Title page

Contents

Resume 4

Kurzfassung 5

Acknowledgements 6

Executive Summary 9

Synthese 10

Zusammenfassung 12

1. Introduction 14

2. Setting the scene 16

3. Data and methodology 18

Identifying AI-related patents among all patents 18

Identifying 'core' AI patents 18

Grouping core AI patents into different topics 19

4. Exploring the characteristics of core AI patents 22

Measuring the quality of core AI patents: scope, originality, generality and radicalness 22

Exploring core AI innovations thematically 24

Exploring the role of core AI patents for subsequent innovations 25

5. Concluding remarks and next steps 32

Endnotes 34

References 36

Annex A. Additional information about data and methodology 41

Annex B. Additional results 44

Figure 3.1. AI-related patents and core AI patents, 2000-18 19

Figure 3.2. Pairwise correlation of AI keywords, core AI patents, 2000-18 20

Figure 4.1. Patent quality indicators, different patent groups, 2000-18 23

Figure 4.2. Word clouds of AI main topics, 2000-18 25

Figure 4.3. Core AI patents with AI citations, 2000-18 26

Figure 4.4. Algorithms-related AI patents and their citations, 2000-18 27

Figure 4.5. Recognition- or Detection-related AI patents and their citations, 2000-18 29

Figure 4.6. Autonomous Driving-related AI patents and their citations, 2000-18 30

Figure 4.7. Robotics-related AI patents and their citations, 2000-18 31

Boxes

Box 4.1. Four indicators measuring the quality of core AI patents 24

Annex Tables

Table A A.1. List of AI keywords and corresponding AI topics 42

Annex Figures

Figure A B.1. Patent quality indicators, core AI vs. top cited non-AI patents, 2000-18 44

Figure A B.2. Patent quality indicators, different patent groups, 2000-09 vs. 2010-18 45