Innovation and Regulation: Attaining and Maintaining Transparency and Accountability in Automated Decision-Making

在社会中，自动决策（ADM）（没有人类参与的决策）正在增加。零售业使用ADM通过推荐系统（RS）为其客户配置和个性化内容。RS获取组织所知道或感知的关于个人或群体的信息，并应用一种算法来预测和推荐未来的购买。RS的感知好处是，个人看到的内容与他们更相关，对组织来说是具有成本效益的，并增加了销售。随着2015年至2018年在线零售增长超过30%，2020年4月电子商务销售额增长了209%，了解消费者的兴趣和习惯对于组织优化生存机会尤为重要。相比之下，RS对个人和社会提出了许多挑战，主要是围绕隐私。当他们在幕后工作时，许多人可能不知道RS，研究继续了解更多关于伦理挑战及其对社会的真实的影响。这项研究集中在RS与一般数据保护条例的政策方面。我们试图了解组织如何根据GDPR解释这项技术的使用，特别是一般分析活动，基于分析的决策和自动化个人决策（包括分析）之间的应用和区别。GDPR是一个关键的起点，因为它的总体设计是为个人带来更明确的权利，并为组织提供更多的义务和后果。虽然GDPR没有定义透明度，但我们的意思是对所有人“清晰、开放和诚实”，而问责制则是指“承担责任”和“展示合规性”。为了了解有关透明度和问责制要素的当前立场，我们将考虑组织的公开信息，并利用我们的个人权利。与组织、个人和监管机构进行的半结构化访谈也将提供广泛的利益相关者对RS透明度和问责制的反馈，包括目前的状况以及他们认为可以改善的问题。总的来说，该项目旨在了解组织，个人和社会之间的平衡，并提出适当的建议，为社会利益为斯普斯卡共和国ADM的未来监管提供信息。

Within society, Automated Decision-Making (ADM) (a decision with no human involvement) is increasing. The retail sector uses ADM to profile and personalise content to its customers through Recommender Systems (RS). RS takes the information the organisation knows or perceives about an individual or group of people and applies an algorithm that predicts and recommends future purchases. RS's perceived benefits are that the individual sees content that is more relevant to them and is cost-effective to organisations, and increases sales. As online retail grew over 30% between 2015-2018, and April 2020 saw e-commerce sales grow by 209%, keeping abreast of consumer interests and habits is particularly important for organisations to optimise survival chances. In contrast, RS present many challenges for individuals and society, primarily around privacy. As they work behind the scenes, many individuals may be unaware of RS, and studies continue to understand more about the ethical challenges and their real impact upon society. This research concentrates on the policy aspects of RS with the General Data Protection Regulation. We seek to understand how organisations interpret this technology's use in line with the GDPR, particularly the application and distinction between general profiling activity, decision-making based upon profiling and automated individual decision-making, including profiling. The GDPR is a key starting point as its general design is to bring clearer rights for individuals and provide more obligations and consequences for organisations. While the GDPR does not define transparency, we mean being 'clear, open and honest' to all and by accountability, we mean 'taking responsibility' and 'demonstrating compliance'.To understand the current position regarding elements of transparency and accountability, we will consider the organisation's publicly available information and make use of our individual rights. Semi-structured interviews with organisations, individuals and regulators will also provide wide stakeholder feedback on RS transparency and accountability, both on the present position and what they feel may improve matters. Overall, the project aims to understand the balance between organisations, individuals and society and make appropriate recommendations to inform the future regulation of ADM in RS for societal good.