商业自动化决策的算法解释权研究

The research on right to explanation of algorithm in auto decision-making

自动化决策广泛应用,算法的评分、排序决定和影响着相对人贷款、就业、信誉等诸多权益。但当算法作出不利甚至错误决策时,相对人无从知晓原因更遑论改正其错误决策,严重缺乏应有的救济。通过梳理现有法律资源发现,传统制度在自动化决策场景下均不敷适用,无法为相对人提供算法决策的解释。法律应配置独立的算法解释权,用以衡平自动化决策使用者与相对人不对称的权力关系,作为合同制度在人工智能时代的因应性变革。算法解释权的理论正当性亦可在贯彻意思自治原则、矫正信息不对称、分配风险负担等层面得到充分证成。算法解释权的内在构造具有独特的内容要求和双层架构,其具体制度应在兼顾效率和公平的原则下进行设计。

The increasing prevalence of algorithm in decision - making processes raises very important and urgent ques- tions regarding how law protects rights of people, such as eligibility of loan, employment and reputation. The paper assumes that there ought to be a right to explanation due to existing law cannot provide proper protection. The fight to explanation is an adjustment of law complying with the power distance of contract in the era of AI. And yet, the traditional civil law theory justifies such a right as moraL/ethical foundation of a right to explanation. It should be designed as a right with two - tier structure. Also, a proper scope and due process should be considered on compatibility of efficiency and equity.