Artificial General Intelligence

354023733X......Page 1
Cognitive Technologies......Page 2
Artificial
General
Intelligence......Page 3
Contents......Page 8
1 A Brief History of AGI......Page 15
1.1 Some Historical AGI-Related Projects......Page 16
2.1 The Psychology of Intelligence......Page 20
2.3 A Control Theory Approach to Defining Intelligence......Page 22
2.4 Efficient Intelligence......Page 24
3 The Abstract Theory of General Intelligence......Page 25
4 Toward a Pragmatic Logic......Page 29
5 Emulating the Human Brain......Page 31
6 Emulating the Human Mind......Page 33
7 Creating Intelligence by Creating Life......Page 36
8 The Social Nature of Intelligence......Page 38
9 Integrative Approaches......Page 40
10 The Outlook for AGI......Page 41
References......Page 42
1.1 To Define Intelligence......Page 45
1.2 A Working Definition of Intelligence......Page 47
1.3 Comparison With Other Definitions......Page 49
1.4 Logic and Reasoning Systems......Page 54
2.1 Experience-Grounded Semantics......Page 57
2.2 Inheritance Statement......Page 59
2.3 Categorical Language......Page 61
2.4 Syllogistic Inference Rules......Page 62
2.5 Controlled Concurrency in Dynamic Memory......Page 64
3.1 Reasonable Solutions......Page 66
3.2 Unified Uncertainty Processing......Page 67
3.3 NARS as a Parallel and Distributed Network......Page 68
3.4 Resources Competition......Page 70
3.5 Flexible Behaviors......Page 71
3.6 Autonomy and Creativity......Page 72
References......Page 74
1 Introduction......Page 77
1.1 The Novamente AGI System......Page 78
1.2 Novamente for Knowledge Management and Data Analysis......Page 79
2 Enabling Software Technologies......Page 81
2.1 A Distributed Software Architecture for Integrative AI......Page 82
2.2 Database Integration and Knowledge Integration......Page 84
3 What Is Artificial General Intelligence?......Page 86
3.1 What Is General Intelligence?......Page 87
3.2 The Integrative Approach to AGI......Page 89
3.3 Experiential Interactive Learning and Adaptive Self-modification......Page 91
4 The Psynet Model of Mind......Page 94
5 The Novamente AGI Design......Page 97
5.1 An Integrative Knowledge Representation......Page 98
5.2 The Mind OS......Page 102
5.3 Atom Types......Page 105
5.4 Novamente Maps......Page 108
5.5 Mind Agents......Page 109
5.6 Map Dynamics......Page 110
5.7 Functional Specialization......Page 113
5.8 Novamente and the Human Brain......Page 114
5.9 Emergent Structures......Page 116
6 Interacting with Humans and Data Stores......Page 118
6.1 Data Sources......Page 119
6.2 Knowledge Encoding......Page 120
6.3 Querying......Page 121
6.4 Formal Language Queries......Page 122
6.6 Report Generation......Page 123
7 Example Novamente AI Processes......Page 124
7.1 Probabilistic Inference......Page 126
7.2 Nonlinear-Dynamical Attention Allocation......Page 129
7.3 Importance Updating......Page 130
7.4 Schema and Predicate Learning......Page 131
7.5 Pattern Mining......Page 134
7.6 Natural Language Processing......Page 136
8 Conclusion......Page 138
Appendix: Novamente Applied to Bioinformatic Pattern Mining......Page 139
References......Page 141
2 General Intelligence......Page 144
2.1 Core Requirements for General Intelligence......Page 146
2.2 Advantages of Intelligence Being General......Page 147
3 Shortcuts to AGI......Page 148
4 Foundational Cognitive Capabilities......Page 155
5 An AGI in the Making......Page 157
5.1 AGI Engine Architecture and Design Features......Page 158
6 From Algorithms to General Intelligence......Page 160
6.1 Sample Test Domains for Initial Performance Criteria......Page 161
6.2 Towards Increased Intelligence......Page 162
7 Other Research......Page 163
8 Fast-track AGI: Why So Rare?......Page 165
9 Conclusion......Page 168
References......Page 169
1 Introduction......Page 171
2 Evolvable Hardware......Page 173
2.1 Neural Network Models......Page 174
3 The CAM-Brain Machine (CBM)......Page 178
3.1 Evolved Modules......Page 179
3.2 The Kitten Robot "Robokitty"......Page 180
4 Short- and Long-Term Future......Page 183
5 Postscript – July 2002......Page 184
References......Page 186
1 Introduction......Page 187
2 More Formally......Page 188
3 Prediction Using a Universal Algorithmic Prior Based on the Shortest Way of Describing Objects......Page 189
4 Super Omegas and Generalizations of Kolmogorov Complexity & Algorithmic Probability......Page 191
5 Computable Predictions Through the Speed Prior Based on the Fastest Way of Describing Objects......Page 193
6 Speed Prior-Based Predictions for Our Universe......Page 194
7 Optimal Rational Decision Makers......Page 196
8 Optimal Universal Search Algorithms......Page 197
9 Optimal Ordered Problem Solver (OOPS)......Page 198
10 OOPS-Based Reinforcement Learning......Page 202
11 The Gödel Machine......Page 203
12 Conclusion......Page 204
References......Page 206
1 Introduction and Outline......Page 211
2 Basic Overview, Relation to Previous Work, and Limitations......Page 212
2.1 Notation and Set-up......Page 213
2.3 Proof Techniques and an O()-optimal Initial Proof Searcher......Page 215
2.4 Relation to Hutter's Previous Work......Page 216
2.5 Limitations of Gödel Machines......Page 217
3.1 Proof Techniques......Page 218
4.1 Alternative Relaxed Target Theorem......Page 224
5 Bias-Optimal Proof Search (BIOPS)......Page 225
5.1 How a Surviving Proof Searcher May Use Biops to Solve Remaining Proof Search Tasks......Page 226
6.1 Possible Types of Gödel Machine Self-improvements......Page 227
6.3 Probabilistic Gödel Machine Hardware......Page 229
6.4 More Relations to Previous Work on Less General Self-improving Machines......Page 230
6.5 Are Humans Probabilistic Gödel Machines?......Page 232
6.7 Frequently Asked Questions......Page 233
7 Conclusion......Page 234
References......Page 235
1 Introduction......Page 239
2.1 The Cybernetic Agent Model......Page 242
2.3 AI Model for Known Deterministic Environment......Page 244
2.4 AI Model for Known Prior Probability......Page 245
2.5 Probability Distributions......Page 247
2.6 Explicit Form of the AIμ Model......Page 248
2.7 Factorizable Environments......Page 250
2.8 Constants and Limits......Page 251
2.9 Sequential Decision Theory......Page 252
3.1 Introduction......Page 253
3.2 Algorithmic Information Theory......Page 254
3.3 Uncertainty & Probabilities......Page 255
3.4 Algorithmic Probability & Universal Induction......Page 256
3.5 Loss Bounds & Pareto Optimality......Page 257
4.1 The Universal AIξ Model......Page 258
4.2 On the Optimality of AIXI......Page 261
4.3 Value Bounds and Separability Concepts......Page 263
4.4 Pareto Optimality of AIξ......Page 266
4.5 The Choice of the Horizon......Page 267
4.6 Outlook......Page 269
4.7 Conclusions......Page 270
5.1 Sequence Prediction (SP)......Page 271
5.2 Strategic Games (SG)......Page 273
5.3 Function Minimization (FM)......Page 277
5.4 Supervised Learning from Examples (EX)......Page 281
5.5 Other Aspects of Intelligence......Page 283
6 Time-Bounded AIXI Model......Page 284
6.1 Time-Limited Probability Distributions......Page 285
6.2 The Idea of the Best Vote Algorithm......Page 287
6.4 Valid Approximations......Page 288
6.6 The Universal Time-Bounded AIXItl Agent......Page 289
6.7 Limitations and Open Questions......Page 290
6.8 Remarks......Page 291
7.1 General Remarks......Page 292
7.2 Outlook & Open Questions......Page 294
7.3 The Big Questions......Page 295
7.4 Conclusions......Page 296
Annotated Bibliography......Page 297
References......Page 299
1 Intelligence and the Search for Programs......Page 303
2 Theoretical Results......Page 306
2.1 Program Search in the Standard AI Model......Page 307
2.2 Self-improving Program Search......Page 308
2.3 Discussion of Efficiency Definitions......Page 310
3 Convenient Model of Computation......Page 311
3.1 Extended Program Notation......Page 318
3.2 Compiling Typed Rewriting Systems......Page 323
4 Reasoning Using Games......Page 326
4.1 Reason and Search Game for Terms......Page 330
5 Conclusions......Page 336
References......Page 337
1 Introduction......Page 339
2 The Epistemological Problem......Page 340
3 Approaches to the Theory of Evolutionary Origin of Human Intelligence......Page 342
3.1 "Intelligent Inventions" of Biological Evolution......Page 343
3.2 Methodological Approaches......Page 346
3.3 Role of Investigations of "Artificial Life" and "Simulation of Adaptive Behavior"......Page 349
4.1 Alife Model of Evolutionary Emergence of Purposeful Adaptive Behavior......Page 350
4.2 Model of Evolution of Web Agents......Page 355
5 Towards the Implementation of Higher Cognitive Abilities......Page 359
References......Page 361
1 Introduction......Page 364
2.3 The Humble Earthworm......Page 365
3 Consciousness......Page 366
3.1 Feeling and Qualia......Page 367
4 General Intelligence......Page 369
4.1 Human Intelligence......Page 371
5 3D Simulation and Language......Page 374
6 Epistemology......Page 377
7 Instantiation: the Heart of Consciousness......Page 378
8 In a Nutshell......Page 381
9 Real-World AI......Page 385
9.1 Examples and Metaphors......Page 389
9.3 Barcode Example......Page 391
9.4 Software Design......Page 394
10 Conclusion......Page 396
References......Page 397
1 Foundations of General Intelligence......Page 399
2.1 Concepts: An Illustration of Principles......Page 407
2.2 Levels of Organization in Deliberation......Page 417
2.3 The Code Level......Page 419
2.4 The Modality Level......Page 426
2.5 The Concept Level......Page 436
2.6 The Thought Level......Page 454
2.7 The Deliberation Level......Page 471
3 Seed AI......Page 486
3.1 Advantages of Minds-in-General......Page 490
3.2 Recursive Self-enhancement......Page 494
3.3 Infrahumanity and Transhumanity: "Human-Equivalence" as Anthropocentrism......Page 499
4 Conclusions......Page 503
References......Page 506
B......Page 512
E......Page 513
I......Page 514
O......Page 515
S......Page 516
W......Page 517