AI Workers’ Inquiry 2026

2026年人工智能从业者调查报告

The UTAW Tech Workers’ Inquiry outlines how generative AI is changing work across the tech sector, written with the perspectives and experiences of workers who build, deploy, manage, evaluate and use this technology. UTAW（英国科技工会）的这份科技从业者调查报告概述了生成式人工智能如何改变整个科技行业的工作方式。报告基于那些负责构建、部署、管理、评估及使用该技术的从业者的视角与经验撰写而成。

Executive Summary

执行摘要

AI is widely used to describe a disparate, and in many cases contradictory, set of technologies. While the term defies consensus, a clear picture of its impacts is emerging. Our participants described it as a largely negative force reorganising their work: increasing output expectations, changing skills requirements, expanding monitoring, degrading expertise, shifting accountability, and creating divisions between colleagues. AI applications rarely remove work; they redistribute and intensify it. “人工智能”一词被广泛用于描述一套迥异、甚至在许多情况下相互矛盾的技术。尽管该术语尚无统一定义，但其影响已逐渐清晰。参与调查的从业者将其描述为一种重塑工作的负面力量：它提高了产出预期、改变了技能需求、扩大了监控范围、削弱了专业性、转移了问责制，并在同事之间制造了隔阂。人工智能应用很少能真正减少工作量；它们只是在重新分配并加剧工作强度。

Workers stressed that positive uses depend on their ability to decide for themselves how to engage with AI. The central problem is not simply whether AI works well, but how it is introduced, controlled, measured and used by employers. Put simply, it is useful only when one gets to choose if and how to use them. 从业者强调，人工智能的积极应用取决于他们能否自主决定如何与人工智能互动。核心问题不仅仅在于人工智能本身是否好用，而在于雇主如何引入、控制、衡量和使用它。简而言之，只有当人们能够选择是否使用以及如何使用这些工具时，它们才是有益的。

The problems identified are not derived from any technical limits of today’s models. Poor quality outputs, hallucinations and weak code do have an impact, but they are only part of the picture. This is because employers use AI to intensify work, rationalise redundancies, eliminate junior hiring, increase surveillance, deskill and downskill workers, shift risk onto employees and weaken their agency. Even as models improve, the underlying workplace issues will remain if workers do not gain enforceable rights over how AI is used, at work and in society at large. 所发现的问题并非源于当前模型的技术局限。低质量的输出、幻觉和薄弱的代码确实有影响，但这只是问题的一部分。根本原因在于，雇主利用人工智能来加剧工作强度、合理化裁员、取消初级岗位招聘、加强监控、使员工技能退化、将风险转嫁给员工并削弱其自主权。即便模型不断改进，如果从业者无法在工作及社会层面获得关于人工智能使用方式的可执行权利，潜在的职场问题依然会存在。

Tech workers are deeply concerned over AI’s harmful impacts on the environment, in military and state surveillance contexts, and on mental health. This is why the Inquiry’s recommendations focus less on functionality and more on who controls and benefits from it. 科技从业者对人工智能在环境、军事与国家监控领域以及心理健康方面产生的负面影响深感担忧。因此，本调查报告的建议重点不在于功能性，而在于谁在控制这些技术以及谁从中获益。

Policy Proposals

政策建议

These proposals expand upon the carried composite motion at TUC Congress 2026 titled “Artificial Intelligence at work,” moved by the CWU, seconded by Aegis and supported by: UNISON; National Education Union; USDAW. 这些建议扩展了2026年英国工会联盟（TUC）大会上通过的题为“工作中的人工智能”的综合动议。该动议由英国通信工会（CWU）提出，Aegis工会附议，并得到了UNISON、英国国家教育工会（NEU）及USDAW的支持。

New regulatory authority on AI at work

建立工作场所人工智能监管机构

We need a regulatory framework with teeth to hold employers and technology companies accountable for their use of AI, along the following lines: 我们需要一个强有力的监管框架，以确保雇主和科技公司对其人工智能的使用负责，具体方向如下：

• Establish a new statutory authority for AI at work, with powers to investigate, enforce and sanction employers and technology providers. 建立一个新的工作场所人工智能法定监管机构，拥有调查、执法以及制裁雇主和技术提供商的权力。
• The authority should cover all AI systems, including genAI tools, automated decision systems, monitoring tools, recruitment systems, productivity systems and performance management tools. 该机构应涵盖所有人工智能系统，包括生成式AI工具、自动化决策系统、监控工具、招聘系统、生产力系统及绩效管理工具。
• Workers and their unions should be able to report AI-related harms directly to the authority, including unsafe deployment, discriminatory outcomes, excessive monitoring, misleading productivity claims, work intensification, deskilling and downskilling, job displacement, and ethical and environmental concerns. 从业者及其工会应能直接向该机构报告与人工智能相关的危害，包括不安全的部署、歧视性结果、过度监控、误导性的生产力声明、工作强度增加、技能退化、岗位流失以及伦理和环境问题。

Skills, training and junior pathways

技能、培训与初级人才培养路径

Skills and development must be prioritised where AI is deployed to eliminate AI-induced deskilling and downskilling: 在部署人工智能的地方，必须优先考虑技能与发展，以消除人工智能导致的技能退化：

• Protect junior roles and learning pathways by ensuring AI does not replace training, mentoring, supervision or foundational skill development. 保护初级岗位和学习路径，确保人工智能不会取代培训、指导、监督或基础技能培养。
• Employers should not use AI as a substitute for subject-matter expertise, proper staffing or professional development. 雇主不应将人工智能作为专业知识、合理人员配置或职业发展的替代品。
• Workers should have protected time to learn, practise, review and develop skills without being forced to rely on AI tools. 从业者应拥有受保障的时间来学习、实践、复盘和发展技能，而不应被迫依赖人工智能工具。

Right to refuse, question and challenge

拒绝、质疑与挑战的权利

Workers must retain the right to refuse and be protected from retaliation in doing so: 从业者必须保留拒绝的权利，并在此过程中免受报复：

• Codify a right to choose whether and how to use genAI tools where there are professional, ethical, environmental, equality, accessibility, safety, religious, mental health or quality concerns. 在涉及专业、伦理、环境、平等、无障碍、安全、宗教、心理健康或质量问题时，应通过法律明确从业者选择是否使用以及如何使用生成式AI工具的权利。
• Workers should be protected from detriment for refusing, questioning, criticising or reporting concerns about AI use or development. 从业者因拒绝、质疑、批评或报告有关人工智能使用或开发的问题而免受不利待遇。
• Workers should have the right to human review where AI is used in decisions affecting their work, pay, performance, promotion, discipline, hiring or redundancy. 当人工智能被用于影响工作、薪酬、绩效、晋升、纪律、招聘或裁员的决策时，从业者应拥有要求人工复核的权利。