Tag
Gergely Orosz recommends Dr Cat Hicks's recent book "Psychology of Software Teams," highlighting how companies often misunderstand how developers work — now compounded by the sudden arrival of AI in the workplace.
一条推文感慨如今的软件工程师都变成了「高级智能体管理员」，反映了 AI 编程助手对开发者工作方式的影响。
@poteto (Lauren) from the Grok Bot team proposes a four-layer "inverted pyramid" framework for trusting AI Agent outputs: prioritize investing in codebase improvements and static analysis over rule files and manual review, and engineer the environment so Agents do the right thing by default.
开发者并未被 AI 取代，但 AI 取代了「理解自己代码」的环节：代码由模型生成后无人真正读过，调试退化为把报错喂回模型。文章引用 Veracode 对 100+ 模型的测试（45% 生成代码存在已知安全问题），质疑这种「用理解换速度」的权衡是否会反噬整个行业。
Zach Lloyd shares his talk from AI Engineer NYC arguing that software engineering is becoming factory engineering, and highlights the emerging need for factory benchmarking as a focus area for engineers.
Gergely Orosz highlights engineer Cindy Sridharan's claim that Claude, when given proper instructions and context via a tailored claude.md, now writes better code than she does almost always.
斯坦福HAI的一篇报道指出，AI编程代理在团队协作方面表现不佳，单兵作战尚可但难以在协作式软件开发场景中有效配合。
SVPino argues that AI tools increasing developer efficiency will follow Jevons paradox, raising rather than reducing overall demand for software and software engineers.
SlopCodeBench 的 10/2026 更新显示，Opus 5.5 与 Astra 在整体表现中领先，而 xhigh 被认为存在负面影响。该基准通过逐步披露功能检查点、让模型在继承同一代码库的多阶段挑战中工作，更贴近真实软件开发流程。
一篇个人博客文章，作者分享自己宗教研究学位如何在软件工程职业生涯中发挥作用，探讨意识形态与编程如何同时作用于人类与社会，以及内部人/外部人视角在理解复杂信念体系中的价值。
OpenHands 宣布成为 NVIDIA Nemotron 3.5 Super VL 的发布合作伙伴，将这款开放多模态模型引入长期运行的软件工程智能体工作流，使智能体能够理解截图、仪表盘、文档和图表等非文本上下文。
Scott Aaronson reflects on how AI's rapid advances in mathematics make earlier skepticism about AI replacing software engineers seem untenable, arguing math represents the pinnacle of human intellect and is now being challenged by AI.
Gergely Orosz 的 Pragmatic Engineer 通讯文章回顾 2025 年软件工程与 LLM 的结合现状，涵盖 Anthropic、Cursor、Google、Amazon 等公司如何使用 Claude Code、Cursor、Windsurf 等 AI 开发工具，以及资深工程师们的观察。
An introductory tutorial on the C++26 contracts feature (P2900), showing how preconditions and postconditions replace traditional assert-based checks, along with compiler options, evaluation semantics, and common pitfalls.
一位刚晋升为工程主管的博主反思：如今技术面试普遍要求使用 LLM，但 AI 编码工具的订阅费用却可能把经济拮据的优秀工程师挡在门外，他分享了如何调整招聘流程来降低候选人的成本负担。
A practitioner argues that AI agent builders are rediscovering problems that basic software engineering—input validation, data modeling, process isolation, access control—already solved, and that vibe coding skips these cross-cutting concerns. The post calls for applying established design and architecture principles to agent workflows.
该论文通过系统文献综述与128位从业者评估，提出并验证了AI鲁棒性审计框架的44项元需求（分为16个类别），以帮助组织构建和评估AI鲁棒性审计框架。
一篇深度分析文章探讨为什么给软件项目增加更多 AI 代理并不能缩短交付周期，借弗雷德·布鲁克斯的《人月神话》类比指出瓶颈在于决策、验证与协调，并结合 Google Research、Cursor 和 Anthropic 的实验说明输出规模只有在有良好组织结构时才能真正扩展。
一条带有讽刺意味的推文，感慨“软件已死”，哀叹多年积累的解决难题的技能似乎不再有用，暗指 AI 编程工具对软件开发行业的影响。
trq212 and fnthawar discuss how the key skill when working with coding agents is 'being in the details'—like running a company, you can't know everything, but obsessively caring about the right details (code, architecture, design, UX) is what makes something great.