"CREAATIF: Crafting Responsive Assessments of AI and Tech-Impacted Futures"

“CREAATIF：对人工智能和技术影响的未来进行响应式评估”

• 批准号: AH/Z505584/1
• 负责人: David Leslie
• 金额: $ 28.04万
• 依托单位: Queen Mary University of London
• 依托单位国家: 英国
• 项目类别: Research Grant
• 财政年份: 2024
• 资助国家: 英国
• 起止时间: 2024 至 无数据
• 项目状态: 未结题
• 来源: https://gtr.ukri.org/projects?ref=AH%2FZ505584%2F1
• 关键词: CREAATIF; Crafting; Responsive; Assessments; AI

Generative AI (GenAI) burst into the popular imagination in late 2022 with the release of ChatGPT - a chat agent that has proven not only to be very popular but also signifies a major leap forward in technological capabilities. ChatGPT is just one of several GenAI technologies that has entered the scene in recent years; others can generate (or alter) video, images, music, dialogue, and computer code. These developments have the potential to change the nature of work for many, including for workers previously deemed immune to direct competition from technology.There is urgency to studying the impact of these tools in the specific context of creative work, in which technologically-mediated worker precarity is an ongoing but increasingly acute concern. Worker resistance, as exemplified by recent industrial action by the Writers Guild of America, highlights that impacts go beyond 'displacement' of or access to work, and can impact established notions of authorship while also affecting worker discretion and dignity. The creative sector is at the coalface of the GenAI transformation in which emerging technologies potentially devalue labour materially (wages) and socially (recognition of contribution).Our understanding of the transformative effects of GenAI in creative work is still emerging but present; the experience and perspective of those whose lives and livelihoods are increasingly threatened by these new technologies have not been properly factored into AI policy planning and change. What is needed is to bring these perspectives into view where they can influence labour policy in the area of data-driven technologies. To achieve this requires the building of new architectures that bridge this divide between experience and application and which promote involvement by building on the strength of UK labour law, comparable historical precedents like Scandinavian participatory design, and recent turns toward participatory algorithmic impact assessments.Algorithmic impact assessments hold promise as accountability tools that can surface core concerns about the effects of data-driven technologies while pointing towards governance strategies for mitigating those concerns. Where impact assessments are designed to foreground the voices of people affected by emerging technologies, they can also serve as frameworks for surfacing and crystalising perspectives that reflect the lived experience of technology-mediated lives, which in turn can be channelled into policy guidance.In this project, we bring together two leading and relevant methods of impact assessment: the Human Rights, Democracy, and the Rule of Law Assurance Framework for AI Systems (HUDERIA), and the Good Work Algorithmic Impact Assessment (GWAIA). The GWAIA has been selected as a focal point because of its specific application to questions of worker dignity. Its current design is relevant to algorithmic management tools within a 'conventional' employment context. We will cross-reference this with insights from HUDERIA, which brings specific insights with regards to structuring accountability in the relationship between individuals and technology producers, public and private. A central feature these tools share is the participatory engagement model of surfacing, assessing, and mitigating individual and collective risks to workers by drawing on the experiences, testimony, and ideas of workers themselves.

2022 年底，随着 ChatGPT 的发布，生成式人工智能 (GenAI) 突然进入了大众的想象。ChatGPT 是一种聊天代理，事实证明它不仅非常受欢迎，而且标志着技术能力的重大飞跃。 ChatGPT 只是近年来进入市场的几种 GenAI 技术之一；其他人可以生成（或更改）视频、图像、音乐、对话和计算机代码。这些发展有可能改变许多人的工作性质，包括以前被认为不受技术直接竞争影响的工人。迫切需要研究这些工具在创造性工作的特定背景下的影响，其中技术介导的工人不稳定是一个持续存在但日益严重的问题。美国作家协会最近采取的罢工行动表明，工人的抵抗运动所产生的影响不仅仅是“取代”工作或获得工作机会，还可能影响既定的作者身份观念，同时也影响工人的自由裁量权和尊严。创意行业处于 GenAI 转型的核心，其中新兴技术可能会在物质（工资）和社会（贡献认可）方面降低劳动力的价值。我们对 GenAI 在创意工作中的变革影响的理解仍在萌芽中，但已经存在；那些生活和生计日益受到这些新技术威胁的人的经验和观点尚未适当地纳入人工智能政策规划和变革中。我们需要将这些观点纳入视野，从而影响数据驱动技术领域的劳工政策。为了实现这一目标，需要建立新的架构，弥合经验和应用之间的鸿沟，并通过建立在英国劳动法、斯堪的纳维亚参与式设计等类似历史先例以及最近转向参与式算法影响评估的基础上来促进参与。算法影响评估有望成为问责工具，可以揭示对数据驱动技术影响的核心关切，同时指出缓解这些关切的治理战略。如果影响评估旨在突出受新兴技术影响的人们的声音，那么它们也可以作为浮现和具体化观点的框架，反映技术介导的生活的生活经验，而这些经验又可以转化为政策指导。在该项目中，我们汇集了两种领先且相关的影响评估方法：人工智能系统的人权、民主和法治保障框架（HUDERIA）以及良好工作算法影响评估（GWAIA）。 GWAIA 因其在工人尊严问题上的具体应用而被选为焦点。其当前的设计与“传统”就业环境中的算法管理工具相关。我们将把这一点与 HUDERIA 的见解进行交叉引用，HUDERIA 带来了关于构建个人与技术生产者、公共和私营部门之间关系的责任的具体见解。这些工具共有的一个核心特征是参与式参与模型，通过利用工人本身的经验、证词和想法来呈现、评估和减轻工人面临的个人和集体风险。