AI 추천 알고리즘 편향성과 규제에 관한 연구 = A Study on the Bias and Regulation of AI Algorithm

다국어 초록 (Multilingual Abstract)

With the rapid development of artificial intelligence technology, user-customized AI recommendation algorithms are being used on various platforms. AI recommendation algorithms suggest content deemed of interest to users based on user history and are necessary, to a large extent, to allow ‘selection’ from a vast pool of information. As platforms offer recommendations based on user preference profiling, there are concerns that users increasingly interact only with like-minded individuals and thus fall into an echo chamber. Such a problem, combined with UGC, leads to a polarization of public opinion, caused by false facts and misinformation on common social issues such as COVID-19.
This study confirms, through experiments, that YouTube’s AI recommendation algorithm makes it difficult for users to obtain accurate information through reliable content, once they begin watching UGC containing false or unconfirmed facts, as the algorithm continues to recommend UGC of similar nature. This study also examines AI recommendation algorithms to check for AI bias and compares relevant policies and regulations in different countries.
Our analysis of regulatory proposals and guidelines on algorithms shows considerable similarity between Korea and other countries, regarding ‘user-centricity’ and ‘transparent disclosure’ of AI algorithms. And, although differences have been observed in the scope of regulation, the examined proposals and guidelines are found to be drawing influence from one another. Based on such findings, this study provides a comparative analysis of South Korea’s policies and regulatory directions for AI recommendation algorithms, considers their effects, and proposes such measures as establishing concrete guidelines and supporting user choices to solve the bias problem of AI recommendation algorithms.

참고문헌 (Reference)

1 양소리, "홍준표 “틀튜브가 한국 보수 망쳐…안 보면 된다”"

2 서봉원, "콘텐츠 추천 알고리즘의 진화" 05 : 19-, 2016

3 이문한, "코로나19 관련 가짜뉴스 등 ‘공공의 이익’을 해하는 허위사실 표현에 대한 형사처벌과 헌법상 언론⋅출판의 자유" 11 : 2021

4 방송통신위원회, "인공지능 기반 미디어(매체) 추천 서비스 이용자 보호 기본원칙"

5 오세욱, "알고리즘으로 본 유튜브의 미디어 지향, "관훈저널"" (봄) : 2019

6 홍윤지, "알고리즘 설명요구권’ 놓고 영업기밀 침해 논란도"

7 김유미, "소셜미디어의 가짜뉴스(Fake News)에 대한 제3자 효과: 감염병 관련 허위정보를 중심으로" 한국방송학회 35 (35): 5-32, 2021

8 AI 기반 추천 서비스 이용자 보호를 위한 기본원칙(안) 공개 토론회, "방통위 추진 ‘AI 기반 추천서비스 기본원칙’에 업계⋅소비자 반응 엇갈려"

9 갤럽리포트, "미디어, 콘텐츠, 소셜 네트워크 서비스 이용률"

10 정연오 ; 이성우 ; 이지형, "개인화된 전문가 그룹을 활용한 추천 시스템" 한국지능시스템학회 23 (23): 7-11, 2013

11 이상훈, "가짜뉴스의 법적 규제에 대한 고찰" 사단법인 한국법이론실무학회 8 (8): 95-137, 2020

12 최승필, "가짜뉴스에 대한 규제법적 검토 - 언론관련법 및 정보통신망법상 규제를 중심으로 -" 한국비교공법학회 21 (21): 115-147, 2020

13 파이낸셜뉴스, "“플랫폼 알고리즘은 영업비밀” 전자상거래법 손질 들어가나"

14 Lewis, P., "‘Fiction is outperforming reality’: how YouTube’s algorithm distorts truth"

15 Apple, "iPhone에서 Apple이 광고를 제공하는 방식 제어하기"

16 김청희, "[초점] 성별, 연령대별 유튜브 및 넷플릭스 콘텐츠 이용행태 분석" (3) : 1-21, 2021

17 Bonefeld-Dahl, C., "[BLOG] Seven questions to ask when looking at the EU’s new AI regulation tomorrow"

18 윤영찬, "[2113509] 알고리즘 및 인공지능에 관한 법률안"

19 Smith, A., "Using Artificial Intelligence and Algorithms"

20 Sustein, C., "The Echo Chamber Is the Enemy of Democracy"

21 European Institute of Public Administration, "The Artificial Intelligence Act Proposal and its Implications for Member States"

22 Connor, J., "Reduced Demand Uncertainty and the Sustainability of Collusion: How AI Could Affect Competition"

23 Central Digital and Data Office, "Provide information on how you use algorithmic tools to support decisions (pilot version)"

24 European Commission, "Proposal for a regulation of the european parliament and of the council laying down harmonised rules on artificial intelligence (artificial intelligence act) and amending certain union legislative acts"

25 EUROPEAN CENTRAL BANK, "Opinion of the European Central Bank of 29 December 2021 on a proposal for a regulation laying down harmonised rules on artificial intelligence"

26 네이버 Agenda Research, "NAVER-SAPI AI REPORT"

27 김춘식 ; 홍주현, "How Fake News Become ‘Real’ News on Youtube: A Case Study of Political Propagandization and Reaction Among Political Actors Related to the 〈Gosung Wildfire〉" The Korean Association of Political Science and Communication 23 (23): 403-439, 2020

28 Federal Trade Commision, "Hearing #7: The Competition and Consumer Protection Issues of Algorithms, Artificial Intelligence, and Predictive Analytics"

29 Kahn, J., "Europe proposes strict A.I. regulation likely to have an impact around the world"

30 양종모, "Effects of bias and opacity of artificial intelligence algorithms on legal decision making and its discipline" 법조협회 66 (66): 60-105, 2017

31 Arguedes, A., "Echo chambers, filter bubbles, and polarisation: a literature review"

32 정원섭, "Discrimination and Bias of Artificial Intelligence" 인간환경미래연구원 null (null): 55-73, 2020

33 Martínez, A., "Are Facebook Ads Discriminatory? It’s Complicated"

34 Central Digital and Data Office, "Algorithmic transparency template"

35 Jilson, E., "Aiming for truth, fairness, and equity in your company’s use of AI"

36 강성원, "AI 편집하는 네이버 뉴스 ‘감성 뉴스’ 사라질라"

37 IBM Research Trusted AI, "AI Faireness 360"

38 한국언론진흥재단, "2021 소셜미디어 이용자 조사 결과표" 한국언론진흥재단 351 : 2021