Eray Tüzün

Vahid Haratian, Atakan Akar, Berk Çakar et al.

Automated test generation has a substantial body of work, yet most studies focus on generating tests for complete software units, such as classes, and rely on metrics such as code coverage for assessment. In contrast, modern software development primarily evolves through small, targeted changes introduced in pull requests (PRs). Despite this, the crucial task of generating tests specifically for these PRs has been overlooked, and the performance of state-of-the-art tools for this purpose remains unknown. This study evaluates two distinct approaches for PR-aware test generation: EvoSuite, a leading search-based tool, and GPT-4o, one of the widely used large language models (LLMs). To measure their effectiveness at validating PR-specific changes, we assess their ability to generate fail-to-pass (F2P) test cases, meaning tests that fail on the code before the change and pass on the code after the change. Our evaluation shows that EvoSuite outperformed GPT-4o, producing at least one F2P test for a significantly higher percentage of PRs (36 percent vs. 13 percent). The performance of GPT-4o was significantly hampered by a high rate of compilation errors (63 percent), whereas only 2 percent of EvoSuite's generated tests failed to run. Despite EvoSuite's relative success, our findings indicate that both tools are largely ineffective for this task, as they failed to generate any meaningful change-capturing tests for the large majority of the PRs (64 percent). Although both generators could not achieve a high F2P ratio in our evaluation, and EvoSuite outperformed GPT-4o, we believe that agentic code generation methods may have significant potential for this task. Ultimately, our work highlights a critical gap in tooling and calls for the development of high-performance test generators tailored to the incremental nature of modern software development.

Hüseyin Özgür Kamalı, Erdem Tuna, Vahid Haratian et al.

Code review has evolved for decades, from informal peer checking to today's pull request (PR) workflows, yet it remains a largely manual, uneven, and cognitively demanding process. The rise of Artificial Intelligence (AI) coding assistants has intensified this challenge: while these tools increase code production velocity, they also expand the volume of code requiring review, turning code review into a growing bottleneck. Current AI support remains fragmented, with tools focusing on isolated tasks such as reviewer recommendation, PR description generation, or comment suggestion rather than the end-to-end PR review workflow. In this paper, we review the historical evolution of code review practices and examine the shift driven by large language models (LLMs) and agentic AI systems. We then present a vision for an AI-powered code review workflow combining specialized agents with human-controlled quality gates. Our framework spans five stages: PR Creation, PR Augmentation, Reviewer Selection, AI-Assisted Code Review, and PR Retrospective, with humans retained at key decision points to preserve judgment, accountability, and team-level understanding. We identify major open challenges for responsible adoption, including reliability, bias, privacy, automation bias, transparency, and evaluation, and offer a research agenda for more effective human-AI collaboration in software engineering.

Eray Yapağcı, Yavuz Alp Sencer Öztürk, Eray Tüzün

Reproducing game bugs, particularly crash bugs in continuously evolving games like Minecraft, is a notoriously manual, time-consuming, and challenging process to automate; insights from a key decision maker from Minecraft we interviewed confirm this, highlighting that a substantial portion of crash reports necessitate manual scenario reconstruction. Despite the success of LLM-driven bug reproduction in other software domains, games, with their complex interactive environments, remain largely unaddressed. This paper introduces BugCraft, a novel end-to-end framework designed to automate the reproduction of crash bugs in Minecraft directly from user-submitted bug reports, addressing the critical gap in automated game bug reproduction. BugCraft employs a two-stage approach: first, a Step Synthesizer leverages LLMs and Minecraft Wiki knowledge to transform bug reports into high-quality, structured steps to reproduce (S2R). Second, an Action Model, powered by a vision-based LLM agent and a custom macro API, executes these S2R steps within Minecraft to trigger the reported crash. To facilitate evaluation, we introduce BugCraft-Bench, a curated dataset of Minecraft crash bug reports. On BugCraft-Bench, our framework end-to-end reproduced 34.9% of crash bugs with GPT-4.1, outperforming baseline computer-use models by 37%. BugCraft demonstrates the feasibility of automated reproduction of crash bugs in complex game environments using LLMs, opening promising avenues for game testing and development. Finally, we make our code open at https://bugcraft2025.github.io

Emre Furkan Akyol, Mehmet Dedeler, Eray Tüzün

Bug tracking systems play a crucial role in software maintenance, yet developers frequently struggle with low-quality user-submitted reports that omit essential details such as Steps to Reproduce (S2R), Observed Behavior (OB), and Expected Behavior (EB). We propose ImproBR, an LLM-based pipeline that automatically detects and improves bug reports by addressing missing, incomplete, and ambiguous S2R, OB, and EB sections. ImproBR employs a hybrid detector combining fine-tuned DistilBERT, heuristic analysis, and an LLM analyzer, guided by GPT-4o mini with section-specific few-shot prompts and a Retrieval-Augmented Generation (RAG) pipeline grounded in Minecraft Wiki domain knowledge. Evaluated on Mojira, ImproBR improved structural completeness from 7.9% to 96.4%, more than doubled the proportion of executable S2R from 28.8% to 67.6%, and raised fully reproducible bug reports from 1 to 13 across 139 challenging real-world reports.

Robin Gröpler, Steffen Klepke, Jack Johns et al.

Generative AI (GenAI) has recently emerged as a groundbreaking force in Software Engineering, capable of generating code, identifying bugs, recommending fixes, and supporting quality assurance. While its use in coding tasks shows considerable promise, applying GenAI across the entire Software Development Life Cycle (SDLC) has not yet been fully explored. Critical uncertainties in areas such as reliability, accountability, security, and data privacy demand deeper investigation and coordinated action. The GENIUS project, comprising over 30 European industrial and academic partners, aims to address these challenges by advancing AI integration across all SDLC phases. It focuses on GenAI's potential, the development of innovative tools, and emerging research challenges, actively shaping the future of software engineering. This vision paper presents a shared perspective on the future of GenAI-driven software engineering, grounded in cross-sector dialogue as well as experiences and findings within the GENIUS consortium. The paper explores four central elements: (1) a structured overview of current challenges in GenAI adoption across the SDLC; (2) a forward-looking vision outlining key technological and methodological advances expected over the next five years; (3) anticipated shifts in the roles and required skill sets of software professionals; and (4) the contribution of GENIUS in realising this transformation through practical tools and industrial validation. This paper focuses on aligning technical innovation with business relevance. It aims to inform both research agendas and industrial strategies, providing a foundation for reliable, scalable, and industry-ready GenAI solutions for software engineering teams.

Utku Boran Torun, Veli Karakaya, Ali Babar et al.

Large Language Models (LLMs) are increasingly embedded in software engineering (SE) tools, powering applications such as code generation, automated code review, and bug triage. As these LLM-based AI for Software Engineering (AI4SE) systems transition from experimental prototypes to widely deployed tools, the question of what it means to evaluate their behavior reliably has become both critical and unanswered. Unlike traditional SE or machine learning systems, LLM-based tools often produce open-ended, natural language outputs, admit multiple valid answers, and exhibit non-deterministic behavior across runs. These characteristics fundamentally challenge long-standing evaluation assumptions such as the existence of a single ground truth, deterministic outputs, and objective correctness. In this paper, we examine LLM evaluation as a general, task-dependent concept through the lens of SE tasks. We discuss why reliable evaluation is essential for trust, adoption, and meaningful assessment of LLM-based tools, summarize the current state of evaluation practices, and highlight their limitations in realistic AI4SE settings. We then identify key challenges facing current approaches, including the absence of stable ground truth, subjectivity and multi-dimensional quality, evaluation instability due to non-determinism, limitations of automated and model-based evaluation, and fragmentation of evaluation practices. Finally, we outline future directions aimed at advancing LLM evaluation toward more robust, scalable, and trustworthy methodologies, to stimulate discussion on principled evaluation practices that can keep pace with the growing role of LLMs in SE.

Semih Çağlar, Şükrü Eren Gökırmak, Eray Tüzün

Context: Code reviews are essential for maintaining software quality, yet many human review comments suffer from issues such as redundancy, vagueness, or lack of constructiveness. These types of comments may slow down feedback and obscure important insights. Prior work on code review comments mostly explore the detection and categorization of useful comments, while fine-grained categorization of comment issues remains underexplored. Objective: This work aims to design and evaluate an automated system for classifying code review comments according to specific categories of issues. Methodology: We introduced a nine-label taxonomy for code review comments, covering six review comment smells and three common useful intents, and manually labeled 448 comments from a publicly available dataset. We benchmarked zero-shot and one-shot single-label classification over each comment and its associated unified diff hunk, comparing GPT-5-mini, LLaMA-3.3, and DeepSeek-R1. We reported macro-F1 as the primary metric. Results: Zero-shot performance was moderate under class imbalance (macro-F1 0.360 to 0.374). One-shot exemplar conditioning had model-dependent effects: GPT-5-mini and DeepSeek-R1 macro-F1 scores improved, however LLaMA-3.3 suffered a slight decrease. Exemplars most consistently helped intent-boundary labels, whereas classification of evidence-sensitive labels remain challenging. Conclusion: Our results indicate that comment--diff evidence is sufficient for some labels but limited for evidence-sensitive smells. Future work includes adding thread context, improving intent-preserving rewrites, and validating robustness across platforms.

Veli Karakaya, Utku Boran Torun, Baykal Mehmet Uçar et al.

Automated code review (ACR) bots are increasingly used in industrial software development to assist developers during pull request (PR) review. As adoption grows, a key challenge is how to evaluate the usefulness of bot-generated comments reliably and at scale. In practice, such evaluation often relies on developer actions and annotations that are shaped by contextual and organizational factors, complicating their use as objective ground truth. We examine the feasibility and limitations of automating the evaluation of LLM-powered ACR bots in an industrial setting. We analyze an industrial dataset from Beko comprising 2,604 bot-generated PR comments, each labeled by software engineers as fixed/wontFix. Two automated evaluation approaches, G-Eval and an LLM-as-a-Judge pipeline, are applied using both binary decisions and a 0-4 Likert-scale formulation, enabling a controlled comparison against developer-provided labels. Across Gemini-2.5-pro, GPT-4.1-mini, and GPT-5.2, both evaluation strategies achieve only moderate alignment with human labels. Agreement ratios range from approximately 0.44 to 0.62, with noticeable variation across models and between binary and Likert-scale formulations, indicating sensitivity to both model choice and evaluation design. Our findings highlight practical limitations in fully automating the evaluation of ACR bot comments in industrial contexts. Developer actions such as resolving or ignoring comments reflect not only comment quality, but also contextual constraints, prioritization decisions, and workflow dynamics that are difficult to capture through static artifacts. Insights from a follow-up interview with a software engineering director further corroborate that developer labeling behavior is strongly influenced by workflow pressures and organizational constraints, reinforcing the challenges of treating such signals as objective ground truth.

Utku Boran Torun, Mehmet Taha Demircan, Mahmut Furkan Gön et al.

Traditional bug tracking systems rely heavily on manual reporting, reproduction, triaging, and resolution, each carried out by different stakeholders such as end users, customer support, developers, and testers. This division of responsibilities requires significant coordination and widens the communication gap between non-technical users and technical teams, slowing the process from bug discovery to resolution. Moreover, current systems are highly asynchronous; users often wait hours or days for a first response, delaying fixes and contributing to frustration. This paper examines the evolution of bug tracking, from early paper-based reporting to today's web-based and SaaS platforms. Building on this trajectory, we propose an AI-powered bug tracking framework that augments existing tools with intelligent, large language model (LLM)-driven automation. Our framework addresses two main challenges: reducing time-to-fix and minimizing human overhead. Users report issues in natural language, while AI agents refine reports, attempt reproduction, and request missing details. Reports are then classified, invalid ones resolved through no-code fixes, and valid ones localized and assigned to developers. LLMs also generate candidate patches, with human oversight ensuring correctness. By integrating automation into each phase, our framework accelerates response times, improves collaboration, and strengthens software maintenance practices for a more efficient, user-centric future.

Elgun Jabrayilzade, Mikhail Evtikhiev, Eray Tüzün et al.

Bus factor is a metric that identifies how resilient is the project to the sudden engineer turnover. It states the minimal number of engineers that have to be hit by a bus for a project to be stalled. Even though the metric is often discussed in the community, few studies consider its general relevance. Moreover, the existing tools for bus factor estimation focus solely on the data from version control systems, even though there exists other channels for knowledge generation and distribution. With a survey of 269 engineers, we find that the bus factor is perceived as an important problem in collective development, and determine the highest impact channels of knowledge generation and distribution in software development teams. We also propose a multimodal bus factor estimation algorithm that uses data on code reviews and meetings together with the VCS data. We test the algorithm on 13 projects developed at JetBrains and compared its results to the results of the state-of-the-art tool by Avelino et al. against the ground truth collected in a survey of the engineers working on these projects. Our algorithm is slightly better in terms of both predicting the bus factor as well as key developers compared to the results of Avelino et al. Finally, we use the interviews and the surveys to derive a set of best practices to address the bus factor issue and proposals for the possible bus factor assessment tool.

Marco Kuhrmann, Paolo Tell, Regina Hebig et al.

Together with many success stories, promises such as the increase in production speed and the improvement in stakeholders' collaboration have contributed to making agile a transformation in the software industry in which many companies want to take part. However, driven either by a natural and expected evolution or by contextual factors that challenge the adoption of agile methods as prescribed by their creator(s), software processes in practice mutate into hybrids over time. Are these still agile? In this article, we investigate the question: what makes a software development method agile? We present an empirical study grounded in a large-scale international survey that aims to identify software development methods and practices that improve or tame agility. Based on 556 data points, we analyze the perceived degree of agility in the implementation of standard project disciplines and its relation to used development methods and practices. Our findings suggest that only a small number of participants operate their projects in a purely traditional or agile manner (under 15%). That said, most project disciplines and most practices show a clear trend towards increasing degrees of agility. Compared to the methods used to develop software, the selection of practices has a stronger effect on the degree of agility of a given discipline. Finally, there are no methods or practices that explicitly guarantee or prevent agility. We conclude that agility cannot be defined solely at the process level. Additional factors need to be taken into account when trying to implement or improve agility in a software company. Finally, we discuss the field of software process-related research in the light of our findings and present a roadmap for future research.

Jil Klünder, Regina Hebig, Paolo Tell et al.

Software development methods are usually not applied by the book. Companies are under pressure to continuously deploy software products that meet market needs and stakeholders' requests. To implement efficient and effective development processes, companies utilize multiple frameworks, methods and practices, and combine these into hybrid methods. A common combination contains a rich management framework to organize and steer projects complemented with a number of smaller practices providing the development teams with tools to complete their tasks. In this paper, based on 732 data points collected through an international survey, we study the software development process use in practice. Our results show that 76.8% of the companies implement hybrid methods. Company size as well as the strategy in devising and evolving hybrid methods affect the suitability of the chosen process to reach company or project goals. Our findings show that companies that combine planned improvement programs with process evolution can increase their process' suitability by up to 5%.

Vahid Garousi, Görkem Giray, Eray Tüzün et al.

According to different reports, many recent software engineering graduates often face difficulties when beginning their professional careers, due to misalignment of the skills learnt in their university education with what is needed in industry. To address that need, many studies have been conducted to align software engineering education with industry needs. To synthesize that body of knowledge, we present in this paper a systematic literature review (SLR) which summarizes the findings of 33 studies in this area. By doing a meta-analysis of all those studies and using data from 12 countries and over 4,000 data points, this study will enable educators and hiring managers to adapt their education / hiring efforts to best prepare the software engineering workforce.