重構黑箱社會 / Restructuring a Black Box Society

「演算法時代下的規範與程序正義」(Regulation and Due Process in the Era of the Algorithm)

In an era where big data prevails, tech firms and financial sectors poured themselves into the maniac of mass data collection. No rigid reason needed -- quantity is all that matters when the miraculous algorithms came to turn data into digital gold. However, as automatic systems took over from credit reporting, social media timeline arrangement, search engine result ranking to the extent of spotting anti-terrorism suspects, the tech industry found themselves impossible to shrug off responsibilities by claiming their results “machine learning magic.”

Despite the threat of mass data collection and “black box” algorithms sitting at the core of governmental and commercial activities, regulators are not underequipped to solve the problem. The talk would address human factors affecting different aspects of machine learning algorithms, the relevant social issues, and potential regulation directions. (https://coscup.org/2020/zh-TW/agenda/7ALFQQ)

Remote Talk @ COSCUP 2020, National Taiwan University of Science & Technology (NTUST), Taipei, Taiwan.

Transcript

1. 重構黑箱社會 演算法時代下的規範與程序正義 RSChiang @ COSCUP 2020 Slide CC BY-SA 4.0
2. 「人工智慧」 ≈ 自動化決策系統 輸入資料→計算→輸出決策
3. 輸入資料→計算→產生決策 資料從何而來？ 由誰輸入？ 由誰將 判斷基準 程式化？ 是否符合公平期待？
4. 資料蒐集→處理→利用→產生結果 以 ML 為出發點思考，環節無非為
5. 資料蒐集→處理→利用→產生結果 誰？目的為何？ 範圍是否有所偏誤？ 非量化資料如何判斷？ 標記、勾稽、整合的限制？ 販售：資料盤商 (data broker) 決策透明度？ 是否符合公平期待？
6. 機器決策看似排除了人為偏誤， 實則將左右決策的人為因素隱藏於黑箱中。 電腦揀的哦
7. 規則 (rule) vs. 標準 (standard)
8. 重建自動決策的信賴 正當程序 (due process) 作為解方
9. Restructuring a Black Box Society: Regulation and Due Process in the Era of the Algorithm RSChiang (Poren Chiang) @ COSCUP 2020 E-mail: hi@poren.tw 參考資料 Frank Pasquale, THE BLACK BOX SOCIETY (2015). Danielle Keats Citron, Technological Due Process, 85 WASH. U. L. REV. 1249 (2008). Slide released under CC BY-SA 4.0.