywords. Core search terms included âAI-assisted codingâ, âGitHub Copilotâ, âGenerative AIâ, âlarge language modelsâ, âagile frameworkâ and âsoftware development toolsâ. These were complemented by phase-specific terms corresponding to the SDLC, such as ârequirements engineeringâ, âsoftware designâ, âcode generationâ, âtestingâ, âdebuggingâ, âmaintenanceâ and âproject managementâ. In addition, selected industry reports from consulting and technology firms such as McKinsey, Boston Consulting Group (BCG), Deloitte, and Accenture were included, as these publications often provide early insights into technological adoption patterns and practical challenges that precede academic research. The screening and selection process followed the three-stage framework of Levy and J. Ellis [2006] following prior literature screening approaches. In the first stage (input), the initial search across all databases yielded 2 The State of Generative AI in Software Development: Insights from Literature and a Developer Survey 959,044 publications. In the second stage (processing), duplicate records across databases, non-English manuscripts, and publications prior to 2019 were removed, and results were filtered by publication type following established literature screening procedures [for example Fröhnel et al., 2024]. This step reduced the corpus to 1,712 candidate publications. The remaining records were then screened in two stages: first, titles and abstracts were reviewed for topical relevance, resulting in 160 candidate publications. Second, the remaining articles underwent full-text screening. Selection criteria included relevance to the impact of GenAI on one or more SDLC phases, methodological transparency, and substantive contribution to understanding technical, organizational, or managerial implications. Studies focusing solely on unrelated AI applications or lacking sufficient methodological clarity were excluded. We identified additional publications through backward search [Webster and Watson, 2002] by reviewing the reference lists of the remaining studies and applying the same selection criteria as before. The screening by full-text resulted in 63 publications included in the review. In the third stage, the final corpus of publications was organized according to global observations, individual SDLC phases, agile methodologies, and risk analysis, as presented subsequently. 3 Literature Review The rapid advancement of GenAI is increasingly transforming software development practices across the SDLC. Among AI-assisted development tools, conversational LLMs such as ChatGPT and assistants integrated directly in the development environment like GitHub Copilot, Amazon CodeWhisperer, JetBrains AI Assistant, and Tabnine have become widely adopted, supporting tasks such as code generation, explanation, and review [Sergeyuk et al., 2025]. Empirical studies and field research consistently report productivity gains across development activities, particularly in code optimization (approximately 60 %), bug fixing (26 %), and documentation support (12 %) [Collante et al., 2025]. Evidence further suggests that less experienced developers benefit disproportionately from AI assistance, achieving higher relative productivity improvements [Brynjolfsson et al., 2023, DellâAcqua et al., 2023, Ng et al., 2024]. At the same time, studies indicate that code quality generally remains stable or improves despite reduced development time [Shihab et al., 2025, Yadav and Mondal, 2025]. However, correctness cannot be guaranteed and long-term risks remain, making human validation and oversight essential [YetiĆtiren et al., 2023, Atif et al., 2025]. Beyond individual productivity, industry reports highlight broader organizational implications, including accelerated prototyping, shorter time-to-market, and shifting accountability toward product and project leadership [Deniz et al., 2023, Gnanasambandam et al., 2025]. The following subsections synthesize the literature across the phases of the SDLC, before addressing agile software development as well as the risks and roadblocks associated with GenAI adoption. 3.1 GenAI in the Software Development Lifecycle 3.1.1 Planning. In the planning phase, GenAI supports the creation of foundational project artifacts such as scope definitions, timelines, milestone structures, role descriptions, communication plans, and risk assessments [Barcaui and Monat, 2023, Hughes et al., 2025, Maggoo et al., 2025]. AI systems are also used to conduct market analyses, synthesize competitive information, and formulate product goals and visions [Lechner et al., 2025, DellâAcqua et al., 2023]. Cost estimation and story point approximation have likewise been explored as application areas [Wagner and Wagner, 2024]. Empirical evidence suggests considerable productivity gains during project initiation and planning, including faster setup of new projects and improved alignment between technical teams and management [Aramali et al., 2025, Brynjolfsson et al., 2023, Hughes et al., 2025]. GenAI is typically deployed via structured prompts and increasingly integrated into project management tools such as Jira, Confluence, or Microsoft Project [Barcaui and Monat, 2023, Ng et al., 2024]. These capabilities contribute to early identification of risks and structured project preparation. 3.1.2 Requirements Analysis. During requirements analysis, GenAI assists in drafting, refining, and validating functional and technical requirements. LLMs are used to generate and elaborate user stories and use cases based on high-level project documentation [Cico et al., 2023, Pinto et al., 2023, Wei et al., 2025] and to create preliminary architectural drafts and structured feature specifications [Glever et al., 2024, Gong et al., 2025, Nitin, 2024]. AI-assisted approaches reduce conceptualization time and accelerate the creation of prototypes, including 3 The State of Generative AI in Software Development: Insights from Literature and a Developer Survey visually oriented UI/UX artifacts [Wei, 2024, Wei et al., 2025, Gnanasambandam et al., 2025]. Tools such as Figma AI and related plugins seem promising implementations for rapid prototyping [Petridis et al., 2024]. Furthermore, AI systems can analyze requirement documents to identify inconsistencies, ambiguities, and missing elements, enabling earlier error detection and reducing downstream rework [Gong et al., 2025, Porter et al., 2025]. Improved consistency between business and IT stakeholders has also been reported [Pinto et al., 2023, Brynjolfsson et al., 2023]. 3.1.3 Design and Implementation. The design and implementation phase represents the most extensively studied application domain for GenAI. AI-assisted coding tools such as GitHub Copilot provide real-time code completion, refactoring suggestions, boilerplate generation, and context-aware optimization [Peng et al., 2023, YetiĆtiren et al., 2023, Solohubov et al., 2023, Glever et al., 2024, Gerdemann et al., 2024]. They also support automatic documentation generation, including code comments and API documentation [Atif et al., 2025, Pinto et al., 2023, Ndiaye et al., 2025], as well as recommendations to improve code structure and maintainability [Gong et al., 2025, Zhong et al., 2025]. Experimental and field studies consistently report increased development speed [Peng et al., 2023, Cui et al., 2024, Shihab et al., 2025, Struever et al., 2025, Deniz et al., 2023]. Routine tasks are particularly affected, with automation reducing manual effort for standardized coding activities [Solohubov et al., 2023]. Several analyses also indicate improvements in review efficiency and defect reduction, although human oversight remains essential [Collante et al., 2025, YetiĆtiren et al., 2023, Ndiaye et al., 2025]. Implementation typically occurs via integration into Integrated Development Environments (IDEs) such as Visual Studio Code or IntelliJ, often combined with secure APIs or isolated models to address data protection and compliance requirements [Ng et al., 2024, Ebert and Louridas, 2023]. 3.1.4 Testing and Integration. In the testing and integration phase, GenAI is used to automatically generate unit, integration, and end-to-end test cases [Atif et al., 2025, Glever et al., 2024, Maggoo et al., 2025, Zhong et al., 2025]. Additional applications include the creation of mock objects and synthetic test data [YetiĆtiren et al., 2023, Gong et al., 2025], suggestions for missing test paths and coverage gaps [Cico et al., 2023, Nitin, 2024], and AI-assisted security analysis [Ding et al., 2024, Pearce et al., 2025, Ndiaye et al., 2025]. The literature documents reductions in manual testing effort for routine scenarios and broader, faster test coverage [Kathiresan, 2024, Solohubov et al., 2023, Zhong et al., 2025] as well as earlier detection of edge cases and integration issues [YetiĆtiren et al., 2023, Kathiresan, 2024]. In practice, GenAI is often integrated into CI/CD pipelines and connected to test frameworks or embedded directly into development environments for automated code analysis and test case generation [Ng et al., 2024, Pinto et al., 2023]. 3.1.5 Operation and Maintenance. In the operation and maintenance phase, GenAI supports the analysis of logs, error messages, and stack traces, providing diagnostic insights and suggesting potential remediation steps based on learned patterns [Ng et al., 2024, Gong et al., 2025, Zhong et al., 2025]. It is also used to explain, summarize, and restructure poorly documented or complex legacy code, improving maintainability and knowledge transfer [Nitin, 2024, Atif et al., 2025, Maggoo et al., 2025]. Reported effects include faster issue detection and resolution, improved traceability, and enhanced resilience to staff turnover due to automated documentation and knowledge preservation [Ng et al., 2024, Pinto et al., 2023, Nitin, 2024]. Integration commonly occurs within incident and log management platforms such as Jira, ServiceNow, or Grafana, and in sensitive environments through locally deployed or access-controlled AI systems to mitigate compliance risks [Ng et al., 2024, Ebert and Louridas, 2023]. 3.2 GenAI in Agile Software Development GenAI influences the dynamics of agile development practices beyond productivity gains. The literature reports high perceived usefulness and satisfaction in agile teams when using AI-assisted tools [Geyer et al., 2025] and accelerated task completion, leading to shorter iteration cycles and more frequent releases [Zhang et al., 2024, Ulfsnes et al., 2024, Bahi et al., 2024]. By automating routine coding, documentation, and testing tasks, GenAI enables teams to focus on value-generating increments, reinforcing core agile principles such as rapid feedback and continuous delivery. At the same time, role expectations shift: Product owners, 4 The State of Generative AI in Software Development: Insights from Literature and a Developer Survey developers, and scrum masters increasingly engage in strategic, creative, and evaluative activities, while routine tasks are partially delegated to AI systems [Diebold, 2025, Bahi et al., 2024, Nasir and Hussain, 2024]. The literature recommends adapting role profiles and strengthening AI-related competencies, particularly critical evaluation skills in AI-supported environments. 3.3 Risks and Roadblocks in the Adoption of GenAI Despite substantial productivity gains, the literature identifies a range of short- and long-term risks associated with the integration of GenAI into software development. In the short term, concerns relate to reliability, quality control, and security. AI-generated code suggestions are frequently erroneous, with manual correction efforts averaging approximately ten minutes per identified defect [YetiĆtiren et al., 2023]. A significant proportion of outputs requires post-editing before productive use [Ziegler et al., 2022]. Developers often accept AI-generated snippets with limited scrutiny, suggesting superficial quality control practices [Ziegler et al., 2022, Chen et al., 2021]. Perceived productivity improvements do not always align with objective activity metrics, raising the risk of overestimating efficiency gains [Stray et al., 2025]. Security vulnerabilities represent a further concern: coding assistants can be susceptible to prompt injection attacks, potentially leading to data contamination, unintended code execution, or data leakage [Liu et al., 2025, Cotroneo et al., 2024, Norton et al., 2025]. Long-term risks are more structural. Continuous reliance on AI-generated suggestions may erode deep technical knowledge and increase dependency on AI systems [Ng et al., 2024, Brynjolfsson et al., 2023, Ding et al., 2025]. Automatically generated code can introduce opaque dependencies and architectural inconsistencies, contributing to technical debt [Ebert and Louridas, 2023, Atif et al., 2025, Anderson et al., 2025, Moreschini et al., 2026]. Even syntactically correct code may contain subtle semantic or security-relevant weaknesses [YetiĆtiren et al., 2023, Atif et al., 2025]. Organizational dependence on specific vendors poses additional strategic risks, including technology lock-in and compliance exposure [Ng et al., 2024]. At the team level, some studies indicate reduced direct human interaction when AI tools mediate commu- nication, code explanation or problem-solving [Ulfsnes et al., 2024, Cabrero-Daniel et al., 2024], entailing risks of knowledge silos and diminished interpersonal trust. To mitigate these effects, research recommends maintaining structured exchange formats such as code reviews and sprint retrospectives. More broadly, gover- nance demands increase as GenAI becomes embedded in production workflows. Issues of quality assurance, accountability, compliance, and traceability become more salient when AI-generated artifacts are incorporated into production systems [Geyer et al., 2025, Zhang et al., 2024, Ulfsnes et al., 2024, Diebold, 2025, Bahi et al., 2024]. The literature emphasizes the need to formalize review obligations, clarify responsibility for AI-assisted outputs, and establish dedicated governance frameworks to prevent quality degradation and unmanaged risk exposure. Several roadblocks constrain the realization of GenAIâs full potential. A large-scale industry survey [Ahlawat et al., 2024] shows that coding accounts for only 10â15 % of total development time and less than 50 % of developers use GenAI. They conclude, that improvements in code generation alone have limited overall impact. To prevent automation risks from offsetting productivity gains, a human-in-the-loop approach (including quality review, domain judgment, and requirements oversight) is emphasized by industry reports [Deniz et al., 2023, Ahlawat et al., 2024, Struever et al., 2025]. Overall, the literature indicates that GenAI exerts measurable influence across all phases of the SDLC, with the strongest effects observed in design, implementation, and testing. However, these gains are neither uniform nor automatically realizedâshort-term limitations in reliability and security, long-term risks such as skill erosion and technical debt as well as organizational roadblocks may constrain or offset productivity improvements. Effective governance frameworks, structured human-in-the-loop validation, and deliberate organizational adaptation remain critical to translating GenAIâs technical capabilities into lasting benefits. 4 Survey Design To complement the literature analysis with practitioner perspectives, we conduct a standardized online survey informed by the TAM [Davis, 1989, Venkatesh and Davis, 2000]. TAM provides a widely used framework for examining how users perceive the usefulness and adoption of new technologies in organizational contexts. The survey investigates how software practitioners use GenAI tools, how they assess productivity effects across different SDLC phases, how governance mechanisms for GenAI adoption are implemented in practice, and which risks practitioners associate with the use of GenAI. 5 The State of Generative AI in Software Development: Insights from Literature and a Developer Survey Data were collected using a self-developed online questionnaire. The instrument comprised five sections: (1) demographic and professional background variables (age, education, role, team size); (2) primary development methodology (agile, traditional, hybrid) and degree of adherence; (3) GenAI tool usage behavior and frequency; (4) assessments of productivity, quality, and security implications across SDLC phases; and (5) time-based efficiency estimations and open-ended questions on perceived opportunities and risks. A pretest (n=3) was conducted to refine wording and improve the distinction between process model categories. The target population comprised professionals involved in software development and related disciplines, including software engineers, architects, team leads, DevOps engineers, product owners, QA engineers, as well as students and interns. Data collection took place between 31 December 2025 and 19 January 2026 using a multi-stage snowball sampling approach [Döring, 2023] through professional and academic networks, including direct outreach on LinkedIn. While the primary focus was the DACH region, participation from other regions was not restricted. 5 Survey Results In the following we present the findings of our quantitative online survey (n = 65). Although the sample exhibits a comparatively young age structure, it demonstrates a strong professional profile. The majority of respondents are industry practitioners, including 30 professional software developers, one team lead, and four product owners. Notably, ten participants report more than ten years of professional experience, indicating that the sample includes senior expertise despite its youthful demographic composition. Figure 5.1: Development methodology and age distribution With regard to development methodologies, the agile approaches Scrum and Kanban constitute the largest share at about 42 %. Approximately 34 % report operating without a clearly defined process model, followed by 17 % of respondents who use a hybrid approach with both agile and traditional (waterfall-based) methodologies. The comparably smallest group with about 8 % of respondents work in formalized or plan-driven (âTraditionalâ) environments. Figure 5.2: GenAI tool usage 6 The State of Generative AI in Software Development: Insights from Literature and a Developer Survey As Figure 5.2 presents, the survey reveals widespread integration of GenAI into daily development activities. Approximately 79 % of respondents report using AI tools at least once per day, indicating that GenAI has become a routine component of professional practice. In terms of tools, browser-based LLMs dominate. ChatGPT is used by 77 % of participants, making it the most frequently adopted tool, followed by Google Gemini/Bard at 52 %. GitHub Copilot, which combines IDE-integrated code completion with a conversational chat interface, is used by about 48 % of respondents. IDE-integrated coding assistants such as JetBrains AI Assistant (11 %), Tabnine (3 %), and Amazon CodeWhisperer (2 %) remain marginal by comparison. 17 % of respondents reported using other tools while 3 % did not respond to the tool-related question. This distribution suggests that developers favor dialog-based interaction with general-purpose LLMs over specialized IDE-integrated alternatives. Figure 5.3: Impact of GenAI by SDLC phase Figure 5.3 displays the distribution of responses regarding the SDLC phases in which participants perceived the greatest utility of GenAI (multiple choice). 85 % of respondents identify âDesign and Implementationâ as the phase with the strongest perceived benefit, followed by âOperation and Maintenanceâ at 75 %. Earlier lifecycle phases receive notably lower ratings, with âRequirements Analysisâ at 43 %, âTesting and Integrationâ at roughly 37 %, and âPlanningâ at 18 %. Although the low value for âPlanningâ might be attributed to the high proportion of junior software developers in the sample, none of the Product Owners and only 5 % of respondents with more than ten years of experience report a perceived impact in this phase, suggesting the opposite. Figure 5.4: Estimated time saving with GenAI by coding task (100 % = no time saving, 0 % = full automation) 7 The State of Generative AI in Software Development: Insights from Literature and a Developer Survey To quantify these benefits at the task level, respondents were asked to estimate the time required for six common development activities when using GenAI in relation to working without AI support (see Figure 5.4). The results show that the strongest effects are reported for writing boilerplate code and documentation, where 72 % and 69 % of respondents, respectively, estimate at least halving the required time. These two tasks fall under design and implementation as well as operation and maintenance; the two SDLC phases also rated highest in Figure 5.3. Writing new code shows comparably more moderate positive effects, with the largest group (37 %) reporting a notable reduction to approximately 75 %. Creating unit tests also shows substantial gains among the respondents, although the high âcannot assessâ rate (28 %) suggests that not all respondents regularly perform this activity. Understanding unfamiliar or legacy code produces the most polarized responses, with 29 % reporting extreme speedup to less than 25 % but also the highest âno changeâ rate (11 %) across all tasks. Debugging yields the weakest perceived savings and the highest share of respondents reporting increased effort (6 %), consistent with the cognitively complex nature of this task. Together, those findings confirm the primary role of GenAI as a coding assistant that accelerates repetitive and standardized development tasks, while also suggesting substantial perceived value in operation and maintenance activities such as debugging, log analysis, and legacy code comprehension. Figure 5.5: GenAI policies and guidelines in the respondentâs companies As Figure 5.5 shows, the survey results indicate a comparatively mature governance landscape for GenAI usage across organizations. The largest share of respondents (40 %) report the existence of official tools combined with a formal AI policy. An additional 26 % indicate the presence of informal recommendations, suggesting that in roughly two thirds of organizations AI usage is at least partially embedded within recognized guidance structures. However, governance is not uniformly institutionalized. Around 20 % of participants state that they are unaware of any existing policies, and approximately 8 % report that no regulation exists at all. Smaller proportions indicate no response, other miscellaneous categories or that AI tools are explicitly forbidden in the company. This distribution reveals a dual dynamic: While structured governance mechanisms are already widespread, a non-negligible segment of organizations still exhibits regulatory ambiguity or limited transparency regarding AI usage policies. The qualitative responses largely reinforce the quantitative findings and provide deeper insight into perceived opportunities, risks, and anticipated role changes associated with GenAI in software development. Across roles and experience levels, the dominant theme is efficiency. Participants repeatedly emphasize time savings through automation of repetitive and low-complexity tasks such as boilerplate code generation, documentation, test creation, debugging support, and code explanation. Many respondents describe GenAI as a âsparring partnerâ that accelerates learning, facilitates onboarding into new technologies, and supports rapid prototyping. Experienced developers highlight its value in reviewing novel code, interfacing with unfamiliar systems, and exploring alternative solution paths. Several comments state that AI reduces âtoilâ and allows developers to focus more on architecture, system design, and complex problem solving. At the same time, concerns are substantial and remarkably consistent. The most frequently mentioned risks are uncritical adoption of AI-generated code, loss of deep technical understanding, and data protection and security vulnerabilities. Many respondents warn against âblind copyingâ and emphasize that AI outputs often appear syntactically correct but may contain logical, architectural, or security flaws. Senior practitioners stress that experience remains indispensable for evaluating AI-generated artifacts. A recurring theme is the 8 The State of Generative AI in Software Development: Insights from Literature and a Developer Survey danger of cognitive offloading: if AI is used as a shortcut rather than as a support tool, individual learning curves and long-term problem-solving capabilities may deteriorate. Data securityâparticularly the risk of exposing sensitive or company-specific information to external modelsâis also repeatedly cited as a major concern. Regarding the future of the profession, most participants do not anticipate the disappearance of software development but rather a structural transformation of the role. A common expectation is a shift from pure code writing toward architectural thinking, requirement formulation, quality assurance, and orchestration of AI systems. Several respondents predict a polarization effect: junior roles may decline in number, while senior and architect-level competencies become more valuable. Others foresee new hybrid roles, such as AI supervisors. While a minority expresses skepticism or uncertainty about long-term effects, the prevailing view is that GenAI will become a standard tool, and that developers who fail to integrate it into their workflows may fall behind. 6 Conclusion We examined the impact of GenAI on the SDLC by integrating a systematic literature review with empirical evidence from a cross-sectional survey of 65 software developers. The findings consistently demonstrate that AI-assisted coding tools exert substantial influence across multiple SDLC phases, with the most pronounced and empirically validated effects observed in design, implementation, testing, and documentation. We determine that prior research reports measurable productivity gains, reductions in routine workload, and generally stable or improved code quality when human validation mechanisms remain in place. The survey results largely corroborate these findings. Respondents identify implementation, boiler plate code, and documentation as the domains with the highest perceived value, confirming the role of GenAI as a productivity multiplier in syntactically structured and repetitive tasks. In contrast, less structured and evaluative tasks such as planning, requirements analysis, debugging, and testing show considerably lower perceived productivity gains. Beyond phase-specific effects, the study highlights structural and organizational dimensions of AI integration. GenAI has become deeply embedded in daily workflows, yet security concerns remain moderately elevated, indicating that developers are aware of potential risks despite widespread adoption. The governance analysis reveals that a majority of organizations have already established either formal AI policies or at least informal usage guidelines, yet, a meaningful minority report regulatory ambiguity or limited awareness of existing policies, pointing to communication and implementation gaps. Importantly, the findings suggest that GenAI does not eliminate the need for expertise but instead shifts value creation within the SDLC. As routine coding activities become increasingly automated, higher-order compe- tencies, such as requirements precision, architectural reasoning, critical validation, and governance oversight, gain importance. Experience level moderates perception: early-career professionals report particularly strong perceived learning benefits, whereas experienced practitioners tend to frame AI primarily as an efficiency- enhancing tool. This dynamic underscores the importance of structured training and human-in-the-loop mechanisms to prevent overreliance and ensure sustainable skill development. Several limitations warrant consideration. The survey sample is skewed toward younger, developer-centric respondents primarily from the DACH region. Moreover, the reliance on self-reported productivity estimates introduces potential response biases, as perceived and objective productivity gains may diverge. In summary, GenAI is not merely an incremental productivity tool but a transformative force reshaping workflows, role profiles, and governance structures within software development. However, the magnitude and durability of its benefits depend on deliberate organizational integration, transparent policy frameworks, and continued human accountability. Promising future research avenues lie in examining longitudinal effects on skill development, architectural quality, and organizational performance. 9 The State of Generative AI in Software Development: Insights from Literature and a Developer Survey References Michael Chui, Roger Roberts, Lareina Yee, Eric Hazan, Alex Singla, Kate Smaje, Alex Sukharevsky, and Rodney Zemmel. The economic potential of generative AI: The next productivity frontier. Technical report, McKinsey, June 2023. URLhttps://w.mckinsey.com/capabilities/tech-and-ai/our-insights/ the-economic-potential-of-generative-ai-the-next-productivity-frontier. Ian Goodfellow, Yoshua Bengio, and Aaron Courville. Deep Learning. Adaptive Computation and Machine Learning series. MIT Press, Cambridge, MA, USA, November 2016. ISBN 978-0-262-03561-3. URL https://mitpress.mit.edu/9780262035613/deep-learning/. Stefan Feuerriegel, Jochen Hartmann, Christian Janiesch, and Patrick Zschech. Generative AI. Business & Information Systems Engineering, 66:111â126, 2024. ISSN 2363-7005. doi:10.1007/s12599-023-00834-7. Fred Davis. Perceived Usefulness, Perceived Ease of Use, and User Acceptance of Information Technology. Management Information Systems Quarterly, 13(3), December 1989. ISSN ISSN 0276-7783/ISSN 2162-9730. URL https://aisel.aisnet.org/misq/vol13/iss3/6. Viswanath Venkatesh and Fred D. Davis. A Theoretical Extension of the Technology Acceptance Model: Four Longitudinal Field Studies. Management Science, 46(2):186â204, February 2000. ISSN 0025-1909, 1526-5501. doi:10.1287/mnsc.46.2.186.11926. URLhttps://pubsonline.informs.org/doi/10.1287/ mnsc.46.2.186.11926. Jane Webster and Richard T. Watson. Analyzing the Past to Prepare for the Future: Writing a Literature Review. MIS Quarterly, 26(2):xiiiâxxiii, 2002. URL http://w.jstor.org/stable/4132319. Chitu Okoli. A guide to conducting a standalone systematic literature review. Communications of the Association for Information Systems, 37, 2015. Guy ParĂ©, Marie-Claude Trudel, Mirou Jaana, and Spyros Kitsiou. Synthesizing information systems knowledge: A typology of literature reviews. Information & Management, 52(2):183â199, March 2015. ISSN 03787206. doi:10.1016/j.im.2014.08.008. URLhttps://linkinghub.elsevier.com/retrieve/pii/ S0378720614001116. Yair Levy and Timothy J. Ellis. A Systems Approach to Conduct an Effective Literature Review in Support of Information Systems Research. Informing Science: The International Journal of an Emerging Transdiscipline, 9:181â212, 2006. ISSN 1547-9684, 1521-4672. doi:10.28945/479. URLhttps://w. informingscience.org/Publications/479. Kim Fröhnel, Gabriel Benedikt Thomas Adams, and RĂŒdiger Zarnekow. Facing the Multifaceted Impact of Fake Reviews: A Comprehensive Literature Review. Wirtschaftsinformatik 2024 Proceedings, January 2024. URL https://aisel.aisnet.org/wi2024/39. Agnia Sergeyuk, Yaroslav Golubev, Timofey Bryksin, and Iftekhar Ahmed. Using AI-based coding assistants in practice: State of affairs, perceptions, and ways forward. Information and Software Technology, 178:107610, February 2025. ISSN 0950-5849. doi:10.1016/j.infsof.2024.107610. URLhttps://w.sciencedirect. com/science/article/pii/S0950584924002155. Antonio Collante, Samuel Abedu, SayedHassan Khatoonabadi, Ahmad Abdellatif, Ebube Alor, and Emad Shihab. The Impact of Large Language Models (LLMs) on Code Review Process, August 2025. URL http://arxiv.org/abs/2508.11034. arXiv:2508.11034 [cs]. Erik Brynjolfsson, Danielle Li, and Lindsey R Raymond. Generative AI at work. 2023. URLhttps: //w.nber.org/system/files/working_papers/w31161/w31161.pdf. Fabrizio DellâAcqua, Edward McFowland, Ethan R. Mollick, Hila Lifshitz-Assaf, Katherine Kellogg, Saran Rajendran, Lisa Krayer, François Candelon, and Karim R. Lakhani. Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality. SSRN Electronic Journal, 2023. ISSN 1556-5068. doi:10.2139/ssrn.4573321. URLhttps: //w.ssrn.com/abstract=4573321. Kevin KB Ng, Liyana Fauzi, Leon Leow, and Jaren Ng. Harnessing the Potential of Gen-AI Coding Assistants in Public Sector Software Development, September 2024. URLhttp://arxiv.org/abs/2409.17434. arXiv:2409.17434 [cs]. Md Istiak Hossain Shihab, Christopher Hundhausen, Ahsun Tariq, Summit Haque, Yunhan Qiao, and Brian Wise Mulanda. The Effects of GitHub Copilot on Computing Studentsâ Programming Effectiveness, Efficiency, and Processes in Brownfield Coding Tasks. In Proceedings of the 2025 ACM Conference on International Computing Education Research V.1, pages 407â420, Charlottesville USA, August 2025. 10 The State of Generative AI in Software Development: Insights from Literature and a Developer Survey ACM. ISBN 979-8-4007-1340-8. doi:10.1145/3702652.3744219. URLhttps://dl.acm.org/doi/10.1145/ 3702652.3744219. Devansh Yadav and Shouvick Mondal. Evaluating Pre-trained Large Language Models on Zero Shot Prompts for Parallelization of Source Code, January 2025. URL https://osf.io/jmevx_v1. Burak YetiĆtiren, IĆık Ăzsoy, Miray Ayerdem, and Eray TĂŒzĂŒn. Evaluating the Code Quality of AI-Assisted Code Generation Tools: An Empirical Study on GitHub Copilot, Amazon CodeWhisperer, and ChatGPT, October 2023. URL http://arxiv.org/abs/2304.10778. arXiv:2304.10778 [cs]. Mohammad Atif, Kriti Chopra, Ozgur Kilic, Tianle Wang, Zhihua Dong, Charles Leggett, Meifeng Lin, Paolo Calafiura, and Salman Habib. CelloAI: Leveraging Large Language Models for HPC Software Development in High Energy Physics, August 2025. URL http://arxiv.org/abs/2508.16713. arXiv:2508.16713 [cs]. Begum Karaci Deniz, Chandra Gnanasambandam, Martin Harrysson, Alharith Hussin, and Shivam Srivastava.Unleash developer productivity with generative AI.Technical report, McK- insey, June 2023. URLhttps://w.mckinsey.com/capabilities/tech-and-ai/our-insights/ unleashing-developer-productivity-with-generative-ai. Chandra Gnanasambandam, Martin Harrysson, and Rikki Singh. How an AI-enabled software product development life cycle will fuel innovation. Technical report, McKinsey, October 2025. URLhttps: //w.mckinsey.com/industries/technology-media-and-telecommunications/our-insights/ how-an-ai-enabled-software-product-development-life-cycle-will-fuel-innovation. AndrĂ© Barcaui and AndrĂ© Monat. Who is better in project planning?Generative artificial intelligence or project managers? Project Leadership and Society, 4:100101, December 2023. ISSN 2666- 7215. doi:10.1016/j.plas.2023.100101. URLhttps://w.sciencedirect.com/science/article/pii/ S2666721523000224. Laurie Hughes, Reza Kiani Mavi, Masoud Aghajani, Keith Fitzpatrick, Senali Madugoda Gunaratnege, Seyed Ashkan Hosseini Shekarabi, Richard Hughes, Ahmad Khanfar, Ahdieh Khatavakhotan, Neda Kiani Mavi, Keyao Li, Moataz Mahmoud, Tegwen Malik, Sashah Mutasa, Farzaneh Nafar, Ross Yates, Rasha Alahmad, Il Jeon, and Yogesh K. Dwivedi. Impact of artificial intelligence on project management (PM): Multi-expert perspectives on advancing knowledge and driving innovation toward PM2030. Journal of Innovation & Knowledge, 10(5):100772, September 2025. ISSN 2444-569X. doi:10.1016/j.jik.2025.100772. URL https://w.sciencedirect.com/science/article/pii/S2444569X25001179. Laksh Maggoo, Charles De Pommerol, Ashik Ardeshna, Emmanuel Amiot, Laurent Bensoussan, Jad Haddad, Lindy Kouyoumji, Ivan Palencia, and Benjamin Tubiana. Next Frontier in Transform- ing Telecom: Practical Benefits of Agentic AI and Digital Agents. Technical report, Oliver Wyman, 2025. URLhttps://w.oliverwyman.com/content/dam/oliver-wyman/v2/publications/ 2025/march/next-frontier-in-transforming-telecom.pdf. Christoph Lechner, Nikolaus Lang, Siegfried Handschuh, Olivier Bouffault, and Julian Cooper. Can GenAI do your next strategy task? Not yet. California Management Review, January 2025. URLhttps: //cmr.berkeley.edu/2024/09/can-genai-do-your-next-strategy-task-not-yet/. Phillip Wagner and Reinhard Wagner. The Evolution of Technology in Artificial Intelligence and Its Impact on Project Management. In D. Jude Hemanth, Utku Kose, Bogdan Patrut, and Mevlut Ersoy, editors, Innovative Methods in Computer Science and Computational Applications in the Era of Industry 5.0, pages 268â293, Cham, 2024. Springer Nature Switzerland. ISBN 978-3-031-56322-5. doi:10.1007/978-3-031-56322- 5_19. Vartenie Aramali, Namho Cho, Falguni Pande, M. K. S. Al-Mhdawi, Udechukwu Ojiako, and Abroon Qazi. Generative AI in project management: Impacts on corporate values, employee perceptions, and organizational practices. Project Leadership and Society, 6:100191, December 2025. ISSN 2666- 7215. doi:10.1016/j.plas.2025.100191. URLhttps://w.sciencedirect.com/science/article/pii/ S266672152500016X. Orges Cico, Betim Cico, and Andja Cico. AI-assisted Software Engineering: a tertiary study. In 2023 12th Mediterranean Conference on Embedded Computing (MECO), pages 1â6, June 2023. doi:10.1109/MECO58584.2023.10154972. URLhttps://ieeexplore.ieee.org/document/10154972. ISSN: 2637-9511. Gustavo Pinto, Cleidson de Souza, Thayssa Rocha, Igor Steinmacher, Alberto de Souza, and Edward Monteiro. Developer Experiences with a Contextualized AI Coding Assistant: Usability, Expectations, and Outcomes, November 2023. URL http://arxiv.org/abs/2311.18452. arXiv:2311.18452 [cs]. 11 The State of Generative AI in Software Development: Insights from Literature and a Developer Survey Jialiang Wei, Anne-Lise Courbis, Thomas Lambolais, GĂ©rard Dray, and Walid Maalej. On AI-Inspired User Interface Design. IEEE Software, 42(3):50â58, May 2025. ISSN 1937-4194. doi:10.1109/MS.2025.3536838. URL https://ieeexplore.ieee.org/document/10886968. Pierre-Yves Glever, Arnaud Balssa, Stephane Girard, Thilo Hermann, Serge Baccou, and Lorna Neville. Redefining software engineering: The impact of generative AI. Techni- cal report, Capgemini, 2024. URLhttps://w.capgemini.com/insights/research-library/ redefining-software-engineering-the-impact-of-generative-ai/. Jingzhi Gong, Vardan Voskanyan, Paul Brookes, Fan Wu, Wei Jie, Jie Xu, Rafail Giavrimis, Mike Basios, Leslie Kanthan, and Zheng Wang. Language Models for Code Optimization: Survey, Challenges and Future Directions, January 2025. URL http://arxiv.org/abs/2501.01277. arXiv:2501.01277 [cs]. Vikram Nitin. Using AI to Automate the Modernization of Legacy Software Applications. In Proceedings of the 39th IEEE/ACM International Conference on Automated Software Engineering, ASE â24, pages 2514â2517, New York, NY, USA, October 2024. Association for Computing Machinery. ISBN 979-8-4007-1248-7. doi:10.1145/3691620.3695610. URL https://doi.org/10.1145/3691620.3695610. Bingyang Wei. Requirements are All You Need: From Requirements to Code with LLMs. In 2024 IEEE 32nd International Requirements Engineering Conference (RE), pages 416â422, June 2024. doi:10.1109/RE59067.2024.00049. URLhttps://ieeexplore.ieee.org/abstract/document/10628487. ISSN: 2332-6441. Savvas Petridis, Michael Terry, and Carrie J Cai. PromptInfuser: How Tightly Coupling AI and UI Design Impacts Designersâ Workflows. In Proceedings of the 2024 ACM Designing Interactive Systems Conference, DIS â24, pages 743â756, New York, NY, USA, July 2024. Association for Computing Machinery. ISBN 979-8- 4007-0583-0. doi:10.1145/3643834.3661613. URL https://dl.acm.org/doi/10.1145/3643834.3661613. Dylan Porter, Joanna F. DeFranco, and Phil Laplante. Requirements Specification Automated Quality Analysis: Past, Present, and Future. Computer, 58(1):101â104, January 2025. ISSN 0018-9162, 1558-0814. doi:10.1109/MC.2024.3480629. URL https://ieeexplore.ieee.org/document/10834143/. Sida Peng, Eirini Kalliamvakou, Peter Cihon, and Mert Demirer. The Impact of AI on Developer Pro- ductivity: Evidence from GitHub Copilot, February 2023. URLhttp://arxiv.org/abs/2302.06590. arXiv:2302.06590 [cs]. Illia Solohubov, Artur Moroz, Mariia Yu Tiahunova, Halyna H Kyrychek, and Stepan Skrupsky. Accelerating software development with AI: exploring the impact of ChatGPT and GitHub Copilot. 2023. URL https://ceur-ws.org/Vol-3679/paper17.pdf. Dieter Gerdemann, Sven Garrels, Sophie Glusac, and Konstantin Bykhanov.Why software + generative AI could take value creation to a whole new level | Kear- ney, April 2024.URLhttps://w.kearney.com/service/digital-analytics/article/ why-software-generative-ai-could-take-value-creation-to-a-whole-new-level. Alassane Ndiaye, Xhorxhina Taraj, Bostjan Robeznik, Gerd Herzog, Christian Gerber, and Hendrik Purwins. Generative AI in Software Engineering: Transforming the Software Development Process. Technical report, Accenture & German Research Center for Artificial Intelligence, October 2025. Ming Zhong, Xin Sun, Fang Lv, Lulin Wang, Hongna Geng, Lei Qiu, Huimin Cui, and Xiaobing Feng. BePilot: An AI Programming Assistant for Compiler Backend Development. ACM Trans. Softw. Eng. Methodol., August 2025. ISSN 1049-331X. doi:10.1145/3764585. URLhttps://dl.acm.org/doi/10.1145/3764585. Just Accepted. Kevin Zheyuan Cui, Mert Demirer, Sonia Jaffe, Leon Musolff, Sida Peng, and Tobias Salz. The Productivity Effects of Generative AI: Evidence from a Field Experiment with GitHub Copilot. An MIT Exploration of Generative AI, March 2024. doi:10.21428/e4baedd9.3ad85f1c. URLhttps://mit-genai.pubpub.org/ pub/v5iixksv/release/2. Max Struever, Matthew Kropp, and Julie Bedard. From Dev Speed to Business Impact: The Case for AI- Assisted Coding and Generative Engineering, July 2025. URLhttps://w.bcg.com/x/the-multiplier/ ai-assisted-coding-generative-engineering. Christof Ebert and Panos Louridas. Generative AI for Software Practitioners. IEEE Software, 40(4):30â38, July 2023. ISSN 0740-7459, 1937-4194. doi:10.1109/MS.2023.3265877. URLhttps://ieeexplore.ieee. org/document/10176168/. Ao Ding, Gaolei Li, Xiaoyu Yi, Xi Lin, Jianhua Li, and Chaofeng Zhang. Generative AI for Software Security Analysis: Fundamentals, Applications, and Challenges. IEEE Software, 41(6):46â54, November 2024. ISSN 12 The State of Generative AI in Software Development: Insights from Literature and a Developer Survey 1937-4194. doi:10.1109/MS.2024.3416036. URLhttps://ieeexplore.ieee.org/abstract/document/ 10634314. Hammond Pearce, Baleegh Ahmad, Benjamin Tan, Brendan Dolan-Gavitt, and Ramesh Karri. Asleep at the Keyboard? Assessing the Security of GitHub Copilotâs Code Contributions. Commun. ACM, 68(2):96â105, January 2025. ISSN 0001-0782. doi:10.1145/3610721. URLhttps://dl.acm.org/doi/10.1145/3610721. Gopinath Kathiresan. Automated Test Case Generation with AI: A Novel Framework for Improving Software Quality and Coverage. World Journal of Advanced Research and Reviews, 23(2):2880â 2889, 2024. ISSN 2581-9615. doi:10.30574/wjarr.2024.23.2.2463. URLhttps://wjarr.com/content/ automated-test-case-generation-ai-novel-framework-improving-software-quality-and-coverage. Last Modified: 2025-03-24T17:45+05:30. Werner Geyer, Jessica He, Daita Sarkar, Michelle Brachman, Chris Hammond, Jennifer Heins, Zahra Ashktorab, Carlos Rosemberg, and Charlie Hill. A Case Study Investigating the Role of Generative AI in Quality Evaluations of Epics in Agile Software Development, May 2025. URLhttp://arxiv.org/abs/ 2505.07664. arXiv:2505.07664 [cs]. Sai Zhang, Zhenchang Xing, Ronghui Guo, Fangzhou Xu, Lei Chen, Zhaoyuan Zhang, Xiaowang Zhang, Zhiyong Feng, and Zhiqiang Zhuang. Empowering Agile-Based Generative Software Development through Human-AI Teamwork, November 2024. URLhttp://arxiv.org/abs/2407.15568. arXiv:2407.15568 [cs]. Rasmus Ulfsnes, Nils Brede Moe, Viktoria Stray, and Marianne Skarpen. Transforming Software Development with Generative AI: Empirical Insights on Collaboration and Workflow, February 2024. URLhttp: //arxiv.org/abs/2405.01543. arXiv:2405.01543 [cs]. Anas Bahi, Jihane Gharib, and Youssef Gahi. Integrating Generative AI for Advancing Agile Software Development and Mitigating Project Management Challenges. International Journal of Advanced Computer Science and Applications, 15(3), 2024. ISSN 21565570, 2158107X. doi:10.14569/IJACSA.2024.0150306. URL http://thesai.org/Publications/ViewPaper?Volume=15&Issue=3&Code=IJACSA&SerialNo=6. Philipp Diebold. From Backlogs to Bots: Generative AIâs Impact on Agile Role Evolution. J. Softw. Evol. Process, 37(1), January 2025. ISSN 2047-7473. doi:10.1002/smr.2740. URLhttps://doi.org/10.1002/ smr.2740. Bilal Nasir and Ijaz Hussain. Automating Agile Workflows: The Role of Multi-Agent LLMs in Modern Software Engineering. Technical report, December 2024. Albert Ziegler, Eirini Kalliamvakou, Shawn Simister, Ganesh Sittampalam, Alice Li, Andrew Rice, Devon Rifkin, and Edward Aftandilian. Productivity Assessment of Neural Code Completion, May 2022. URL http://arxiv.org/abs/2205.06537. arXiv:2205.06537 [cs]. Mark Chen, Jerry Tworek, Heewoo Jun, Qiming Yuan, Henrique Ponde de Oliveira Pinto, Jared Kaplan, Harri Edwards, Yuri Burda, Nicholas Joseph, Greg Brockman, Alex Ray, Raul Puri, Gretchen Krueger, Michael Petrov, Heidy Khlaaf, Girish Sastry, Pamela Mishkin, Brooke Chan, Scott Gray, Nick Ryder, Mikhail Pavlov, Alethea Power, Lukasz Kaiser, Mohammad Bavarian, Clemens Winter, Philippe Tillet, Felipe Petroski Such, Dave Cummings, Matthias Plappert, Fotios Chantzis, Elizabeth Barnes, Ariel Herbert- Voss, William Hebgen Guss, Alex Nichol, Alex Paino, Nikolas Tezak, Jie Tang, Igor Babuschkin, Suchir Balaji, Shantanu Jain, William Saunders, Christopher Hesse, Andrew N. Carr, Jan Leike, Josh Achiam, Vedant Misra, Evan Morikawa, Alec Radford, Matthew Knight, Miles Brundage, Mira Murati, Katie Mayer, Peter Welinder, Bob McGrew, Dario Amodei, Sam McCandlish, Ilya Sutskever, and Wojciech Zaremba. Evaluating Large Language Models Trained on Code, July 2021. URLhttp://arxiv.org/abs/2107.03374. arXiv:2107.03374 [cs]. Viktoria Stray, Elias Goldmann BrandtzĂŠg, Viggo Tellefsen Wivestad, Astri Barbala, and Nils Brede Moe. Developer Productivity With and Without GitHub Copilot: A Longitudinal Mixed-Methods Case Study, September 2025. URL http://arxiv.org/abs/2509.20353. arXiv:2509.20353 [cs]. Yue Liu, Yanjie Zhao, Yunbo Lyu, Ting Zhang, Haoyu Wang, and David Lo. "Your AI, My Shell": Demystifying Prompt Injection Attacks on Agentic AI Coding Editors, September 2025. URLhttp: //arxiv.org/abs/2509.22040. arXiv:2509.22040 [cs]. Domenico Cotroneo, Cristina Improta, Pietro Liguori, and Roberto Natella. Vulnerabilities in AI Code Generators: Exploring Targeted Data Poisoning Attacks. In Proceedings of the 32nd IEEE/ACM Inter- national Conference on Program Comprehension, ICPC â24, pages 280â292, New York, NY, USA, June 2024. Association for Computing Machinery. ISBN 979-8-4007-0586-1. doi:10.1145/3643916.3644416. URL https://dl.acm.org/doi/10.1145/3643916.3644416. 13 The State of Generative AI in Software Development: Insights from Literature and a Developer Survey Kieran Norton, Tim Li, Tim Davis, Emily Mossburg, Diana Kearns-Manolatos, Saurabh Ban- sode, Chris Thomas, and Alfons BuxĂł Ferrer. How can tech leaders manage emerging gener- ative AI risks today while keeping the future in mind?Technical report, Deloitte, Febru- ary 2025. URLhttps://w.deloitte.com/us/en/insights/topics/digital-transformation/ four-emerging-categories-of-gen-ai-risks.html. Zhiguo Ding, Songyang Wu, Yilin Zhang, Tao Yang, Yingna Li, Zhuhua Li, Hang Shao, and Yicong Shi. CodingCare: AI Code Generation Security Framework for Common Vulnerability Mitigation. In Proceedings of the 2025 4th International Conference on Intelligent Systems, Communications and Computer Networks, ISCCN â25, pages 307â312, New York, NY, USA, June 2025. Association for Computing Machinery. ISBN 979-8-4007-1520-4. doi:10.1145/3732945.3732990. URLhttps://dl.acm.org/doi/10.1145/3732945. 3732990. Edward Anderson, Geoffrey Parker, and Burcu Tan. The Hidden Costs of Coding With Generative AI. MIT Sloan Management Review, 67(1), August 2025. URLhttps://sloanreview.mit.edu/article/ the-hidden-costs-of-coding-with-generative-ai/. Sergio Moreschini, Elvira-Maria Arvanitou, Elisavet-Persefoni Kanidou, Nikolaos Nikolaidis, Ruoyu Su, Apostolos Ampatzoglou, Alexander Chatzigeorgiou, and Valentina Lenarduzzi. The Evolution of Technical Debt from DevOps to Generative AI: A multivocal literature review. Journal of Systems and Software, 231: 112599, January 2026. ISSN 0164-1212. doi:10.1016/j.jss.2025.112599. URLhttps://w.sciencedirect. com/science/article/pii/S0164121225002687. Beatriz Cabrero-Daniel, Tomas Herda, Victoria Pichler, and Martin Eder. Exploring Human-AI Collaboration in Agile: Customised LLM Meeting Assistants, April 2024. URLhttp://arxiv.org/abs/2404.14871. arXiv:2404.14871 [cs]. Pranay Ahlawat, Julie Bedard, Sankalp Damani, Ben Feldman, Elizabeth Lucero, Abhishek Samdaria, and Geoffrey Sipperly. The Art of Scaling GenAI in Software. Technical report, Boston Consulting Group, May 2024. URL https://w.bcg.com/publications/2024/the-art-of-scaling-genai-in-software. Nicola Döring. Forschungsmethoden und Evaluation in den Sozial- und Humanwissenschaften. Springer, Berlin, Heidelberg, 2023. ISBN 978-3-662-64761-5 978-3-662-64762-2. doi:10.1007/978-3-662-64762-2. URL https://link.springer.com/10.1007/978-3-662-64762-2. 14