국회도서관 국가전략정보포털

Title page 1

Contents 7

Foreword 7

Acknowledgment 7

Executive Summary 9

References 11

1. Overview of current AI capabilities 12

Comparative table of current AI capabilities 13

Commentary on current ratings 15

References 17

2. Constructing a framework to measure AI capabilities 18

AI has progressed beyond our understanding of its capabilities 19

Methodology: A novel and unique approach 20

Constructing and developing the indicators 22

Limitations 24

Next steps 24

The role of the AI Capability Indicators 25

References 27

3. OECD AI Capability Indicators 28

Language scale 29

Social interaction scale 31

Problem-solving scale 33

Creativity scale 35

Metacognition and critical thinking scale 37

Knowledge, learning and memory 39

Vision scale 41

Manipulation scale 43

Robotic intelligence scale 45

References 47

4. Policy use cases for the AI Capability Indicators 48

Mapping the indicators to occupational demand for human abilities 49

Transformational change in education 52

Conclusion 53

References 54

Tables 8

Table 1.1. Overview of current AI capability levels 14

Table 3.1. AI language scale 30

Table 3.2. AI social interaction scale 32

Table 3.3. AI problem-solving scale 34

Table 3.4. AI creativity scale 36

Table 3.5. AI metacognition and critical thinking scale 38

Table 3.6. AI knowledge, learning and memory scale 40

Table 3.7. AI vision scale 42

Table 3.8. AI manipulation scale 44

Table 3.9. AI robotic intelligence scale 46

Figures 8

Figure 2.1. The development of the OECD AI Capability Indicators 21

Figure 2.2. OECD AI Capability Indicators 21

Figure 2.3. Overview of the five levels of AI 23

Figure 4.1. Mapping the AI Capability Indicators of language, problem solving and social interaction to the capability requirements of teachers' tasks 50

Figure 4.2. A framework for analysing implications of increasing AI capabilities in education 52

Boxes 8

Box 4.1. Vignette of AI at mid-level of the Problem-solving scale supporting pandemic response 51