人工智慧科研倫理與風險之基本認識

AI Literacy: Ethics and Risks in Research and Development

10.6464/TJSSTM.202310_(37).0004

甘偵蓉(Zhen-Rong GAN)

人工智慧 ； 倫理與風險 ； 機器學習 ； 模型 ； 國科會科研發展指引 ； Artificial intelligence (AI) ； ethical considerations ； risk assessment ； machine learning technologies ； AI Technology R&D Guidelines

科技醫療與社會

37期（2023 / 10 / 01）

167 - 219

繁體中文;英文

生活周遭越來越多事物使用人工智慧（artificial intelligence，以下簡稱AI）技術或相關產品，但人們對於AI科技的看法，往往不是太樂觀天真，就是太悲觀或帶有刻板印象。本文旨在整理與概述AI的倫理議題與風險有哪些面向，內容上主要分成2個部分。第一部分先簡介目前AI系統最常使用的機器學習技術，以及可能涉及的倫理議題與風險。第二部分則以國科會2019年公布的「人工智慧科研發展指引」為例，針對文件中所提及的3項核心價值以及8項行為指引，逐一說明每項理解方式及相關案例；其中8項行為指引又被區分為：任何新興科研都應該遵守的、有關AI技術特性、以及有關AI應用等3個類別。期待本文有助於促進非AI技術人員對於AI倫理與風險的認識。

Artificial intelligence (AI) and its related products are increasingly being integrated into our daily lives. However, people's perceptions of AI, often influenced by media hype or sci-fi films, range from excessively optimistic and naive to overly pessimistic and even stereotypical. These polarized viewpoints ultimately hinder the development of trustworthy AI, which by definition should be legal, ethically compliant, and technically robust. This paper aims to help non-AI technical professionals understand AI-related ethical issues and risks by dividing the discussion into two parts. The first part introduces the most commonly used machine learning technologies in AI systems and discusses potential ethical issues and risks. The second delves into the "AI Technology R&D Guidelines," published by the National Science and Technology Council in Taiwan in 2019. Interpretations of the document's three core values and eight guidelines are provided. The eight guidelines are categorized based on their relation to emerging technology R&D, AI technology characteristics, and AI application outcomes. Real-world case studies illustrate each guideline, while also highlighting their interplay within each category. In conclusion, this paper offers a foundational understanding of the ethics and risks involved in AI research and development.

1. Guidance on AI and data protection (n. d.). https://ico.org.uk/for-organisations/guide-to-data-protection/key-dp-themes/guidance-on-ai-and-data-protection/ (Retrieved April 7, 2023).
2. The AI Act (2021). The Artificial Intelligence Act. https://artificialintelligenceact.eu/ (Retrieved April 7, 2023).
3. General Data Protection Regulation (n. d.). https://gdpr.eu/ (Retrieved April 7, 2023)
4. California Privacy Rights Act (2020). https://thecpra.org/ (Retrieved April 7, 2023).
5. Arrieta, Alejandro Barredo,Díaz-Rodríguez, Natalia,Ser, Javier Del,Bennetot, Adrien,Tabik, Siham,Barbado, Alberto,Garcia, Salvador,Gil-Lopez, Sergio,Molina, Daniel,Benjamins, Richard,Chatila, Raja,Herrera, Francisco(2020).Explainable Artificial Intelligence (XAI): Concepts, Taxonomies, Opportunities and Challenges toward Responsible AI.Information Fusion,58,82-115.
6. Bender, Emily M.,Gebru, Timnit,McMillan-Major, Angelina,Shmitchell, Shmargaret(2021).On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?.Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency
7. Buolamwini, Joy,Gebru, Timnit(2018).Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification.Proceedings of Machine Learning Research,81,77-91.
8. Cabinet Secretariat of Japan (2019). Social Principles of Human Centric AI. https://www.cas.go.jp/jp/seisaku/jinkouchinou/pdf/humancentricai.pdf (Retrieved: July 4, 2023).
9. Chatila, Raja, Kay Firth-Butterfield and John C. Havens (2018). Ethically Aligned Design: A Vision for Prioritizing Human Well-being with Autonomous and Intelligent Systems (Version 2). University of Southern California Los Angeles. https://apps.dtic.mil/sti/pdfs/AD1170922.pdf (Retrieved July 28, 2023).
10. D’ignazio, Catherine,Klein, Lauren F.(2020).Data Feminism.Cambridge, MA:MIT Press.
11. David, Scott L, Jean-Claude Goldenstein, Ali G. Hessami, Patricia Shaw, Eleanor (Nell) Watson and Gerlinde Weger (2020). IEEE Use Case-Criteria for Addressing Ethical Challenges in Transparency, Accountability, and Privacy of Contract Tracing-Draft (Call for Global Consultation). IEEE Standards Association - The Ethics Certification Program for Autonomous and Intelligent Systems (ECPAIS). https://engagestandards.ieee.org/rs/211-FYL-955/images/ECPAIS_USECASE_10132020_DRAFT.pdf (Retrieved July 28, 2023).
12. European Commission (2021). Proposal for a Regulation Laying Down Harmonized Rules on Artificial Intelligence. https://digital-strategy.ec.europa.eu/en/library/proposal-regulation-laying-down-harmonised-rules-artificial-intelligence (Retrieved: July 4, 2023).
13. European Commission, Directorate-General for Communications Networks, Content and Technology(2019).Ethics Guidelines for Trustworthy AI.Publications Office.
14. Eykholt, Kevin,Evtimov, Ivan,Fernandes, Earlence,Li, Bo,Rahmati, Amir,Xiao, Chaowei,Prakash, Atul,Kohno, Tadayoshi,Song, Dawn(2018).Robust Physical-world Attacks on Deep Learning Visual Classification.2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition
15. Fjeld, Jessica,Achten, Nele,Hilligoss, Hannah,Nagy, Adam,Srikumar, Madhulika(2020).Berkman Klein Center Research PublicationBerkman Klein Center Research Publication,Berkman Klein Center Research.
16. Floridi, Luciano(2019).Translating Principles into Practices of Digital Ethics: Five Risks of being Unethical.Philosophy and Technology,32,185-193.
17. Floridi, Luciano,Holweg, Matthias,Taddeo, Mariarosaria,Silva, Javier Amaya,Mökander, Jakob,Wen, Yuni(2022).,未出版
18. Frey, Carl Benedikt,許恬寧（譯）(2020).技術陷阱：從工業革命到AI 時代,技術創新下的資本、勞動力與權力.臺北:八旗文化.
19. Gan, Zhen-Rong,Israel, Mark(2020).Transnational Policy Migration, Interdisciplinary Policy Transfer and Decolonization: Tracing the Patterns of Research Ethics Regulation in Taiwan.Developing World Bioethics,20(1),5-15.
20. Hagendorff, Thilo(2020).The Ethics of AI Ethics: An Evaluation of Guidelines.Minds & Machines,30,99-120.
21. Hao, Karen (2019). Training a Single AI Model Can Emit as Much Carbon as Five Cars in Their Lifetimes. MIT Technology Review. https://www.technologyreview.com/2019/06/06/239031/training-a-single-ai-model-can-emit-as-much-carbon-as-five-cars-in-their-lifetimes/ (Retrieved July 28, 2023).
22. Heaven, Will Douglas (2022). Why Meta’s Latest Large Language Model Survived Only Three Days Online. MIT Technology Review. https://www.technologyreview.com/2022/11/18/1063487/meta-large-language-model-ai-only-survived-three-days-gpt-3-science (Retrieved April 7, 2023).
23. Hellström, Thomas,Dignum, Virginia,Bensch, Suna(2020).,未出版
24. Hind, Michael,Wei, Dennis,Campbell, Murray,Codella, Noel C. F.,Dhurandhar, Amit,Mojsilović, Aleksandra,Ramamurthy, Karthikeyan Natesan,Varshney, Kush R.(2019).TED: Teaching AI to Explain Its Decisions.Proceedings of the 2019 AAAI/ACM Conference on AI, Ethics, and Society
25. HLEG AI (High-Level Expert Group on Artificial Intelligence) (2019). Ethics Guidelines for Trustworthy AI. https://digital-strategy.ec.europa.eu/en/library/ethics-guidelines-trustworthy-ai (Retrieved April 7, 2023).
26. Hussain, Waheed (2018). The Common Good. In The Stanford Encyclopedia of Philosophy, edited by Edward N. Zalta. https://plato.stanford.edu/archives/spr2018/entries/common-good/ (Retrieved: September 27, 2023).
27. Jobin, Anna,Ienca, Marcello,Vayena, Effy(2019).The Global Landscape of AI Ethics Guidelines.Nature Machine Intelligence,1(9),389-399.
28. Kazim, Emre and Adriano Koshiyama (2020). Explaining Decisions Made with AI: a Review of the Co-badged Guidance by the ICO and the Turing Institute. SSRN: https://ssrn.com/abstract=3656269 (Retrieved: July 4, 2023).
29. Kosinski, Michal,Wang, Yilun(2017).Deep Neural Networks are More Accurate than Humans at Detecting Sexual Orientation from Facial Images.Journal of Personality and Social Psychology,114(2),246-257.
30. Martin, Kirsten(Ed.)(2018).Ethics of Data and Analytics: Concepts and Cases.London:Routledge.
31. Mitchell, Margaret,Wu, Simone,Zaldivar, Andrew,Barnes, Parker,Vasserman, Lucy,Hutchinson, Ben,Spitzer, Elena,Raji, Inioluwa Deborah,Gebru, Timnit(2019).Model Cards for Model Reporting.Proceedings of the Conference on Fairness, Accountability, and Transparency,NY USA:
32. Noble, Safiya Umoja(2018).Algorithms of Oppression.New York:New York University Press.
33. Obermeyer, Ziad,Powers, Brian,Vogeli, Christine,Mullainathan, Sendhil(2019).Dissecting Racial Bias in an Algorithm Used to Manage the Health of Populations.Science,366(6464),447-453.
34. pnw.ai (2021). America Needs AI Literacy Now. ACM NEWS. https://cacm.acm.org/news/257309-america-needs-ai-literacy-now/fulltext (Dec 9, 2021) (Retrieved: July 4, 2023)
35. Prunkl, Carina,Ashurst, Carolyn,Anderljung, Markus,Webb, Helena,Leike, Jan,Dafoe, Allan(2021).Institutionalizing Ethics in AI Through Broader Impact Requirements.Nature Machine Intelligence,3(2),104-110.
36. Reese, Hope (2016). Why Microsoft’s “Tay” AI Bot Went Wrong. TechRepublic. https://www.techrepublic.com/article/why-microsofts-tay-ai-bot-went-wrong/ (Retrieved: July 14, 2022).
37. Richards, Neil. M.,King, Jonathan H.(2013).Three Paradoxes of Big Data.Stanford Law Review,66,41.
38. Schwartz, Reva,Vassilev, Apostol,Greene, Kristen,Perine, Lori,Burt, Andrew,Hall, Patrick(2022).,National Institute of Standards and Technology, U.S. Department of Commerce.
39. Srikumar, Madhulika,Finlay, Rebecca,Abuhamad, Grace,Ashurst, Carolyn,Campbell, Rosie,Campbell-Ratcliffe, Emily,Hongo, Hudson,Jordan, Sara R.,Lindley, Joseph,Ovadya, Aviv,Pineau, Joelle(2022).Advancing Ethics Review Practices in AI Research.Nature Machine Intelligence,4(12),1061-1064.
40. Strubell, Emma,Ganesh, Ananya,McCallum, Andrew(2019).Energy and Policy Considerations for Deep Learning in NLP.Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics
41. Tomašev, Nenad,Cornebise, Julien,Hutter, Frank,Mohamed, Shakir,Picciariello, Angela,Connelly, Bec,Belgrave, Danielle C. M.,Ezer, Daphne,van der Haert, Fanny Cachat,Mugisha, Frank,Abila, Gerald,Arai, Hiromi,Almiraat, Hisham,Proskurnia, Julia,Snyder, Kyle,Otake-Matsuura, Mihoko,Othman, Mustafa,Glasmachers, Tobias,de Wever, Wilfried,Teh, Yee Whye,Khan, Mohammad Emtiyaz,De Winne, Ruben,Schaul, Tom,Clopath, Claudia(2020).AI for Social Good: Unlocking the Opportunity for Positive Impact.Nature Communication,11,2468.
42. Tutt, Andrew(2017).An FDA for Algorithms.Administrative Law Review,69(1),83-124.
43. Urbina, Fabio,Lentzos, Filippa,Invernizzi, Cédric,Ekins, Sean(2022).Dual Use of Artificial Intelligence-Powered Drug Discovery.Natural Machine Intelligence,4,189-191.
44. Wakefield, Jane (2021). Amazon Faces Spying Claims over AI Cameras in Vans. BBC News. https://www.bbc.co.uk/news/technology-55938494 (Retrieved April 7, 2023).
45. Wikipedia Contributors (2022). Data Re-identification. Wikipedia. https://en.wikipedia.org/w/index.php?title=Data_re-identification&ol-did=1085774172 (Retrieved April 7, 2023).
46. Yeung, Karen(2020).Recommendation of the Council on Artificial Intelligence (OECD).International Legal Materials,59(1),27-34.
47. Zanzotto, Fabio Massimo(2019).Viewpoint: Human-in-the-loop Artificial Intelligence.Journal of Artificial Intelligence Research,64,243-252.
48. Zou, James,Schiebinger, Londa(2018).AI Can Be Sexist and Racist-It’s Time to Make It Fair.Nature,559(7714),324-326.
49. Zuboff, Shoshana,溫澤元（譯）,林怡婷（譯）,陳思穎（譯）(2020).監控資本主義時代.臺北:時報出版.
50. 尤聰光（2020），〈臺東大學圖書館人臉辨識借書服務〉。《東大新聞網》：9 月 3 日。https://enews.nttu.edu.tw/p/406-1045-101249,r923.php（檢索日期 : 2023 年 9 月 27 日）。
51. 甘偵蓉(2022)。個人隱私在 AI 時代面臨哪些風險？資料治理的重要性。科學月刊，492
52. 何之行,廖貞(2020)。AI 個資爭議在英國與歐盟之經驗—以 Google DeepMind一案為例。法律思維與制度的智慧轉型，臺北:
53. 李世德(2018)。GDPR 與我國個人資料保護法之比較分析。臺灣經濟論衡，16(3)，69-93。
54. 李建良（2020），〈人社學人與國科會 AI 科研發展指引的對話〉。《臺灣人工智慧行動網》：1 月 8 日。https://ai.iias.sinica.edu.tw/most-ai-tech-rd-guidelines-dialogue/（檢索日期：2023 年 4 月 7 日）。
55. 李建良（編）(2020).法律思維與制度的智慧轉型.臺北:元照.
56. 國科會（2019），〈人工智慧科研發展指引〉。https://www.nstc.gov.tw/nstc/attachments/53491881-eb0d-443f-9169-1f434f7d33c7 （檢索日期：2022 年 12 月 16 日）。
57. 國科會前瞻及應用科技司（2019），〈國科會訂定「人工智慧科研發展指引」 完善我國 AI 科研發展環境〉。國家科學及技術委員會：9 月 23 日。https://www.nstc.gov.tw/folksonomy/detail/dbf8da09-22be-4ef1-8294-8832fc6e8a26?l=ch（檢索日期：2023 年 4 月 7 日）。
58. 陳芳毓、許鈺屏、李鈺淇、高偉倫、阮怡婷（2021），〈什麼是永續發展目標SDGs？17項目標一次掌握〉。《未來城市FutureCity@天下》。https://futurecity.cw.com.tw/article/1867（檢索日期：2023 年 4 月 7 日）。
59. 陳秋雲、喻文玟、趙宥寧、游振昇（2020），〈中女中宿舍裝人臉辨識系統爆爭議〉。《聯合新聞網》：9 月 3 日。https://udn.com/news/story/6898/4833571（檢索日期：2023 年 4 月 7 日）。
60. 廖英凱(2022)。當人工智慧的傷害已經出現，誰來補救？。科學月刊，481
61. 臺灣能源永續研究基金會（2020），〈AI 可以幫助拯救地球的 8 種方法 〉。https://www.taise.org.tw/post-view.php?ID=160（ 檢索日期：2023 年 4 月 7 日）。
62. 潘璿安、周倩（2020），〈研究資料管理概念〔線上課程編號0116〕〉及〈研究資料處理：案例探討〔線上課程編號 0205〕〉。教育部臺灣學術倫理教育資源中心。https://ethics.moe.edu.tw（檢索日期：2023 年 4 月 7 日）。
63. 鄭宇君,陳恭,陳百齡(2017)。社群媒體巨量資料蒐集與分析—以 Facebook 與 Twitter 為例。人文與社會科學簡訊，19(1)，37-46。
64. 憲法法庭（2022），111 年憲判字第 13 號【健保資料庫案】。https://cons.judicial.gov.tw/docdata.aspx?fid=97&id=309956&rn=9897（ 檢索日期：2023 年 4 月 7 日）。

1. (2024)。AI開發過程的倫理權衡：自駕車決策案例研究。歐美研究，54(1)，1-67。