{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,11]],"date-time":"2026-08-11T06:27:50Z","timestamp":1786429670432,"version":"build-2736575974"},"reference-count":30,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,8,11]],"date-time":"2026-08-11T00:00:00Z","timestamp":1786406400000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Artif. Intell."],"abstract":"\n Introduction<\/jats:title>\n We investigate how expressive inline code comments written in various developer styles, functional to progressively poetic, philosophical, and misleading, affect large language model (LLM) behavior during code optimization.<\/jats:p>\n <\/jats:sec>\n \n Methods<\/jats:title>\n In this pilot study, we used a controlledmerge sort implementation across five stylistic variants and evaluated GPT-5 and Claude Opus 4.1 under standardized console prompts, isolating the effect of embedded comment semiotic variation. Seven expert developers (three senior, four mid-level) scored model outputs against adapted ISO\/IEC 25010 criteria and novel LLM suggestibility index (LSI) framework.<\/jats:p>\n <\/jats:sec>\n \n Results<\/jats:title>\n \n Semiotic character of comments measurably altered code quality, with consensus-score reliability ICC\n (2, k)<\/jats:sub>\n = 0.65\u20130.81 for six of seven dimensions; single-rater Krippendorff's \u03b1 = 0.232 reflects substantial interpretive variability. Claude exhibited higher interpretive sensitivity (mean behavioral divergence 4.00; SD 1.16), while GPT-5 maintained stronger architectural fidelity (mean divergence 3.58; SD 1.26). Reflective comments (philosophical, conversational) were associated with Claude's highest maintainability scores in our panel (both\n M<\/jats:italic>\n = 4.00, ~8% above stock\n M<\/jats:italic>\n = 3.71), while the same philosophical comments reduced GPT-5 maintainability (\n M<\/jats:italic>\n = 2.86), suggesting asymmetric model responses to expressive context.\n <\/jats:p>\n <\/jats:sec>\n \n Conclusions<\/jats:title>\n These findings position inline comments as model-sensitive latent semantic prompts, with implications for AI-in-the-loop development and design of comment conventions for AI-assisted maintenance.<\/jats:p>\n <\/jats:sec>","DOI":"10.3389\/frai.2026.1784973","type":"journal-article","created":{"date-parts":[[2026,8,11]],"date-time":"2026-08-11T05:35:25Z","timestamp":1786426525000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["What really happens when a dev vibes with the code? An empirical study on LLM behavioral divergence in response to expressive code comments"],"prefix":"10.3389","volume":"9","author":[{"given":"Angela N.","family":"Johnson","sequence":"first","affiliation":[{"name":"College of Professional Studies, Northeastern University","place":["Boston, MA, United States"]},{"name":"Avania","place":["Boston, MA, United States"]},{"name":"The Real Cat Labs","place":["Boston, MA, United States"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,8,11]]},"reference":[{"key":"B1","volume-title":"Python Data Structures and Algorithms","author":"Baka","year":"2017"},{"key":"B2","article-title":"Vibe coding: toward an AI-native paradigm for semantic and intent-driven programming","author":"Bamil","year":"2025","journal-title":"arXiv"},{"key":"B3","doi-asserted-by":"publisher","first-page":"259","DOI":"10.1023\/B:EMSE.0000027786.04555.97","article-title":"The impact of background and experience on software inspections","volume":"9","author":"Carver","year":"2004","journal-title":"Empirical Softw. Eng."},{"key":"B4","article-title":"Evaluating large language models trained on code","author":"Chen","year":"2021","journal-title":"arXiv"},{"key":"B5","doi-asserted-by":"publisher","first-page":"7","DOI":"10.1093\/analys\/58.1.7","article-title":"The extended mind","volume":"58","author":"Clark","year":"1998","journal-title":"Analysis"},{"key":"B6","doi-asserted-by":"publisher","first-page":"654","DOI":"10.1016\/j.infsof.2007.02.011","article-title":"Using grounded theory to understand software process improvement: a study of Irish software product companies","volume":"49","author":"Coleman","year":"2007","journal-title":"Inf. Softw. Technol."},{"key":"B7","article-title":"AI-mediated code comment improvement","author":"Dhakal","year":"2025","journal-title":"arXiv"},{"key":"B8","article-title":"Vibe coding in practice: Motivations, challenges, and a future outlook\u2014a grey literature review","author":"Fawzy","year":"2025","journal-title":"arXiv"},{"key":"B9","doi-asserted-by":"publisher","first-page":"70","DOI":"10.1109\/WCRE.2007.21","article-title":"\u201cDo code and comments co-evolve? On the relation between source code and comment changes,\u201d","author":"Fluri","year":"2007","journal-title":"Proc. 14th working conf. reverse engineering (WCRE)"},{"key":"B10","doi-asserted-by":"publisher","first-page":"556","DOI":"10.32996\/jcsts.2025.7.4.66","article-title":"Democratizing software engineering through generative AI and vibe coding: the evolution of no-code development","volume":"7","author":"Gadde","year":"2025","journal-title":"J. Comput. Sci. Technol. Stud."},{"key":"B11","article-title":"A survey of vibe coding with large language models","author":"Ge","year":"2025","journal-title":"arXiv"},{"key":"B12","volume-title":"The Discovery of Grounded Theory: Strategies for Qualitative Research","author":"Glaser","year":"1967"},{"key":"B13","doi-asserted-by":"publisher","first-page":"1068","DOI":"10.18653\/v1\/2025.acl-long.54","article-title":"\u201cPersonality-guided code generation using large language models,\u201d","author":"Guo","year":"2025","journal-title":"Proc. 63rd annu. meeting assoc. comput. linguistics (ACL), Vol. 1: Long Papers"},{"key":"B14","doi-asserted-by":"publisher","first-page":"169","DOI":"10.18653\/v1\/2020.inlg-1.23","article-title":"\u201cTwenty years of confusion in human evaluation: NLG needs evaluation sheets and standardised definitions,\u201d","author":"Howcroft","year":"2020","journal-title":"Proc. 13th int. conf. natural language generation (INLG)"},{"key":"B15","volume-title":"Vibe Coding as Semiotic Interface: Experimental Insights into LLM Behavior and Co-Creative Code Authorship","author":"Johnson","year":"2025"},{"key":"B16","unstructured":"Jones\n N.\n \n \n The Vibe Coding Bible: How to Build Useful Things with Short Prompts. Nate's Substack\n \n 2025"},{"key":"B17","unstructured":"Karpathy\n A.\n \n \n There's a new kind of coding I call \u2018vibe coding'. X (formerly Twitter)\n \n 2025"},{"key":"B18","volume-title":"Python for BEGINNERS, 1st Edn","author":"Kaswan","year":"2022"},{"key":"B19","doi-asserted-by":"publisher","first-page":"22199","DOI":"10.52202\/068431-1613","article-title":"\u201cLarge language models are zero-shot reasoners,\u201d","author":"Kojima","year":"2022","journal-title":"Advances in neural information processing systems (NeurIPS), Vol. 35"},{"key":"B20","article-title":"Large language models understand and can be enhanced by emotional stimuli","author":"Li","year":"2023","journal-title":"arXiv"},{"key":"B21","article-title":"Prompt stability in code LLMs: Measuring sensitivity across emotion- and personality-driven variations","author":"Ma","year":"2025","journal-title":"arXiv"},{"key":"B22","unstructured":"Mehta\n I.\n \n \n A Quarter of Startups in YC's Current Cohort Have Codebases That Are Almost Entirely AI-Generated\n \n 2025"},{"key":"B23","doi-asserted-by":"crossref","DOI":"10.1145\/3786181.3788724","article-title":"\u201cCode roulette: how prompt variability affects LLM code generation,\u201d","volume-title":"Proc. 3rd int. workshop on large language models for code (LLM4Code \u201826), co-located with ICSE 2026","author":"Paleyes","year":"2026"},{"key":"B24","unstructured":"\u201cRegulatory strategy for digital therapeutics and artificial intelligence-enabled devices,\u201d\n \n \n \n Parks\n A.\n \n \n Butt\n N.\n \n \n Farcas\n S.\n \n \n Johnson\n A.\n \n \n Digital Health Laws and Regulations 2026, International Comparative Legal Guides (ICLG)\n \n 2026"},{"key":"B25","article-title":"Good vibrations? A qualitative study of co-creation, communication, flow, and trust in vibe coding","author":"Pimenova","year":"2025","journal-title":"arXiv"},{"key":"B26","article-title":"\u201cVibe coding: programming through conversation with artificial intelligence,\u201d","volume-title":"Proc. 36th annu. conf. psychology of programming interest group (PPIG)","author":"Sarkar","year":"2025"},{"key":"B27","unstructured":"2025 Stack Overflow Developer Survey\n \n 2025"},{"key":"B28","doi-asserted-by":"crossref","DOI":"10.1145\/2884781.2884833","article-title":"\u201cGrounded theory in software engineering research: a critical review and guidelines,\u201d","volume-title":"Proc. 38th int. conf. software engineering (ICSE)","author":"Stol","year":"2016"},{"key":"B29","doi-asserted-by":"publisher","first-page":"24824","DOI":"10.52202\/068431-1800","article-title":"\u201cChain-of-thought prompting elicits reasoning in large language models,\u201d","author":"Wei","year":"2022","journal-title":"Advances in neural information processing systems (NeurIPS), Vol. 35"},{"key":"B30","unstructured":"Vibe Coding Is the Future. Lightcone Podcast\n \n 2025"}],"container-title":["Frontiers in Artificial Intelligence"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frai.2026.1784973\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,8,11]],"date-time":"2026-08-11T05:35:28Z","timestamp":1786426528000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frai.2026.1784973\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8,11]]},"references-count":30,"alternative-id":["10.3389\/frai.2026.1784973"],"URL":"https:\/\/doi.org\/10.3389\/frai.2026.1784973","relation":{},"ISSN":["2624-8212"],"issn-type":[{"value":"2624-8212","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,8,11]]},"article-number":"1784973"}}