行政治理中的情感计算技术应用及其风险预防

The Application of Affective Computing Technology in Administrative Governance and Its Risk Prevention

情感计算系基于目前的人工智能等技术,对人类此前未用科学描述过的情绪、感情等“主观”内容作出相对理性、精确分析乃至预测的技术。该技术的应用毫无疑问让行政治理迭代、高效,突出表现在催生跨物理空间的新治理场景、实现高精度的治理对象侧写、转型智慧型的行政治理模式等。但是,行政治理的无序利用将会产生至少三个层面的风险:因技术限度造成了无法产生对于治理来说所需要的精确性、因技术本身的要求产生对信息权等基本人权侵犯的高度风险以及因技术的不当应用造成的公权力扩张之风险。有鉴于此,应当基于法治精神实现情感计算技术之于行政治理应用中的内部和外部规制。内部规制上,将个人信息权益等作为保障核心并融入伦理原则形成情感计算的行动方案;外部规制上,搭建风险预防框架并确立规范、价值上的引导路径。

Affective computing has a broad application space in the process of establishing digital administrative governance:firstly,constructing a virtual scene for service administrative governance,so that administrative governance activities can directly penetrate into the emotional or affective dimensions of subjects'personality;secondly,adapting to the upgrading of administrative law enforcement technology tools,so that subjects of digital administrative governance can integrate and analyze multi-modal information and maximally reproduce the inner emotional tendency and position of law enforcement objects;and thirdly,promoting the intelligent transformation of administrative governance,so that administrative governance actions become more concrete and touch people's heart and soul more directly.However,because of the lag of legal research and the complexity of affective computing,it is difficult to regulate and restrict affective computing,leading to a number of risks.Firstly,it is difficult for affective computing to fully guarantee the accuracy of governance results.Secondly,the intrinsic characteristics of affective information make it possible to violate the right to privacy.And thirdly,the application of affective computing can easily lead to the over-expansion of algorithmic public power.In view of the above risks,the following paths need to be constructed to prevent the risks:firstly,to protect information rights and interests by regulating emotional data,that is,to ensure that the use of emotional data complies with the current legal requirements,improve emotional data protection measures,and establish a data internal legitimacy review mechanism;secondly,to embed the logic of emotional computing in the human-machine emotional synergy of the training dimension,that is,to not only improve the transparency and interpretability of affective computing algorithms and establish effective mechanisms for the collection and use of affective data but also promote the establishment of a human-computer interaction system for affective computing with the participation of multiple parties;thirdly,to set standards and limits for the application of affective computing in digital administrative governance,that is,administrative organs should set up standards for affective computing through administrative legislation and policy formulation,and clarify the purpose and limits of affective computing application;fourthly,to delineate the value bottom line and theoretical forbidden zone for the application of affective computing,that is,to determine the principles of value bottom line and ethical forbidden zone,strengthen the ethical assessment and legitimacy review of affective computing,and establish the application supervision mechanism;and fifthly,to add rationality and legitimacy requirements for the iteration of affective computing,that is,to enhance ethical judgment capacity through data collection and learning and through ethical modeling and rule updating and to improve ethical judgment by using feedback from human experts.