Large language models have found their success by scaling up their capabilities to work in general settings. The same can unfortunately not be said for their interpretability methods. The current trend in mechanistic interpretability is to provide precise explanations of specific behaviors in controlled settings. These often do not generalize well into other settings, or are too resource intensive for larger studies. In this work we propose to study repeated behaviors in large language models by mining completion scenarios in Java code datasets, through exploiting the structured nature of source code. We then collect the attention patterns generated in the attention heads to demonstrate that they are scalable signals for global interpretability of model components. We show that vision models offer a promising direction for analyzing attention patterns at scale. To demonstrate this, we introduce the Attention Pattern – Masked Autoencoder (AP-MAE), a vision transformer-based model that efficiently reconstructs masked attention patterns. Experiments on StarCoder2 models (3B–15B) show that AP-MAE (i) reconstructs masked attention patterns with high accuracy, (ii) generalizes across unseen models with minimal degradation, (iii) reveals recurring patterns across a large number of inferences, (iv) predicts whether a generation will be correct without access to ground truth, with accuracies ranging from 55% to 70% depending on the task, and (v) enables targeted interventions that increase accuracy by 13.6% when applied selectively, but cause rapid collapse when applied excessively. These results establish attention patterns as a scalable signal for interpretability and demonstrate that AP-MAE provides a transferable foundation for both analysis and intervention in large language models. Beyond its standalone value, AP-MAE can also serve as a selection procedure to guide more fine-grained mechanistic approaches toward the most relevant components. We release code and models to support future work in large-scale interpretability.

Current in-IDE AI coding tools typically rely on time-consuming manual prompting and context management, whereas proactive alternatives that anticipate developer needs without explicit invocation remain underexplored. Understanding when humans are receptive to such proactive AI assistance during their daily work remains an open question in human-AI interaction research. We address this gap through a field study of proactive AI assistance in professional developer workflows. We present a five-day in-the-wild study with 15 developers who interacted with a proactive feature of an AI assistant integrated into a production-grade IDE that offers code quality suggestions based on in-IDE developer activity. We examined 229 AI interventions across 5,732 interaction points to understand how proactive suggestions are received across workflow stages, how developers experience them, and their perceived impact. Our findings reveal systematic patterns in human receptivity to proactive suggestions: interventions at workflow boundaries (e.g., post-commit) achieved 52% engagement rates, while mid-task interventions (e.g., on declined edit) were dismissed 62% of the time. Notably, well-timed proactive suggestions required significantly less interpretation time than reactive suggestions (45.4s versus 101.4s, W = 109.00, r = 0.533, p = 0.0016), indicating enhanced cognitive alignment. This study provides actionable implications for designing proactive coding assistants, including how to time interventions, align them with developer context, and strike a balance between AI agency and user control in production IDEs.

The integration of Artificial Intelligence (AI) into Integrated Development Environments (IDEs) is reshaping software development, fundamentally altering how developers interact with their tools. This shift marks the emergence of Human-AI Experience in Integrated Development Environment (in-IDE HAX), a field that explores the evolving dynamics of Human-Computer Interaction in AI-assisted coding environments. Despite rapid adoption, research on in-IDE HAX remains fragmented, which highlights the need for a unified overview of current practices, challenges, and opportunities. To provide a structured overview of existing research, we conduct a systematic literature review of 90 studies, summarizing current findings and outlining areas for further investigation. We organize key insights from reviewed studies into three aspects: Impact, Design, and Quality of AI-based systems inside IDEs. Impact findings show that AI-assisted coding enhances developer productivity but also introduces challenges, such as verification overhead and over-reliance. Design studies show that effective interfaces surface context, provide explanations and transparency of suggestion, and support user control. Quality studies document risks in correctness, maintainability, and security. For future research, priorities include productivity studies, design of assistance, and audit of AI-generated code. The agenda calls for larger and longer evaluations, stronger audit and verification assets, broader coverage across the software life cycle, and adaptive assistance under user control.

The recent rise in the popularity of large language models has spurred the development of extensive code datasets needed to train them. This has left limited code available for collection and use in the downstream investigation of specific behaviors, or evaluation of large language models without suffering from data contamination. To address this problem, we release The Heap, a large multilingual dataset covering 57 programming languages that has been deduplicated with respect to other open datasets of code, enabling researchers to conduct fair evaluations of large language models without significant data cleaning overhead.

Large language models have shown impressive performance in various domains, including code generation across diverse open-source domains. However, their applicability in proprietary industrial settings, where domain-specific constraints and code interdependencies are prevalent, remains largely unexplored. We present a case study conducted in collaboration with the leveling department at ASML to investigate the performance of LLMs in generating functional, maintainable code within a closed, highly specialized software environment.We developed an evaluation framework tailored to ASML’s proprietary codebase and introduced a new benchmark. Additionally, we proposed a new evaluation metric, build@k, to assess whether LLM-generated code successfully compiles and integrates within real industrial repositories. We investigate various prompting techniques, compare the performance of generic and code-specific LLMs, and examine the impact of model size on code generation capabilities, using both match-based and execution-based metrics. The findings reveal that prompting techniques and model size have a significant impact on output quality, with few-shot and chain-of-thought prompting yielding the highest build success rates. The difference in performance between the code-specific LLMs and generic LLMs was less pronounced and varied substantially across different model families.

Large Language Models are essential coding assistants, yet their training is predominantly English-centric. In this study, we evaluate the performance of code language models in non-English contexts, identifying challenges in their adoption and integration into multilingual workflows. We conduct an open-coding study to analyze errors in code comments generated by five state-of-the-art code models, CodeGemma, CodeLlama, CodeQwen1.5, GraniteCode, and StarCoder2 across five natural languages: Chinese, Dutch, English, Greek, and Polish. Our study yields a dataset of 12,500 labeled generations, which we publicly release. We then assess the reliability of standard metrics in capturing comment correctness across languages and evaluate their trustworthiness as judgment criteria. Through our open-coding investigation, we identified a taxonomy of 26 distinct error categories in model-generated code comments. They highlight variations in language cohesion, informativeness, and syntax adherence across different natural languages. Our analysis shows that, while these models frequently produce partially correct comments, modern neural metrics fail to reliably differentiate meaningful completions from random noise. Notably, the significant score overlap between expert-rated correct and incorrect comments calls into question the effectiveness of these metrics in assessing generated comments.

Benchmarks are essential for unified evaluation and reproducibility. The rapid rise of Artificial Intelligence for Software Engineering (AI4SE) has produced numerous benchmarks for tasks such as code generation and bug repair. However, this proliferation has led to major challenges: (1) fragmented knowledge across tasks, (2) difficulty in selecting contextually relevant benchmarks, (3) lack of standardization in benchmark creation, and (4) flaws that limit utility. Addressing these requires a dual approach: systematically mapping existing benchmarks for informed selection and defining unified guidelines for robust, adaptable benchmark development. We conduct a review of 247 studies, identifying 273 AI4SE benchmarks since 2014. We categorize them, analyze limitations, and expose gaps in current practices. Building on these insights, we introduce BenchScout, an extensible semantic search tool for locating suitable benchmarks. BenchScout employs automated clustering with contextual embeddings of benchmark-related studies, followed by dimensionality reduction. In a user study with 22 participants, BenchScout achieved usability, effectiveness, and intuitiveness scores of 4.5, 4.0, and 4.1 out of 5. To improve benchmarking standards, we propose BenchFrame, a unified approach to improve benchmark quality. Applying BenchFrame to HumanEval yielded HumanEvalNext, which features corrected errors, improved language conversion, higher test coverage, and greater difficulty. Evaluating 10 state-of-the-art code models on HumanEval, HumanEvalPlus, and HumanEvalNext revealed average pass-at-1 drops of 31.22% and 19.94%, respectively, underscoring the need for continuous benchmark refinement. We further examine BenchFrame's scalability through an agentic pipeline and confirm its generalizability on the MBPP dataset. Lastly, we publicly release the material of our review, user study, and the enhanced benchmark. ¹https://github.com/AISE-TUDelft/AI4SE-benchmarks

In the rapidly evolving world of software development, the surge in developers’ reliance on AI-driven tools has transformed Integrated Development Environments into powerhouses of advanced features. This transformation, while boosting developers’ productivity to unprecedented levels, comes with a catch: increased hardware demands for software development. Moreover, the significant economic and environmental toll of using these sophisticated models necessitates mechanisms that reduce unnecessary computational burdens. We propose HyperSeq - Hyper-Adaptive Representation for Predictive Sequencing of States - a novel, resource-efficient approach designed to model developers’ cognitive states. HyperSeq facilitates precise action sequencing and enables real-time learning of user behavior. Our preliminary results show how HyperSeq excels in forecasting action sequences and achieves remarkable prediction accuracies that go beyond 70%. Notably, the model’s online-learning capability allows it to substantially enhance its predictive accuracy in a majority of cases and increases its capability in forecasting next user actions with sufficient iterations for adaptation. Ultimately, our objective is to harness these predictions to refine and elevate the user experience dynamically within the IDE.

As GenAI becomes embedded in developer toolchains and practices, and routine code is increasingly generated, human creativity will be increasingly important for generating competitive advantage. This article uses the McLuhan tetrad alongside scenarios of how GenAI may disrupt software development more broadly, to identify potential impacts GenAI may have on creativity within software development. The impacts are discussed along with a future research agenda comprising five connected themes that consider how individual capabilities, team capabilities, the product, unintended consequences, and society can be affected.

Token-Level code completion is one of the most critical features in modern Integrated Development Environments (IDEs). It assists developers by suggesting relevant identifiers and APIs during coding. While completions are typically derived from static analysis, their usefulness depends heavily on how they are ranked, as correct predictions buried deep in the list are rarely seen by users. Most current systems rely on handcrafted heuristics or lightweight machine learning models trained on user logs, which can be further improved to capture context information and generalize across projects and coding styles. In this work, we propose a new scoring approach to ranking static completions using language models in a lightweight and model-agnostic way. Our method organizes all valid completions into a prefix tree and performs a single greedy decoding pass to collect token-level scores across the tree. This enables a precise token-aware ranking without needing beam search, prompt engineering, or model adaptations. The approach is fast, architecture-agnostic, and compatible with already deployed models for code completion. These findings highlight a practical and effective pathway for integrating language models into already existing tools within IDEs, and ultimately providing smarter and more responsive developer assistance.

Large Language Models are transforming software engineering, yet prompt management in practice remains ad hoc, hindering reliability, reuse, and integration into industrial workflows. We present Prompt-with-Me, a practical solution for structured prompt management embedded directly in the development environment. The system automatically classifies prompts using a four-dimensional taxonomy that encompasses intent, author role, software development lifecycle stage, and prompt type. To improve prompt reuse and quality, Prompt-with-Me suggests language refinements, masks sensitive information, and extracts reusable templates from a developer’s prompt library.Our taxonomy study of 1,108 real-world prompts demonstrates that modern LLMs can accurately classify software engineering prompts. Furthermore, our user study with 11 participants shows strong developer acceptance, with high usability (Mean SUS=73), low cognitive load (Mean NASA-TLX=21), and reported gains in prompt quality and efficiency through reduced repetitive effort. Lastly, we offer actionable insights for building the next generation of prompt management and maintenance tools for software engineering workflows.

The rise of AI tools is reshaping computer science education, shifting the focus from coding skills to teaching students how to effectively use these technologies. Understanding students' mental models and fostering computational and metacognitive skills are now essential, as over-reliance on AI can weaken critical thinking. This panel explores how a human-centered approach can balance these challenges, sharing strategies to optimize learning while addressing the risks of cognitive offloading in an AI-driven world.

Transformer-based language models are highly effective for code completion, with much research dedicated to enhancing the content of these completions. Despite their effectiveness, these models come with high operational costs and can be intrusive, especially when they suggest too often and interrupt developers who are concentrating on their work. Current research largely overlooks how these models interact with developers in practice and neglects to address when a developer should receive completion suggestions. To tackle this issue, we developed a machine learning model that can accurately predict when to invoke a code completion tool given the code context and available telemetry data. To do so, we collect a dataset of 200k developer interactions with our cross-IDE code completion plugin and train several invocation filtering models. Our results indicate that our small-scale transformer model significantly outperforms the baseline while maintaining low enough latency. We further explore the search space for integrating additional telemetry data into a pre-trained transformer directly and obtain promising results. To further demonstrate our approach’s practical potential, we deployed the model in an online environment with 34 developers and provided real-world insights based on 74k actual invocations.

In the PDF of this article, the pages were incorrectly numbered as ‘2303–2327’ when it should have been ‘3637–3661’. The page range was found to be just correct in the HTML version of the article. The original article has been corrected.

Does the training of large language models potentially infringe upon code licenses? Furthermore, are there any datasets available that can be safely used for training these models without violating such licenses? In our study, we assess the current trends in the field and the importance of incorporating code into the training of large language models. Additionally, we examine publicly available datasets to see whether these models can be trained on them without the risk of legal issues in the future. To accomplish this, we compiled a list of 53 large language models trained on file-level code. We then extracted their datasets and analyzed how much they overlap with a dataset we created, consisting exclusively of strong copyleft code.

Our analysis revealed that every dataset we examined contained license inconsistencies, despite being selected based on their associated repository licenses. We analyzed a total of 514 million code files, discovering 38 million exact duplicates present in our strong copyleft dataset. Additionally, we examined 171 million file-leading comments, identifying 16 million with strong copyleft licenses and another 11 million comments that discouraged copying without explicitly mentioning a license. Based on the findings of our study, which highlights the pervasive issue of license inconsistencies in large language models trained on code, our recommendation for both researchers and the community is to prioritize the development and adoption of best practices for dataset creation and management.

Language model-based code completion models have quickly grown in use, helping thousands of developers write code in many different programming languages. However, research on code completion models typically focuses on imperative languages such as Python and JavaScript, which results in a lack of representation for functional programming languages. Consequently, these models often perform poorly on functional languages such as Haskell. To investigate whether this can be alleviated, we evaluate the performance of two language models for code, CodeGPT and UniXcoder, on the functional programming language Haskell. We fine-tune and evaluate the models on Haskell functions sourced from a publicly accessible Haskell dataset on HuggingFace. Additionally, we manually evaluate the models using our novel translated HumanEval dataset. Our automatic evaluation shows that knowledge of imperative programming languages in the pre-training of LLMs may not transfer well to functional languages, but that code completion on functional languages is feasible. Consequently, this shows the need for more high-quality Haskell datasets. A manual evaluation on HumanEval-Haskell indicates CodeGPT frequently generates empty predictions and extra comments, while UniXcoder more often produces incomplete or incorrect predictions. Finally, we release HumanEval-Haskell, along with the fine-tuned models and all code required to reproduce our experiments on GitHub.

Integrated Development Environments (IDEs) have become central to modern software development, especially with the integration of Artificial Intelligence (AI) to enhance programming efficiency and decision-making. The study of in-IDE Human-AI Experience is critical in understanding how these AI tools are transforming the software development process, impacting programmer productivity, and influencing code quality.

We conducted a literature review to study the current state of in-IDE Human-AI Experience research, bridging a gap in understanding the nuanced interactions between programmers and AI assistants within IDEs. By analyzing 36 selected papers, our study illustrates three primary research branches: Design, Impact, and Quality of Interaction.

The trends, challenges, and opportunities identified in this paper emphasize the evolving landscape of software development and inform future directions for research, and development in this dynamic field. Specifically, we invite the community to investigate three aspects of these interactions: designing task-specific user interface, building trust, and improving readability.

In recent years, Large Language Models (LLMs) have gained significant popularity due to their ability to generate human-like text and their potential applications in various fields, such as Software Engineering. LLMs for Code are commonly trained on large unsanitized corpora of source code scraped from the Internet. The content of these datasets is memorized and emitted by the models, often in a verbatim manner. In this work, we will discuss the security, privacy, and licensing implications of memorization. We argue why the use of copyleft code to train LLMs is a legal and ethical dilemma. Finally, we provide four actionable recommendations to address this issue.

We report on the organization and results of the second edition of the tool competition from the International Workshop on Natural Language-based Software Engineering (NLBSE'23). As in the prior edition, we organized the competition on automated issue report classification, with a larger dataset. This year, we featured an extra competition on au-tomated code comment classification. In this tool competition edition, five teams submitted multiple classification models to automatically classify issue reports and code comments. The submitted models were fine-tuned and evaluated on a benchmark dataset of 1.4 million issue reports or 6.7 thousand code comments, respectively. The goal of the competition was to improve the classification performance of the baseline models that we provided. This paper reports details of the competition, including the rules, the teams and contestant models, and the ranking of models based on their average classification performance across issue report and code comment types.