Understanding Machine Understanding

Table of Contents

Acknowledgements xv

Preface: Co-Writing with AI xvii

Introduction: A Quest for Understanding xxi

CHAPTER 1

A Brief History of Computing and AI 1

1.1 Early Visionaries and Key Milestones 3

1.2 The Birth of Artificial Intelligence as a Field 5

1.3 Paradigm Shifts and Breakthroughs 6

1.4 Back in the Lab 7

References for Chapter 1 9

CHAPTER 2

Theories and Tests of Intelligence 11

2.1 Philosophical Perspectives on the Nature of Understanding 11

2.1.1 Just? 13

2.2 The Turing Test and Its Legacy 14

2.2.1 Understanding Itself 16

2.3 Searle's Chinese Room Thought Experiment 17

2.3.1 The Thought Experiment 17

2.3.2 Searle's Conclusions 18

2.3.3 Responses and Objections 18

2.3.4 Continuing Influence and Debate 19

2.3.5 Transcending the Chinese Room 20

2.4 Limitations of Behavioral Tests and the Symbol Grounding Problem 22

2.5 Turing Enough? 25

References for Chapter 2 28

CHAPTER 3

Knowledge vs. Understanding-A Crucial Distinction 31

3.1 Defining Knowledge as Information Retrieval and Understanding

as Reasoning and Insight 31

3.2 Limitations of Knowledge-Focused AI Benchmarks 32

3.2.1 Taking a Step 32

CHAPTER 5

Implementing the MUTT 51

5.1 Modular Architecture and Component Skills 51

5.1.1 Language Comprehension 51

5.3 Proposed Configuration of the Multifaceted Understanding Test Tool 77

5.3.1 Language Comprehension 77

5.3.2 Reasoning and Abstraction 78

5.3.3 Knowledge Integration 78

5.3.4 Perception and Embodiment 79

5.3.5 Social Cognition 79

References for Chapter 5 91

CHAPTER 6

Verifying and Validating MUTT Results 93

6.1 Importance of Verification and Validation 93

6.2 Verification Strategies 94

6.2.1 Code and Implementation Review 94

6.7 Prototype in Place 104

References for Chapter 6 106

CHAPTER 7

Societal Implications of Machine Understanding 109

7.1 Introduction 109

7.1.1 What if all they do is Think? 110

7.2 Transforming the Nature of Work 112

7.3 Impact on Social Interactions and Relationships 113

7.3.1 Escape from AI-catraz 115

7.4 Governance, Policy, and Regulation 117

7.5 Philosophical Implications and the Future of Human Identity 119

7.5.1 But, Can you Prove it? 122

7.6 Conclusion of Chapter 7 125

References for Chapter 7 127

CHAPTER 8

The Future of AI Evaluation 129

8.1 Introduction 129

8.2 The Limitations of Current Evaluation Paradigms 129

8.6 The Future of Human-AI Collaboration in Evaluation 134

8.7 Wither Goeth Thou? 135

8.8 Am I Conscious, or Do I Just Think I am? 136

APPENDIX 1

The Neuroscience of Human Understanding 145

A1.1 Distributed Representations and Processing in the Brain 145

A1.2 The Role of Context, Prior Knowledge, and Embodiment 147

A1.3 Insights from Cognitive Neuroscience for AI Understanding 148

References for Appendix A1 150