본문 바로가기
KERD
논문KCI 등재

인공지능과 데이터 윤리에 관한 소고(小考): 인공지능에 활용되는 헬스 데이터를 중심으로

A Study on Artificial Intelligence and Data Ethics: Focusing on Health Data Used in Artificial Intelligence

박미정
한국의료윤리학회 (The Korean Society For Medical Ethics)|2019.09|한국의료윤리학회지 (Korean Journal of Medical Ethics)|제22권 제3호|pp. 255-273|DOI 10.35301/ksme.2019.22.3.255

0

조회

0

다운로드

7

인용

0

참고문헌

키워드

인공지능알고리즘healthalgorithms윤리ethicsartificial intelligence익명과 가명헬스anonyms and pseudonyms
주제분류의학일반

초록

다양한 방식과 형태로 수집되는 개인의 헬스 데이터는 과학적 연구목적으로 이차 활용된다. 국내외개인정보보호 법률은 과학적 연구를 위한 데이터 처리의 법적 근거로 정보 주체의 동의를 얻는 것 이외에 데이터의 익명성을 요구한다. 공공이익과 관련된 연구라면 정보 주체의 동의가 면제될 수 있는데, 이러한 양립 가능성을 적용하기 위해서는 법 제도 이외의 측면도 고려해야 한다. 본 논문의 연구 질문은 인공지능을 학습시키는 데 데이터가 활용된다는 점에서 시작한다. 인공지능 알고리즘의 특성을 우선 살펴보고 이러한 기술에 활용되는 헬스 데이터의 사례와 제기되는 문제점을 고찰하였다. 인공지능기술 특성의 틀 안에 갇혀있는 문제는 윤리 논의의 주제가 될 수 있다. 법률을 통해 준비하고 해석하기어려운 문제점으로 사전동의와 책임, 데이터 익명성에 대한 신화, 위험점수와 알고리즘적 차별을 설명하였다. 그리고 문제해결을 위해 변화를 꾀한 국내외 법률의 데이터처리 원칙에 대한 입법 과정과 법률조항을 분석하여 해결을 위한 단서를 찾아보았다. 결론에서 ‘독(毒)’을 제거하는 데이터 처리, 사전적 절차로서의 법 제도와 조화, 사후적 판단을 위한 알고리즘의 투명한 설계를 제언하였다.

Abstract

Health data collected in various ways and forms is secondary use of scientific research purposes. Including Republic of Korea, several country’s Data protection law require the anonymity of data in addition to obtaining the consent of the data subject as a provisions relating to specific data processing situations. In the case of research related to the important objectives of public interest, the informed consent of the subject shall be exempt from that liability. In order to find such compatibility of purposes, consideration will need to be taken in terms other than the lawful processing of personal data. This paper starts with the fact that data is used to train artificial intelligence. First, artificial intelligence needs to focus on specifics on what data are used in the training of the artificial intelligence and how the algorithms are built, and the concerns arising from the mechanism of algorithms are discussed. The data used in the artificial intelligence system are considered as the subject of ethical debate and the ethical problems are discussed. And analyzed the legislative process and legal provisions for data processing principles of EU General Data Protection Regulation to find clues to solving that problems. The problems that can arise due to the characteristics of artificial intelligence technology, which are hard to prepare and interpreter through legislation, are explains to informed consent and responsibility, the myth of data anonymity and, risk scores and algorithmic discrimination. As a conclusion, I suggested about data processing to removes ‘poison’ from the data, harmonization with legal system as an ex ante procedure, and transparent design of the algorithm for human judgment as a whole.

참고문헌 (55)

  1. "Artificial intelligence in society" (2019)[URL]
  2. Constantiou ID, "New games, new rules: big data and the changing context of strategy", Journal of Information Technology, 30(1), pp.44~57 (2015)
  3. 마이클 네그네빗스키, "인공지능 개론", pp.23~25 (2013), 한빛아카데미
  4. "Just don’t call it privacy" (2018)[URL]
  5. "2018 reform of EU data protection rules"[URL]
  6. "Bias detectives: the researchers striving to make algorithms fair"[URL]
  7. "AI can be sexist and racist — it’s time to make it fair"[URL]
  8. "OECD Guidelines on the Protection of Privacy and Transborder Flows of Personal Data"[URL]
  9. "환자 편의를 위한 진료정보교류사업 전국 네트워크 완성", 보건복지부 보도자료, ~ (2019)
  10. Chow-White P, "From the bench to the bedside in the big data age: ethics and practices of consent and privacy for clinical genomics and personalised medicine", Ethics and Information Technology, 17(3), pp.189~200 (2015)
  11. "Data Protection Working Party Article 29, Opinion 03/2013 on purpose limitation" (2013)[URL]
  12. Tene O, "Big Data for All: privacy and user control in the age of analytics", Nw J Tech & Intell Prop, 11(5), pp.239~273 (2013)
  13. "Recommendation of the Council on Health Data Governance" (2016)[URL]
  14. "The Nuremberg Code"[URL]
  15. "Reports to Congress on Breach Notification Program, Annual Report to Congress on Breaches of Unsecured Protected Health Information 2013-2014"[URL]
  16. "Definition of limited data set 2015" (2015)[URL]
  17. Office of the Federal Register, "Federal policy for the protection of human subjects", Federal Register, 82(12), pp.7149~ (2017)
  18. "Revised Common Rule Q&As"[URL]
  19. Selbst AD, "The intuitive appeal of explainable machines", Fordham Law Rev, 87, pp.1094~1096 (2018)
  20. Fears R, "Data protection regulation and the promotion of health research: getting the balance right", Quarterly Journal of Medicine, 107(1), pp.3~5 (2014)
  21. Agrawal A, "Prediction Machines: The Simple Economics of Artificial Intelligence" (2018), Harvard Business School Press
  22. Challen R, "Artificial intelligence, bias and clinical safety", BMJ Qual Saf, 28(3), pp.231~237 (2019)
  23. "Medicaid Managed Long-Term Services and Supports"[URL]
  24. Pennsylvania Health Law Project, "Updates on Community Health Choices", Health Law PA News, 21(5), pp.1~10 (2018)
  25. "How your health information is sold and turned into ‘risk scores’"[URL]
  26. "Fighting the opioid epidemic"[URL]
  27. "Mapping patterns to fight the opioid crisis"[URL]
  28. "LexisNexis Risk Solutions Government CEO: the role of data and analytics in helping to mitigate the opioid epidemic at HIMSS" (2018)[URL]
  29. "알고리즘"[URL]
  30. Lusignan S, "Effective pseudonymisation and explicit statements of public interest to ensure the benefits of sharing health data for research, quality improvement and health service management outweigh the risks", Informatics in Primary Care, 21(2), pp.61~63 (2014)
  31. "he protection of individuals with regard to automatic processing of personal data in the context of profiling Recommendation CM/Rec(2010)13"[URL]
  32. "Making an ethical machine"[URL]
  33. Barocas S, "Big data’s disparate impact", California Law Review, 104(3), pp.671~732 (2016)
  34. "Amazon scraps secret AI recruiting tool that showed bias against women"[URL]
  35. "ow well do IBM, Microsoft, and Face++ AI services guess the gender of a face"[URL]
  36. Wolpert D, "No Free Lunch Theorems for Optimization", IEEE Transactions on Evolutionary Computation, 1(1), pp.67~82 (1997)
  37. "AI has a culturally biased world view that Google has a plan to change"[URL]
  38. "Top 9 ethical issues in artificial intelligence" (2016)[URL]
  39. "Big data:a report on algorithmic systems, opportunity, and Civil Rights" (2016)[URL]
  40. "The hidden biases in big data. Harvard Business Review" (2013)[URL]
  41. "Guidelines on Automated individual decision-making and Profiling for the purposes of Regulation 2016/679"[URL]
  42. 박미정, "보건의료 빅데이터 활용에 관한 법·정책적 개선방안 연구", 한국의료법학회지, 26(1), pp.163~192 (2018), 한국의료법학회[DOI]
  43. Kuner C, "Machine learning with personal data: Is data protection smart enough to meet the challenge", International Data Privacy Law, 7(1), pp.1~2 (2017)
  44. "Artificial intelligence for Europe. COM 2018"[URL]
  45. "Derogations and special conditions"[URL]
  46. Calabresi G, "Property rules, liability rules and inalienability: one view of the cathedral", Harvard Law Review, 85, pp.1089~ (1972)
  47. "Optimization over explanation -maximizing the benefits of machine learning without sacrificing its intelligence" (2018)[URL]
  48. Official Journal of the European Union, "Consolidated version of the treaty on european union", ~ (2012)
  49. "Consolidated version of the Treaty on the Functioning of the European Union" (2012)[URL]
  50. "Playing with AI fairness:Google's new machine learning diagnostic tool lets users try on five different types of fairness"[URL]
  51. Crawford K, "Big data and due process: toward a framework to redress predictive privacy harms", Boston Law Review, 55, pp.101~125 (2014)
  52. Citron DK, "Technological due process", Washington University Law Review, 85, pp.1249~1313 (2008)
  53. Citron DK, "The Scored Society : due process for automated predictions", Washington Law Review, 89(1), pp.15~16 (2014)
  54. "Is ethical A.I. even possible"[URL]
  55. Ricoeur P, "Soi-même comme un autre", pp.279~290 (1990), Ed́ itions du Seuil

관련 논문AI추천

데이터윤리의 의미와 원칙에 대한 연구변순용한국초등도덕교육학회 · 초등도덕교육2023 · 제83호보건의료 분야 인공지능 연구자를 위한 연구윤리 지침 개발김소윤, 김수민, 김은애, 김준혁, 김한나, 심진아, 이유리, 이정임한국의료법학회 · 한국의료법학회지2023 · 제31권 제1호인공지능 정보의 사용에 따른 개인정보보호에 관한 법적 연구 – 데이터 3법의「동의권」관련 분석을 중심으로 –이동건법학연구소 · 홍익법학2023 · 제24권 제3호우리나라의 인공지능(AI)서비스를 위한 개인정보보호 개선방안신영진중소기업융합학회 · 융합정보논문지2021 · 제11권 제3호인공지능시스템과 데이터 프라이버시를 위한 입법 고찰박미정한국의료법학회 · 한국의료법학회지2024 · 제32권 제2호보건의료영역 인공지능 윤리 원칙과 고려사항문기업, 손유미, 양지현, 이일학, 최은경한국의료윤리학회 · 한국의료윤리학회지2023 · 제26권 제2호

이 간행물의 다른 논문 (3건)

연명의료결정법과 의료기관윤리위원회: 현황, 경험과 문제점이일학, 최지연, 김정아, 장승경pp. 209-233중환자실 간호사가 경험한 심폐소생술금지 환자의 윤리적 문제에 대한 윤리적 의사결정장재인pp. 234-254공학의 시선에서 바라본 미래의 의료윤리: ‘치료’인가 ‘수리’인가김문정, 심지원pp. 274-286
원문 보기