1.8 Overview of the Book

The rest of the book explores the design space defined by the dimensions of complexity. It considers each dimension separately, where this can be done sensibly.

Part I considers the big view of agents as a coherent vision of AI.

Chapter 2 analyzes what is inside the black box of Figure 1.4 and discusses the modular and hierarchical decomposition of intelligent agents.

Part II considers the case of no uncertainty, which is a useful abstraction of many domains.

Chapter 3 considers the simplest case of determining what to do in the case of a single agent that reasons with explicit states, no uncertainty, and has goals to be achieved, but with an indefinite horizon. In this case, the task of solving the goal can be abstracted to searching for a path in a graph. It is shown how extra knowledge of the domain can help the search.

Chapters 4 and 5 show how to exploit features. In particular, Chapter 4 considers how to find possible states given constraints on the assignments of values to features represented as variables. Chapter 5 presents reasoning with propositions in various forms.

Chapter 6 considers the task of planning, in particular determining sequences of actions to solve a goal in deterministic domains.

Part III considers learning and reasoning with uncertainty. In particular, it considers sensing uncertainty and effect uncertainty.

Chapter 7 shows how an agent can learn from past experiences and data. It covers the most common case of learning, namely supervised learning with features, where a function from input features into target features is learned from observational data. Chapter 8 studies neural networks and deep learning and how features themselves can be learned from sensory observation.

Chapter 9 shows how to reason with uncertainty, in particular with probability and graphical models of independence. Chapter 10 introduces learning with uncertainty. Chapter 11 shows how to model causality and learn the effects of interventions (which cannot be learned from observation alone).

Part IV considers planning and acting with uncertainty.

Chapter 12 considers the task of planning with uncertainty. Chapter 13 deals with reinforcement learning, where agents learn what to do. Chapter 14 expands planning to deal with issues arising from multiple agents.

Part V extends the state and feature-based representations to deal with relational representations, in terms of relations and individuals.

Chapter 15 shows how to reason in terms of individuals and relations. Chapter 16 discusses how to enable semantic interoperability using knowledge graphs and ontologies. Chapter 17 shows how reasoning about individuals and relations can be combined with learning and probabilistic reasoning.

Part VI steps back from the details and gives the big picture.

In Chapter 18 on the social impact of AI, further ethical and social concerns are addressed, by considering various questions, such as: What are the effects, benefits, costs, and risks of deployed AI systems for society? What are the ethical, equity, and regulatory considerations involved in building intelligent agents? How can you ensure that AI systems are fair, transparent, explainable, and trustworthy? How can AI systems be human-centered? What is the impact on sustainability?

Chapter 19 reviews the design space of AI and shows how the material presented can fit into that design space. It also considers some likely and possible future scenarios for the development of AI science and technology.