内容来源：https://www.geekwire.com/2025/engineering-leader-survey-ai-isnt-leading-to-massive-job-cuts-but-its-siphoning-off-weak-performers/

内容总结：

【科技行业观察】人工智能浪潮下，软件工程师价值分化加剧，中国AI应用步伐领先

近日，西雅图技术人才评估公司Karat发布《AI workforce transformation》报告指出，人工智能工具正在重塑软件开发行业，并加剧工程师之间的能力鸿沟。报告基于对美、印、中三国400名工程团队管理者的调研，揭示了AI时代技术人才市场的深刻变革。

核心发现：

1. 人才价值两极分化：73%的领导者认为顶尖工程师价值至少是其薪酬的三倍；而59%的管理者指出，能力较弱的工程师在AI时代创造零价值甚至负价值。
2. AI工具提升生产力：AI平均提升工程师34%的生产效率，日常应用最广的场景是代码生成（83%）与测试、质量审查（61%）。
3. 中国AI应用领先：在AI技术采纳与准备度上，中国已超越美国和印度。
4. 自主智能体受青睐：多数管理者认为“自主AI工程代理”投资回报率最高，亚马逊AWS高管更公开强调“应将智能体视为不可或缺的同事”。

行业趋势与挑战：
尽管成本压力持续，85%的领导者预计未来三年工程师岗位数量将保持稳定或增长，表明AI短期内不会导致大规模裁员。然而，科技企业正重新评估人才需求——亚马逊、微软等公司近期的裁员均涉及大量软件开发工程师。

与此同时，近70%的工程团队计划通过战略招聘强化AI能力，但约三分之二企业仍禁止在面试中使用AI工具，仅不足30%的公司更新了评估体系以识别适配AI协作的人才。

破局之道：人机协同评估新模式
为应对评估体系滞后问题，Karat推出AI赋能的新一代人才评估平台，采用“人类专家+AI助手”协同面试模式，让候选人在实时协作中处理复杂多文件项目，以考察其真实工程能力与AI协作水平。

DocuSign首席技术官Sagnik Nandy指出：“人类判断力与AI能力的结合才是突破关键，而可靠评估这种协同能力的方法长期缺失。”

背景延伸：
Karat成立于2014年，2021年完成1.1亿美元C轮融资后估值达11亿美元，目前位列北美太平洋西北地区初创公司排行榜第15位。其报告折射出传统依赖代码工程师的科技行业，正步入人机深度协作的新发展阶段。

中文翻译：

那些将扎实基本功与快速涌现的AI工具运用能力相结合的软件工程师，正展现出前所未有的价值。尽管AI平均能将整体生产力提升34%，它也在拉大顶尖工程师与能力较弱者之间的差距。

这些洞察来自总部位于西雅图的技术人才评估初创公司Karat。该公司周三发布的《AI劳动力转型》报告详细阐述了人工智能工具如何改变软件开发模式，以及哪些从业者正受到这项技术的深刻影响。

报告结论基于Karat对美、印、中三国400位工程团队负责人的调研。与其同步发布的Karat NextGen平台，是一款专为人机协作时代设计的AI赋能人才评估解决方案，旨在精准衡量软件工程师的复合能力。

报告核心发现包括：

• 73%的领导者认为优秀工程师创造的价值至少是其总薪酬的3倍
• 59%的领导者指出在AI时代，能力薄弱的工程师产出净值趋近于零甚至为负
• 日常工作中最主要的AI应用场景是代码生成（83%）以及测试、质量保证与代码审查（61%）
• 多数领导者认为具备自主行动力的AI代理/自动化工程代理能带来最高投资回报率
• 尽管面临成本压力，85%的领导者预计未来三年工程师编制将保持稳定或增长，表明AI短期内不会导致大规模裁员
• 中国在AI应用准备度方面已超越美国和印度

科技企业与从业者仍在适应这场AI驱动的行业变革。这个历来依赖程序员构建维护数字基础设施的领域，正经历深刻重塑。

亚马逊10月在全球裁员1.4万名企业员工时，受影响的2303名华盛顿州员工（主要集中于西雅图与贝尔维尤）中，超过600名是软件开发工程师。这一趋势与微软今年初的裁员情况相呼应，表明企业正在AI编程工具崛起的背景下重新评估工程人才需求。

在上周拉斯维加斯的亚马逊re:Invent大会上，AWS高管科琳·奥布里突破性地提出：我们不应只讨论员工如何运用AI工具，而应将具备自主行动力的AI队友视为"与身边同事同等重要的存在"。

Karat报告指出，除解决问题、沟通协作、产品意识等基础能力外，工程师还需具备新型AI原生技能，包括：熟悉自主AI代理、运用AI编程、整合第三方AI接口、提示词工程、评估与规避AI相关风险等。

调研显示近70%的工程领导者计划通过战略性招聘增强AI能力。但矛盾的是，约三分之二企业仍禁止在面试中使用AI，仅不足30%的企业正在更新评估体系并培训面试官识别AI适配型人才。

Karat的NextGen评估平台首创"人类+AI"面试模式：候选人在AI助手辅助下处理复杂的多文件项目，同时与Karat专家面试官实时协作。面试官通过动态追问推理过程、方案权衡与判断依据，精准考察真实工程能力。

Karat客户DocuSign首席技术官萨格尼克·南迪在新闻稿中表示："虽然AI正在重塑工程领域，但真正的突破始终来自人类智慧与AI能力的融合。我们长期缺乏的，正是可靠衡量这种复合能力的方法。这种以人类为主导、AI原生化的面试，正是企业识别新开发模式下顶尖人才的理想解决方案。"

由莫·比亨德与杰夫·斯佩克特于2014年联合创立的Karat，在2021年完成1.1亿美元C轮融资后估值达11亿美元，成为西雅图估值最高的初创企业之一。该公司累计融资额已达1.516亿美元，目前在《GeekWire 200》太平洋西北地区初创企业排行榜中位列第15名。

英文来源：

Strong software engineers who combine their foundational skills with fluency in rapidly emerging AI tools are more valuable than ever. And while AI boosts overall productivity by 34% on average, its widening the gap between top engineers and those considered weaker.
Those are among the findings from Karat, the Seattle-based technical talent evaluation startup, which released its new “AI Workforce Transformation” report on Wednesday, detailing how artificial intelligence tools are changing the way software is developed and what types of workers are most impacted by the technology.
The findings come from Karat’s survey of 400 engineering leaders across the U.S., India, and China. And the report coincides with the release of Karat NextGen, an AI-enabled talent evaluation solution designed to evaluate software engineers in an era of increased human and AI collaboration.
Among the report highlights:

• 73% of leaders now believe a strong engineer is worth at least 3x their total compensation.
• 59% of leaders say weak engineers deliver net zero or negative value in the AI era.
• The top AI use cases for day-to-day work are code generation (83%), and testing, QA, and code review (61%).
• Agentic AI/autonomous engineering agents are highlighted by the majority of leaders as having the highest return on investment.
• Despite cost pressure, 85% of leaders expect engineering headcounts to stay flat or increase over the next three years, signaling that AI isn’t leading to massive job cuts in the near term.
• China is outpacing the U.S. and India in AI adoption and readiness.
  Tech companies and workers are still adjusting to the shifting landscape of an AI-fueled industry that has traditionally relied on coders to help build and maintain the backbone of digital platforms.
  When Amazon laid off 14,000 corporate employees from its global workforce in October, among the 2,303 impacted Washington state workers, mostly in Seattle and Bellevue, more than 600 were software development engineers.
  That trend mirrored layoffs at Microsoft earlier this year, as companies reassess their engineering needs amid the rise of AI-driven coding tools.
  At Amazon’s re:Invent event last week in Las Vegas, AWS executive Colleen Aubrey went beyond discussing how human employees will leverage AI tools, and said instead that it’s time to consider agentic teammates “as essential as the people sitting right next to you.”
  According to Karat’s report, beyond foundational skills such as problem-solving, communication, and product sense, engineers need to be assessed for new AI-native abilities, including familiarity with agentic AI; using AI for coding; integrating 3rd-party AI APIs; prompt engineering; and evaluating and mitigating AI-related risks.
  Karat’s report found that nearly 70% of engineering leaders plan to strengthen their AI capabilities through strategic hiring. Yet, almost two-thirds of companies still prohibit AI use in interviews, and less than 30% are updating assessments and training interviewers to identify AI-ready talent.
  The startup’s NextGen talent evaluation platform features a human + AI interview format where candidates tackle complex, multi-file projects with an integrated AI assistant while collaborating live with Karat’s expert interview engineers, who probe reasoning, trade-offs, and judgment in real time to reveal genuine engineering ability.
  Sagnik Nandy, CTO at DocuSign, a Karat customer, said in a news release that while AI is transforming engineering, “the real breakthroughs happen when human judgment and AI capabilities work together” and a way to measure that combination reliably is what’s been missing.
  “A human-led, AI-native interview is exactly the kind of solution organizations need to understand who can truly excel in this new model of development,” Nandy said.
  Founded in 2014 by Mo Bhende and Jeff Spector, Karat became one of Seattle’s highest-valued startups after it raised $110 million in a Series C round in 2021, which brought its total valuation at the time to $1.1 billion. Total funding to date is $151.6 million.
  Karat currently ranks No. 15 on the GeekWire 200, our list of the top startups in the Pacific Northwest.