自动化行政合法性证成的基本逻辑

The Basic Logic of Legitimacy Certification of Automated Administration

自动化行政是行政机关借助自动化设备作出行政行为的情形,其本质是以机器学习算法为核心的人工智能技术对行政活动的参与。由于机器学习算法可以自我学习、具有黑箱特性且可能缩短甚至直接绕过人类来思考,所以它在直接或辅助行政机关作出行政决定时会对行政法治统合下的课责和监督机制形成一定挑战。即便如此,自动化行政本身的合法性仍然可以被有效证成。一方面,机器学习算法对行政决定作出过程的参与,不会在根本上冲击最能彰显行政法治的“依法行政原则”和“正当行政程序原则”,这使得自动化行政具备了形式上的合法性。另一方面,机器学习算法还有助于实现“最佳行政”之目标,从而增进国家的良善治理,这可以在更深层次证立自动化行政的实质合法性。

Automated administration means that administrative organs make decisions with the help of automatic equipment. Automated administration's essence is the participation of artificial intelligence, which core part refers to machine learning algorithm, in administrative activities. Machine learning algorithms can be self-study, has the characteristics of black box, and may shorten or even bypass the thinking of human beings, so when the machine learning algorithms directly or assist administrative organs to make decisions, it will pose a certain challenge to the class responsibility and supervision mechanism under administrative rule of law. Even so, the legitimacy of automated administration can still be effectively proved.On the one hand, the participation of machine learning in the process of administrative decisions-making will not fundamentally impact the "principle of administration according to law" and "the principle of due process of administration",which makes the automatic administration meet the requirements of formal legitimacy. On the other hand, machine learning algorithm also helps to realize "better administration", so as to promote the good governance of the country, which will further prove the substantive legitimacy of automatic administration.