AI Tools Accelerates Coding, But Not Overall Software Delivery, GitLab Research Finds GitLab 研究发现：AI 工具加速编码，但并未整体提升软件交付

GitLab's 2026 AI Accountability Report highlights an AI Paradox: although 78% of developers say they code faster, overall software delivery has not accelerated due to downstream testing and review bottlenecks, as well as new challenges for enterprise governance and traceability.GitLab 的 2026 年 AI 责任报告揭示了一个 AI 悖论：尽管 78% 的开发人员表示他们编码速度更快，但由于下游测试和审查瓶颈，以及企业治理和可追溯性的新挑战，整体软件交付并未加速。

According to GitLab research, AI has made the task of writing software faster, with 78% of respondents reporting faster code output and 73% noting that overall code quality has improved. However, AI tools have uncovered a deeper issue: organizations cannot easily control what they are shipping, as governance, traceability, and accountability have failed to keep pace, creating a structural imbalance.根据 GitLab 的研究，AI 使编写软件的任务变得更快，78% 的受访者表示代码输出速度更快，73% 的受访者指出整体代码质量有所提高。然而，AI 工具揭示了一个更深层次的问题：组织无法轻易控制他们正在交付的内容，因为治理、可追溯性和问责制未能跟上步伐，造成了结构性失衡。

The report defines AI accountability as the organizational and technical capability to answer three questions about any line of AI-generated code: where did it come from, what was it meant to do, and who is responsible for it once it's in production? Most organizations cannot answer those questions today.该报告将 AI 问责制定义为组织和技术能力，能够回答关于任何一行 AI 生成代码的三个问题：它来自哪里，它的目的是什么，以及一旦投入生产由谁负责？如今，大多数组织都无法回答这些问题。

Indeed, 85% of respondents "agree AI has shifted the bottleneck from writing code to reviewing and validating it". As a result, 79% report that the overall software delivery process has not kept pace with coding.事实上，85% 的受访者“认为 AI 已将瓶颈从编写代码转移到审查和验证代码”。因此，79% 的人报告说，整体软件交付流程未能跟上编码的速度。

As Manav Khurana, chief product and marketing officer at GitLab, notes, recent events such as supply chain attacks, reliability issues, and regulators expectations, show that traceability is a critical concern to prevent organizational exposure. Respondents point to three main factors contributing to make traceability harder: difficulty distinguishing AI-generated from human-written code (43%), fragmented toolchains (40%), and systems that don't track code origin (39%). Reflecting this gap, GitLab's report observes that while:正如 GitLab 的首席产品和营销官 Manav Khurana 指出的那样，最近发生的供应链攻击、可靠性问题和监管机构的期望等事件表明，可追溯性是防止组织暴露的关键问题。受访者指出了使可追溯性更难的三个主要因素：难以区分 AI 生成的代码和人工编写的代码（43%），工具链碎片化（40%），以及不跟踪代码来源的系统（39%）。反映出这种差距，GitLab 的报告观察到：

87% are confident their team could determine within 24 hours whether AI-generated code contributed to a production incident, [only] 34% of organizations that experienced an incident in the past year could not actually make that determination.87% 的人有信心，他们的团队可以在 24 小时内确定 AI 生成的代码是否导致了生产事件，但过去一年中经历过事件的组织中，只有 34% 的组织实际上无法做出这样的判断。

For 85% of respondents, the solution lies in stronger governance, i.e. establishing clear policies to ensure provenance and accountability of AI-generated code. Without it, 83% of organizations view the accumulation of AI-generated code a risk, with 44% ranking it among their top technological concerns.对于 85% 的受访者来说，解决方案在于更强的治理，即建立明确的政策以确保 AI 生成代码的出处和问责制。没有它，83% 的组织认为 AI 生成代码的积累存在风险，其中 44% 的组织将其列为最主要的担忧技术之一。

The findings in GitLab's research echo sentiments from an earlier Reddit thread, where the OP notes that continued investment in AI increased "speed at the text editor/terminal layer", but left them spending most of their time "wading through the quicksand of agile/jira and middle management bloat". Another user, YourMatt, similarly noted that while the gains in coding speed were impressive, they did little to address the broader inefficiencies that ultimately constrain delivery:GitLab 研究中的发现呼应了早期 Reddit 帖子中的观点，其中 OP 指出，对 AI 的持续投资增加了“文本编辑器/终端层面的速度”，但让他们花费大部分时间“在敏捷/Jira 和中层管理臃肿的泥潭中挣扎”。另一位用户 YourMatt 同样指出，尽管编码速度的提升令人印象深刻，但它们对解决最终限制交付的更广泛的低效率问题几乎没有帮助：

sprint after sprint though, nobody in our focus group was churning out more story points than before. It really made it apparent how the mechanics of coding is a relatively small portion of our jobs.尽管一个又一个冲刺，我们焦点小组中没有人比以前完成更多的故事点。这确实清楚地表明，编码的机制在我们工作中所占的比例相对较小。

In a more recent thread, Mestyo reinforces this view, arguing that the majority of work performed by individual contributors cannot be meaningfully accelerated by AI coding tools.在最近的一篇帖子中，Mestyo 强化了这一观点，认为个人贡献者执行的大部分工作无法通过 AI 编码工具得到有意义的加速。

As a final note from the community, Reddit user EveryDay_is_LegDay echoes this perspective, arguing from experience that testing remains the primary bottleneck and that "producing code faster only exacerbates the problems of most development teams".最后，来自社区的 Reddit 用户 EveryDay_is_LegDay 从经验出发，呼应了这一观点，认为测试仍然是主要的瓶颈，并且“更快地生成代码只会加剧大多数开发团队的问题”。

About the Author

Sergio De Simone

Sergio De Simone is a software engineer. Sergio has been working as a software engineer for over twenty five years across a range of different projects and companies, including such different work environments as Siemens, HP, and small startups. For the last 10+ years, his focus has been on development for mobile platforms and related technologies. He is currently working for BigML, Inc., where he leads iOS and macOS development.

Show moreShow less