Artificial Intelligence: A Modern Approach Book Review – Master Ethical AI for Family Resilience in 2025

Part I: Artificial Intelligence

• Artificial Intelligence (AI): The study of systems that perceive, reason, act, and learn to achieve goals.
• Intelligent Agent: An entity that perceives its environment and takes actions to maximize success.
• Rational Agent: An agent that acts to achieve the best possible outcome based on its knowledge.
• Task Environment: The problem space in which an agent operates, defined by properties like observability and determinism.
• Turing Test: A method to evaluate if a machine can exhibit human-like intelligence through conversation.

Part II: Problem-solving

• Search Algorithm: Methods for finding solutions in problem spaces, e.g., breadth-first or depth-first search.
• Heuristic Search: Informed search using estimates to guide toward solutions, like A* algorithm.
• Adversarial Search: Search in competitive environments, such as minimax for games.
• Alpha-Beta Pruning: Optimization technique to reduce branches in game tree search.
• Constraint Satisfaction Problem (CSP): Problems defined by variables, domains, and constraints, solved via backtracking.

Part III: Knowledge, Reasoning, and Planning

• Propositional Logic: A system for representing and reasoning with true/false statements.
• First-Order Logic (FOL): Extends propositional logic with quantifiers and predicates for more expressive knowledge.
• Inference: Deriving new knowledge from existing facts, e.g., resolution or forward chaining.
• Knowledge Representation: Structures for encoding information, like semantic networks or ontologies.
• Automated Planning: Generating sequences of actions to achieve goals, using methods like STRIPS.

Part IV: Uncertain Knowledge and Reasoning

• Probability Distribution: Mathematical description of uncertainty in events or variables.
• Bayesian Network: Graphical model for representing probabilistic dependencies.
• Hidden Markov Model (HMM): Probabilistic model for sequences with hidden states, used in time-series reasoning.
• Markov Decision Process (MDP): Framework for decision-making under uncertainty with states, actions, and rewards.
• Utility Function: Measures preferences or values for outcomes in decision theory.

Part V: Machine Learning

• Supervised Learning: Learning from labeled data to predict outcomes, e.g., decision trees.
• Unsupervised Learning: Finding patterns in unlabeled data, like clustering.
• Reinforcement Learning: Learning optimal actions through trial-and-error with rewards.
• Deep Learning: Multi-layered neural networks for complex pattern recognition.
• Convolutional Neural Network (CNN): Neural architecture for processing grid-like data, e.g., images.

Part VI: Communicating, Perceiving, and Acting

• Natural Language Processing (NLP): Techniques for machines to understand and generate human language.
• Transformer Model: Architecture for sequence tasks like translation, using self-attention.
• Computer Vision: Enabling machines to interpret visual data, e.g., edge detection or object recognition.
• Robotics: Integration of perception, planning, and action for physical agents.
• Localization and Mapping (SLAM): Simultaneous estimation of position and environment map.

Part VII: Conclusions

• AI Ethics: Principles addressing fairness, bias, transparency, and safety in AI systems.
• Existential Risk: Potential long-term threats from advanced AI misalignment with human values.
• Human-Compatible AI: Designing AI to align with human intentions and well-being.

Additional Concepts (Cross-Cutting)

• Overfitting: When a model learns noise instead of patterns, reducing generalization.
• Back-Propagation: Algorithm for training neural networks by adjusting weights.
• Q-Learning: Model-free reinforcement learning method for estimating action values.
• Game Theory: Study of strategic interactions, applied to multiagent systems.
• Ontological Engineering: Building shared conceptualizations for knowledge domains.