Artificial intelligence systems connected to sensor-laden devices are becoming pervasive, which has significant implications for a range of AI risks, including to privacy, the environment, autonomy, and more. There is therefore a growing need for increased accountability around the responsible development and deployment of these technologies. In this paper, we provide a comprehensive analysis of the evolution of sensors, the risks they pose by virtue of their material existence in the world, and the impacts of ubiquitous sensing and on-device AI. We propose incorporating sensors into risk management frameworks and call for more responsible sensor and system design paradigms that address risks of such systems. To do so, we trace the evolution of sensors from analog devices to intelligent, networked systems capable of real-time data analysis and decision-making at the extreme edge of the network. We show that the proliferation of sensors is driven by calculative models that prioritize data collection and cost reduction and produce risks that emerge around privacy, surveillance, waste, and power dynamics. We then analyze these risks, highlighting issues of validity, safety, security, accountability, interpretability, and bias. We surface sensor-related risks not commonly captured in existing approaches to AI risk management, using a materiality lens that reveals how physical sensor properties shape data and algorithmic models. We conclude by advocating for increased attention to the materiality of algorithmic systems, and of on-device AI sensors in particular, and highlight the need for development of a responsible sensor design paradigm that empowers users and communities and leads to a future of increased fairness, accountability and transparency.

翻译：人工智能系统与配备传感器的设备相连接正变得无处不在，这对隐私、环境、自主性等一系列AI风险产生了重大影响。因此，围绕这些技术的负责任开发与部署，日益需要加强问责机制。本文对传感器的演变、其因物质存在于世界所带来的风险，以及无处不在的感知与设备端AI的影响进行了全面分析。我们提议将传感器纳入风险管理框架，并呼吁采用更负责任的传感器与系统设计范式，以应对此类系统带来的风险。为此，我们追溯了传感器从模拟设备演变为具备网络功能、能够在网络极边缘进行实时数据分析与决策的智能系统的历程。研究表明，传感器的激增是由优先考虑数据收集与成本降低的计算模型所驱动的，并产生了围绕隐私、监控、废弃物和权力动态的风险。随后，我们分析了这些风险，重点突出了有效性、安全性、安保、问责性、可解释性和偏见等问题。我们揭示了现有AI风险管理方法中通常未涵盖的与传感器相关的风险，采用物质性视角来阐明物理传感器特性如何塑造数据与算法模型。最后，我们主张应更多关注算法系统的物质性，尤其是设备端AI传感器的物质性，并强调需要发展一种负责任的传感器设计范式，以赋予用户和社区权力，从而迈向一个更加公平、可问责和透明的未来。