Which History of AI

When it comes to learning about the history of artificial intelligence, there are a lot of different options out there. But which one is best for you? In this article, I’ll compare four of the most popular books on the subject:

  1. “A Brief History of Artificial Intelligence: What It Is, Where We Are, and Where We Are Going” by Michael Wooldridge
  2. “Darwin Among The Machines: The Evolution Of Global Intelligence” by George Dyson
  3. “The Quest for Artificial Intelligence” by Nils J. Nillson
  4. “Artificial Intelligence: A Modern Approach” by Stuart Russell and Peter Norvig

All of these books offer a view at the history of artificial intelligence, tracing its origins back to the early days of computing. I sorted them broadly from “most superficial” to “most in-depth”, and give an idea of what they provide you with and what level of knowledge and time you should bring with you.

A Brief History of Artificial Intelligence: What It Is, Where We Are, and Where We Are Going

The Time-Saver

In this short book, Woolridge summarizes the successes and failures we have had in creating artificial intelligence. It is an introductionary book to AI, with an about even balance of history and current issues, and provides a readable and accessible introduction to how the field developed. Woolridge evidently has a lot of knowledge on the matter, and his writing style is easy to read; he doesn’t assume that the reader knows anything. It feels very similar to what you would read in the Economist. The book explains the evolution and issues of the field, and then the second half contains predictions about future developments.

The books has two main advantages from my perspective: a) that it is short, and b) that it is not difficult to understand for someone who has no knowledge of AI. The author covers a very brief version of AI history, ethical concerns, and current/possible abuses of AI. A thing I found less fitting is the last chapter, where the author touches into philosophy, discussing what consciousness is and how it could have developed over tens of millions of years. This is of course very speculative as the author concedes, but adds very little to the book as a whole. Furthermore, the book left me (with CS background) seriously wanting for more detailed information - though I am probably not the target audience of this book.

Read if you:

  • do not want to spend too much time
  • have little to no previous technical knowledge

Don’t read if you:

  • want more detailed insights into the workings of AI

Darwin Among The Machines: The Evolution Of Global Intelligence

The Non-Fiction

In his book, Dyson argues that the evolution of life on Earth is but a prelude to a much grander event: the evolution of intelligent machines. He envisions a future in which human intelligence is surpassed by that of artificial intelligence, and in which machines take on many of the tasks currently performed by humans. Dyson is optimistic about this future, believing that it will bring about greater efficiency and progress. However, he also acknowledges the potential risks involved, and calls for caution and thoughtful regulation.

This book does a good job of collecting historical information on the evolution of computers and artificial life. Each chapter in the book explores a single topic, starting from when it was first introduced in someone’s writings, up to the 1990s. The book is more about artificial life or artificial intelligence than a computer-science book, and how topics like evolution, encoding, hardware, software, and networking are all connected under a common theme. I understand why some of the other reviewers were disappointed in the lack of technical detail, but I don’t think that’s what the book was trying to achieve.

I read the book more as a broad history of the growth of the idea, with the flow of time and growth of the idea as the focus and theme. The book is well written and has sparked my interest to look up the original works behind a few areas of discussion. Like many writers in the 1990s, Dyson expected big results in artificial life over the coming years, which did not come to pass. However, as a comprehensive history of the fields behind artificial life, this book is well worth a visit.

Read if you:

  • are interested in how big topics like evolution, hardware, software and AI are connected via their historical roots

Don’t read if you:

  • expect an in-depth examination of Artificial Intelligence from a Computer-Science point of view

The Quest for Artificial Intelligence

The Computer-Scientist

The Quest for Artificial Intelligence covers both the history and future of AI. Nilsson does an excellent job of explaining the science behind AI in a way that is accessible to the layperson, and discusses the ethical implications of AI and its potential impact on humanity. Nilsson covers the major figures and ideas in AI, from its early days in the 1950s through the present day. He does a good job of explaining the key concepts and ideas in AI, and how they have evolved over time. He also touches on the major applications of AI, including expert systems, natural language processing, and robotics. Interestingly, Heuristic search, planning, and other areas of AI have seen contributions by the author himself. In the panorama of issues, problems, and large scale questions that have shaped AI, Nilsson gives his own work fair coverage.

This book feels for example fitting for aspiring software engineers which wish to develop algorithms that simulate the steps of a thought process. It helps you understand:

  • What is currently possible, what is not possible and what is nearly possible.
  • What approaches have hit dead ends and what alternative approaches superseded them.
  • What subcategories of AI research exist and how they can be integrated.
  • What areas of AI research are being actively investigated today and show promise of further advances.
  • How we stand upon the shoulders of giants, and how many amazing programmatic investigations took place before most script kiddies and raving transhumanists were even conceived.

Read if you:

  • want to understand important steps along the way to AI from a Computer-Science Point of View
  • are interested in more detailed walkthroughs of AI mechanics

Don’t Read if you:

  • are looking for meditations on deeper connections between other fields of science
  • do not want to read about more detailed explanation of AI algorithms

Artificial Intelligence: A Modern Approach

The Textbook

If you really want to know what is going on, there is no way around this textbook Stuart Russell and Peter Norvig. It is used in over 1400 universities worldwide for a good reason. It is also very exhaustive, with the authors stating that it takes about two semesters to cover every chapter - though you can consider yourself at least something of an expert if you understood everything you read. Even though it is exhaustive and used as standard textbook, I found the book to be very accessible. If you’re familiar with the Bayes theorem and the normal distribution, you should be able to understand the book’s mathematical side.

All classical problems of AI such as decision making, search, prediction, and filtering are described in depth. The history of each problem is discussed, as well as the different solutions that can be applied depending on the real world constraints. Here, the book does a great job of explaining the classical problems of AI and the different dimensions of each problem. The history of each problem is also discussed, which I at times found to be nearly more inspirational than the main narrative.

In my opinion, the major accomplishment of AIMA is that Russell and Norvig take the hodge-podge of AI research and manage to fit it sensibly into a narrative structure centered on the notion of different kinds of “agents” (not to be confused with that portion of AI research that explicitly refers to its constructs as “agents!”). This textbook tends to perfection, with no stone left unturned.

As of now however, the next edition looks overdue (sentences like the following are just outdated: “Current Go programs play at the master level on a reduced 9 × 9 board, but are still at advanced amateur level on a full board”).

Read if you:

  • want to know it all and have the perseverance to show it
  • are familiar with basic mathematical concepts like normal distributions and the Bayes theorem

Don’t read if you:

  • want a brief overview
  • do not bring mathematical knowledge to the table and don’t want to learn it