Revisiting Sentiment Analysis for Software Engineering in the Era of Large Language Models

Evaluating gpt-4 for zero-shot classification of bleeding and clotting events: Can large language models serve as second reviewers?

Generative AI systems based on Large Language Models (LLMs) have demonstrated remarkable capabilities in software engineering tasks, from code generation to natural language processing. However, a significant gap persists between curated demonstrations and production-grade deployment. This technical report presents a practitioner-driven analysis of LLM limitations encountered during sustained, real-world use across software development workflows. Drawing from multiple case studies—including natural-language-driven development (“vibe coding”), API integration, multi-file refactoring, and general-purpose question answering—we identify and taxonomize four critical failure modes: (1) the Complexity Cliff, where LLM performance degrades non-linearly as task interdependency grows; (2) Context Window Blindness, where finite attention spans cause silent contract violations across distributed codebases; (3) the Memory Illusion, where session discontinuity erases accumulated architectural knowledge; and (4) Confident Hallucination, where models generate plausible but fabricated outputs indistinguishable in tone from correct ones. We formalize the Verification Paradox—an inverse relationship between a user’s need for AI assistance and their capacity to validate its outputs—and propose a practical five-strategy framework for effective human–AI collaboration in software engineering contexts. We further introduce the concept of Contextual Reasoning Failure, evidenced by cases where LLMs optimize for literal query patterns while ignoring situational logic obvious to any human observer. Our findings suggest that current LLMs, while powerful pattern-matching engines, lack the contextual reasoning, persistent memory, and epistemic self-awareness necessary for reliable autonomous operation, and that practitioner expertise remains the critical safeguard against AI-induced defects in production systems.

Large Language Models (LLMs) have become a crucial tool in Visual Question Answering (VQA) for handling knowledge-intensive questions in few-shot or zero-shot scenarios. However, their reliance on massive training datasets often causes them to inherit language biases during the acquisition of knowledge. This limitation imposes two key constraints on existing methods: (1) LLM predictions become less reliable due to bias exploitation, and (2) despite strong knowledge reasoning capabilities, LLMs still struggle with out-of-distribution (OOD) generalization. To address these issues, we propose Object Attribute Description Promoter (OAD-Promoter), a novel approach for enhancing LLM-based VQA by mitigating language bias and improving domain-shift robustness. OAD-Promoter comprises three components: the Object-concentrated Example Generation (OEG) module, the Memory Knowledge Assistance (MKA) module, and the OAD Prompt. The OEG module generates global captions and object-concentrated samples, jointly enhancing visual information input to the LLM and mitigating bias through complementary global and regional visual cues. The MKA module assists the LLM in handling OOD samples by retrieving relevant knowledge from stored examples to support questions from unseen domains. Finally, the OAD Prompt integrates the outputs of the preceding modules to optimize LLM inference. Experiments demonstrate that OAD-Promoter significantly improves the performance of LLM-based VQA methods in few-shot or zero-shot settings, achieving new state-of-the-art results.

With the rapid advancement of large language models (LLMs), code translation has become a critical yet challenging task in software engineering. Existing evaluations of this task remain unreliable due to two key limitations: the absence of specialized function-level code translation benchmarks and flawed evaluation pipelines. Specifically, current pipelines lack post-processing mechanisms to extract target code from LLMs’ inherently unstable output formats, directly resulting in non-executable translations and false negatives in evaluation. To address these issues, we introduce FuncTransEval, an enhanced multilingual benchmark. It comprises 400×4 functions across four programming languages (Python, C++, Java, and JavaScript) and 12 bidirectional translation pairs, each paired with executable unit tests for fine-grained functional validation. Using FuncTransEval, we systematically assess LLMs’ true capabilities in code translation and identify their core limitations: inadequate understanding of the implicit information such as function logic, type specifications, and cross-language differences. To mitigate these shortcomings, we propose ICAM-Trans (Implicit-Context-Aware Multi-Agent Framework for function-level code translation). ICAM-Trans is a hierarchical, autonomous multi-agents architecture that explicitly uncovers and leverages these implicit contextual information during translation. It employs a Translation Orchestrate Agent (TOA) to autonomously coordinate four specialized Context Analysis Agents (CAAs), which conduct pre-translation analyses on different aspects. Insights from these analyses guide a Context-Aware Translation Agent (CTA) to generate semantically faithful target code. Experiments on FuncTransEval demonstrate that ICAM-Trans consistently outperforms strong baselines, validating its effectiveness in achieving high-fidelity and interpretable function-level code translation.

Recent works have shown that Large Language Models (LLMs) have a tendency to memorize patterns and biases present in their training data, raising important questions about how such memorized content influences model behavior.One such concern is the emergence of political bias in LLM outputs.In this paper, we investigate the extent to which LLMs' political leanings reflect memorized patterns from their pretraining corpora.We propose a method to quantitatively evaluate political leanings embedded in the large pretraining corpora.Subsequently we investigate to whom are the LLMs' political leanings more aligned with, their pretrainig corpora or the surveyed human opinions.As a case study, we focus on probing the political leanings of LLMs in 32 U.S. Supreme Court cases, addressing contentious topics such as abortion and voting rights.Our findings reveal that LLMs strongly reflect the political leanings in their training data, and no strong correlation is observed with their alignment to human opinions as expressed in surveys.These results underscore the importance of responsible curation of training data, and the methodology for auditing the memorization in LLMs to ensure human-AI alignment.

Generative AI in cybersecurity: A comprehensive review of LLM applications and vulnerabilities

Large Language Models (LLMs) commonly report high confidence, even in domains where their underlying knowledge or training data is limited. This mismatch can negatively impact model reliability, particularly affecting educational applications where users may not recognize errors. To detect these knowledge gaps, LLM knowledge must be assessed after training. In this work, we compare LLM prompts to self-assess knowledge of content in two ways: rank-ordering and direct confidence scoring (e.g., 1-5). For human metacognition, rankings or A/B comparisons are more reliable, so we hypothesize that LLMs’ rankings may also be more effective than scores. We compare LLM-generated Overall Rankings and confidence scores for 15 topics against two external estimates of LLM knowledge: expert human ratings and search result counts from Bing, Google, and Wikipedia. We also consider Anchored Rankings in which each document to be rated by the LLM is compared to a set of documents with known expert scores. Comparing across different document representations and different LLMs, Spearman correlations with expert ratings are generally: positive and relatively high with Anchored Rankings having the highest correlation (ρ ranging from 0.74 to 1.0) followed by Overall Rankings and then confidence scores. In contrast, search-based signals have a weaker and variable alignment suggesting that web popularity is a noisy signal for estimating LLM familiarity with content. Overall, these findings suggest that relative self-assessment through rankings provides an interpretable signal of LLM self-knowledge. This can be used to select specialized prompts or workflows for topics where an LLM has less knowledge.

e13630 Background: As many as 60% of prior authorization requests are denied, yet coverage approval occurs for more than 60% of appeals for some therapies. Appeal processes encumber providers and increase burnout, but large language models (LLMs) may aid providers by drafting appeal letters. We evaluated LLM performance at this task for radiotherapy denials. Methods: Three commercially accessible LLMs were evaluated: generative pre-trained transformer 3.5 (GPT3.5), GPT4, and GPT4+web with internet search capacity (OpenAI, Inc., San Francisco, CA). A fourth LLM, GPT3.5-FT, was developed by fine-tuning GPT3.5 in a HIPAA-complaint local environment. The fine-tuning training data comprised 53 insurance denial appeal letters prepared by radiation oncologists and paired prompts describing the clinical history and appeal intent. Training data were enriched in appeal letters for proton radiotherapy, stereotactic body radiotherapy, and image-guided radiotherapy for myriad clinical scenarios. Twenty prompts, each requesting a letter for a simulated patient history, were programmatically presented to the LLMs. Three radiation oncologists, who were blinded to the LLM source, scored letter outputs across four domains: language syntax and semantics, clinical detail inclusion, clinical reasoning validity, and overall readiness for insurer submission. Additionally, one radiation oncologist scored the authenticity and relevance of literature sources cited in output letters, which were requested by several test prompts. Interobserver agreement between radiation oncologist scores was determined by Cohen’s kappa coefficient. Scores were compared between LLMs with non-parametric statistical tests. Results: Agreement between radiation oncologists’ scores was moderate-to-excellent across all domains (median κ = 0.68, minimum κ = 0.41). GPT3.5, GPT4, and GPT4+web drafted letters that, by mode average, were semantically and syntactically clear, included all provided clinical history without confabulation, clinically reasoned with few necessary revisions, and overall were submissible to an insurer with minor revisions. GPT4 and GPT4+web clinically reasoned better than GPT3.5 (p values < 0.001). In contrast, GPT3.5-FT performance was inferior to other LLMs across all domains (p values < 0.001). LLMs were poor at identifying, citing, and summarizing relevant literature unless provided in the prompt. Conclusions: LLMs can draft insurance appeal letters for radiotherapy services that require few revisions yet are poor at referencing relevant literature. Contrary to our hypothesis, fine-tuning with data from our department compromised LLM performance.

Optimizing patent claims is a critical yet challenging task, demanding careful balance between maximizing novelty and preserving legal scope. Manual claim drafting is labor-intensive, costly, and inherently inconsistent, while conventional Large Language Models (LLMs) often lack the structured, iterative reasoning essential for precise claim refinement. To address these challenges, we introduce Tree of Claims (ToC), an innovative framework that redefines claim editing as a guided search problem. ToC synergistically integrates Monte Carlo Tree Search (MCTS) with a collaborative multi-agent system, comprising an LLM-based EditorAgent that proposes contextually grounded edits, and an ExaminerAgent that mimics patent examiner critiques through structured, chain-of-thought analyses of novelty and prior art disclosure. Driven by a carefully designed multi-objective reward function, ToC jointly optimizes novelty, scope retention, and semantic coherence. Experimental evaluation on a benchmark of 1145 claims demonstrates that ToC significantly outperforms standard LLMs in zero-shot and few-shot scenarios, achieving an average composite score improvement of 8%, and up to 9% in certain cases. Extensive experiments, including detailed ablation studies, validate ToC’s efficacy in generating superior, legally robust claim revisions. Overall, ToC establishes a transparent, controllable, and interpretable methodology that effectively bridges advanced LLM reasoning capabilities with strategic MCTS planning for structured patent claim optimization.

Global health is currently facing serious challenges due to the increasing number of chronic disease patients, such as those with heart failure, diabetes, and cancer. This issue arises from the limitations of electronic health record (EHR) systems, which are not yet fully capable of ensuring accurate clinical diagnoses because of potential data input errors and delays in symptom identification by medical personnel. In response to this issue, this paper focuses on the integration of medical tabular data with a classification approach based on classical machine learning (ML) and large language models (LLM) to improve the accuracy of patient diagnosis predictions. This paper aims to develop and compare the performance of various ML models, such as XGBoost, SVM, and logistic regression, as well as LLM models like Gemini, LLaMA, and Qwen in fine-tuning, few-shot, and zero-shot scenarios. The paper results show that the combination of Gemini and the few-shot approach (250 shots) achieved the highest accuracy of up to 99.8% in predicting heart failure risk. The main finding of this study is that the narrative text representation of tabular data processed with LLM significantly enhances contextual understanding and classification accuracy, making this approach highly potent for application in AI-based clinical decision-making.

Code summarization is an important task in software engineering that helps developers understand and maintain code by generating natural language summaries. Existing approaches predominantly rely on single models, facing a dilemma: directly deploying large language models (LLMs) incurs high training costs, while lightweight models specialized for summarization are constrained by the quality of training data and their ability to capture the complex structural semantics of code. This highlights the urgent need for synergistic collaboration between large and small models in cloud computing environments. To address these issues, this paper proposes a cloud-assisted code summarization framework. First, we achieve code enhancement by invoking cloud-deployed LLM services. The specific workflow involves using preset prompt templates to guide the model in evaluating code quality and automatically repairing defects based on its feedback, thereby constructing high-quality datasets Java-QE and Python-QE. Second, for efficient edge deployment, we introduce HiSum: AST Hierarchy-Aware Code Summarization model, a lightweight model. HiSum transforms code AST into Directed Syntax Graphs (DSG) to preserve structural semantics, encodes them via a directed graph convolutional network and decode to improve summary quality. Experimental results show that our framework significantly enhances code summarization performance. On the constructed Java-QE and Python-QE datasets, the HiSum model achieves notable improvements over state-of-the-art baselines in BLEU, METEOR, and ROUGE-L metrics (increases of 1.06%, 1.98%, 3.12% for Java-QE, and 1.46%, 3.24%, 2.20% for Python-QE, respectively). This research provides a solution that utilizes cloud LLM-assisted data enhancement to empower a lightweight hierarchical-aware model.

Background: Recent advancements in large language model (LLM) technologies have introduced powerful open-source instruction-tuned LLMs that match the text generation quality of leading models like GPT-4. Despite accelerating LLM adoption in sensitive-information environments, the lack of disclosed training data hinders replication and makes these achievements exclusive to specific models. Purpose: Given the multilingual nature of the latest iteration of open-source LLMs, the benefits of training language-specific LLMs diminish, leaving computational efficiency as the sole guaranteed advantage of this computationally-expensive procedure. This work aims to address the language-adaptation limitations posed by restricted access to high-quality instruction-tuning data, offering a more cost-effective pipeline. Method: To tackle language-adaptation challenges, we introduce Learned Embedding Propagation (LEP), a novel method with lower training data requirements and minimal disruption of existing LLM knowledge. LEP employs an innovative embedding propagation technique, bypassing the need for instruction-tuning and directly integrating new language knowledge into any instruct-tuned LLM variant. Additionally, we developed Darumeru, a new benchmark for evaluating text generation robustness during training, specifically tailored for Russian adaptation. Results: We applied the LEP method to adapt LLaMa-3-8B and Mistral-7B for Russian, testing four different vocabulary adaptation scenarios. Evaluation demonstrates that LEP achieves competitive performance levels, comparable to OpenChat 3.5 and LLaMa-3-8B-Instruct. Further improvements were observed through self-calibration and additional instruction-tuning steps, enhancing task-solving capabilities beyond the original models. Conclusion: LEP offers a viable and efficient alternative to traditional language-specific instruction-tuning, significantly reducing the costs associated with language adaptation while maintaining or surpassing the performance benchmarks set by contemporary LLMs.

Cognitive biases are systematic deviations in thinking that lead to irrational judgments and problematic decision-making, extensively studied across various fields. Recently, large language models (LLMs) have shown advanced understanding capabilities but may inherit human biases from their training data. While social biases in LLMs have been well-studied, cognitive biases have received less attention, with existing research focusing on specific scenarios. The broader impact of cognitive biases on LLMs in various decision-making contexts remains underexplored. We investigated whether LLMs are influenced by the threshold priming effect in relevance judgments, a core task and widely-discussed research topic in the Information Retrieval (IR) coummunity. The priming effect occurs when exposure to certain stimuli unconsciously affects subsequent behavior and decisions. Our experiment employed 10 topics from the TREC 2019 Deep Learning passage track collection, and tested AI judgments under different document relevance scores, batch lengths, and LLM models, including GPT-3.5, GPT-4, LLaMa2-13B and LLaMa2-70B. Results showed that LLMs tend to give lower scores to later documents if earlier ones have high relevance, and vice versa, regardless of the combination and model used. Our finding demonstrates that LLM's judgments, similar to human judgments, are also influenced by threshold priming biases, and suggests that researchers and system engineers should take into account potential human-like cognitive biases in designing, evaluating, and auditing LLMs in IR tasks and beyond

BackgroundThe launch of ChatGPT (OpenAI) in November 2022 attracted public attention and academic interest to large language models (LLMs), facilitating the emergence of many other innovative LLMs. These LLMs have been applied in various fields, including health care. Numerous studies have since been conducted regarding how to use state-of-the-art LLMs in health-related scenarios.ObjectiveThis review aims to summarize applications of and concerns regarding conversational LLMs in health care and provide an agenda for future research in this field.MethodsWe used PubMed, ACM, and the IEEE digital libraries as primary sources for this review. We followed the guidance of PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) to screen and select peer-reviewed research articles that (1) were related to health care applications and conversational LLMs and (2) were published before September 1, 2023, the date when we started paper collection. We investigated these papers and classified them according to their applications and concerns.ResultsOur search initially identified 820 papers according to targeted keywords, out of which 65 (7.9%) papers met our criteria and were included in the review. The most popular conversational LLM was ChatGPT (60/65, 92% of papers), followed by Bard (Google LLC; 1/65, 2% of papers), LLaMA (Meta; 1/65, 2% of papers), and other LLMs (6/65, 9% papers). These papers were classified into four categories of applications: (1) summarization, (2) medical knowledge inquiry, (3) prediction (eg, diagnosis, treatment recommendation, and drug synergy), and (4) administration (eg, documentation and information collection), and four categories of concerns: (1) reliability (eg, training data quality, accuracy, interpretability, and consistency in responses), (2) bias, (3) privacy, and (4) public acceptability. There were 49 (75%) papers using LLMs for either summarization or medical knowledge inquiry, or both, and there are 58 (89%) papers expressing concerns about either reliability or bias, or both. We found that conversational LLMs exhibited promising results in summarization and providing general medical knowledge to patients with a relatively high accuracy. However, conversational LLMs such as ChatGPT are not always able to provide reliable answers to complex health-related tasks (eg, diagnosis) that require specialized domain expertise. While bias or privacy issues are often noted as concerns, no experiments in our reviewed papers thoughtfully examined how conversational LLMs lead to these issues in health care research.ConclusionsFuture studies should focus on improving the reliability of LLM applications in complex health-related tasks, as well as investigating the mechanisms of how LLM applications bring bias and privacy issues. Considering the vast accessibility of LLMs, legal, social, and technical efforts are all needed to address concerns about LLMs to promote, improve, and regularize the application of LLMs in health care.

Self-bootstrapping automated program repair: using LLMs to generate and evaluate synthetic training data for bug repair