算法规制如何场景化

How Can Algorithm Regulation Be Contextualized

将算法纳入法治化的轨道,需要根据不同的场景设置差异化的规范路径。以高风险场景的算法产品使用效果实验启示,通过算法的解释可以提升用户的理解程度,并且可视化解释的理解效果较公式法更好;算法信任的机制构建,则更多取决于场景的信任基础、人机判断的一致性等因素。实证研究对算法规制如何场景化提供了诸多的启示:一是需要考虑不同场景用户特征的差异,根据用户特点设置权利义务;二是应强调场景中算法以及解释的用途差异,结合考虑相关解释技术的可行性;三是应当考虑不同场景的算法信任基础,算法引入不同场景有不同难度,不同场景应当设置不同的人机分工关系和算法的任务范围。

Bringing algorithms into the orbit of the rule of law requires setting differentiated regulatory paths based on different scenarios.Insights from the use-effect experiments of algorithm products in high-risk scenarios indicate that user understanding can be enhanced through algorithmic explanations,with visual explanations yielding better comprehension than formulaic methods.The construction of algorithm trust mechanisms depends more on the trust foundation of the scenario,the consistency of human-machine judgments and other factors.Empirical research provides many insights into how algorithm regulation can be contextualized:first,it is necessary to consider the differences in user characteristics in different scenarios and establish rights and obligations based on user traits.Second,emphasis should be placed on the differences in the purposes of algorithms and explanations in different scenarios,considering the feasibility of relevant explanation techniques.Third,different scenarios should consider the trust foundation of algorithms.Introducing algorithms into different scenarios poses varying degrees of difficulty,requiring different human-machine division of labor relationships and algorithm's tasks in different scenarios.