본문 바로가기 주메뉴 바로가기
국회도서관 홈으로 정보검색 소장정보 검색

목차보기

Title page 1

Contents 3

I. INTRODUCTION 5

II. TECHNICAL BACKGROUND 8

A. Machine Learning 8

B. Generative Language Models 10

C. Training Data 13

1. Data Characteristics 13

2. Acquisition and Curation 17

D. Training 21

1. Training Phases 21

2. Memorization 23

E. Deployment 25

III. PRIMA FACIE INFRINGEMENT 30

A. Data Collection and Curation 30

B. Training 31

C. RAG 34

D. Outputs 35

IV. FAIR USE 36

A. Factor One 39

1. Identifying the Use 40

2. Transformativeness 41

3. Commerciality 52

4. Unlawful Access 55

B. Factor Two 57

C. Factor Three 58

1. The Amount Used 59

2. Reasonableness in Light of Purpose 59

3. The Amount Made Available to the Public 61

D. Factor Four 65

1. Lost Sales 66

2. Market Dilution 68

3. Lost Licensing Opportunities 70

4. Public Benefits 75

E. Weighing the Factors 78

F. Competition Among Developers 78

G. International Approaches 80

V. LICENSING FOR AI TRAINING 89

A. Voluntary Licensing 89

1. Feasibility of Voluntary Licensing 90

2. Ability to Provide Meaningful Compensation 96

3. Possible Legal Impediments to Collective Licensing 98

B. Statutory Approaches 99

1. Compulsory Licensing 99

2. Extended Collective Licensing 103

3. Opting Out 105

C. Analysis and Recommendations 107

VI. CONCLUSION 111