Psychological Modeling and Testing of AI Systems
This book explores the subject of artificial psychology and how the field must adapt human neuro-psychological testing techniques to provide adequate cognitive testing of advanced artificial intelligence systems. It shows how classical testing methods will reveal nothing about the cognitive nature of the systems and whether they are learning, reasoning, and evolving correctly; for these systems, the authors outline how testing techniques similar to/adapted from human psychological testing must be adopted, particularly in understanding how the system reacts to failure or relearning something it has learned incorrectly or inferred incorrectly. The authors provide insights into future architectures/capabilities that artificial cognitive systems will possess and how we can evaluate how well they are functioning. It discusses at length the notion of human/AI communication and collaboration and explores such topics as knowledge development, knowledge modeling and ambiguity management, artificial cognition and self-evolution of learning, artificial brain components and cognitive architecture, and artificial psychological modeling.
Advertisements