人工智慧之治理──以機器學習公平性為中心

Artificial intelligence governance in the aspect of fairness

人工智慧為科技領域近兩年最受歡迎之技術，各國政府、企業與相關組織莫不力求推動人工智慧之研究、發展及應用。為因應人工智慧對人類社會與經濟可能帶來之衝擊，國際主要國家與國際組織均提出相關監管措施與發展政策，而不論是草案、政策或指引之內容，不約而同均提及人工智慧應具備「公平性」，可見未來相關規範可能將落實於實體法內。然而，法規可以抽象的要求人工智慧應達到公平性，機器卻需要透過明確之定義始得達成，本文擬蒐集整理美國近期法制政策與國際相關文獻，探討現行如何落實人工智慧公平性之要求，以及相關方法與法學上公平性概念之差異，並提出可行之建議以供未來人工智慧相關法律遵循與主管機關進行公平性審查之參考。

Fairness is becoming one of the most popular topics in Machine Learning in recent years. The research community has invested a large amount of effort in this field. Why we should care about it? The main motivation is that it is highly related to our own benefit. We are at an age where many things have become or are becoming automated by machine learning systems. Artificial Intelligence is good but can be us incorrectly. The data provided by human may be highly-biased. Therefore, most of country revealed guidance, rules, even law which trying to mitigate the bias in Machine, but the problem is, are those solution correspond to the concept of principle of equality in law? The purpose of this article is to study relevant fairness solution in the field of machine learning, and compare it to the concept of principle of equality in our justice system.