Barbara Weber, Olaf Zimmermann, Marco Franceschetti · 2026 · arXiv (Cornell University)
DOI: 10.48550/arxiv.2610.30916 · cited by 0 works (OpenAlex) · open-access PDF
Кодирующие ассистенты, выполняющие работу в режиме реального времени, широко сообщают о том, что они помогают ускорить производство кода; большинство учетных записей останавливаются там. Мы сообщаем о строительстве и развертывании системы оценки соответствия-критического, человеческого-в ходу-оценки-оценки, построенной одним разработчиком, который не писал код вручную, защищенный заблокированным процессом от лица заинтересованного лица до аудируемого поведения. Первоначальная система была построена в течение трех рабочих дней и могла стать настоящей когортой. Мы делаем одно заявление из этого опыта: ИИ меняет программное обеспечение, но помогает программированию быстрее, чем изменение того, что редко. Написание кода становится обильным. Наде
Machine translation (Meta M2M100); the English original follows.
Coding assistants are widely reported to speed up code production; most accounts stop there. We report the end-to-end construction and deployment of a compliance-critical, human-in-the-loop exam-grading system, built by a single developer who wrote no code by hand, protected by a gated process from stakeholder persona to audited behavior. The initial system was built in three working days and could grade a real cohort. We draw one claim from this experience: AI changes software engineering less by making programming faster than by changing what is scarce. Writing code becomes abundant. Reliable evaluation-knowing whether a result is correct and meets the actual need-remains necessary at every iteration and becomes the binding constraint. Engineering judgment remains a second scarcity. Several defects passed the test suite: they sat at the boundaries the tests had mocked, the browser and the language-model API, or appeared only in real use. Deployment exposed assumptions that local tests had missed. Moreover, an AI assistant acting across many steps bypassed process checkpoints until we made them enforceable gates. All of this was observed in a triple union: the developer was also the end user and domain expert, making expert judgment immediately available and evaluation as cheap as it can be. This favors the iteration gains, which may shrink where evaluation requires additional people or expertise, but makes the case a hard test for evaluation scarcity: evaluation still constrained progress. We draw five lessons: run a gated, traceable construction process; specify what is knowable and discover the rest against a running prototype; validate in the real environment; gate the assistant, not only the work; record decisions with their assumptions and the process state the assistant forgets. A second project showed similar patterns; an independent single-expert case corroborates the shift.
Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese
Topics: Software engineering, Agile software development, Computer science, Software deployment, Software inspection, Process (computing)
GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.30916, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange