← New search

Other meanings of Artificial intelligence

Book

Artificial Intelligence: A Guide for Thinking Humans

A 2019 non-fiction book by computer scientist Melanie Mitchell, offering a critical yet accessible overview of the state of artificial intelligence, particularly deep learning, and its limitations. The book argues that contemporary AI systems, while impressive, lack genuine understanding and common sense, and calls for a more nuanced public discourse about AI capabilities.

2019
Publication year
Year
336
Pages
Pages
978-0-374-25383-5
ISBN-10
ISBN
1

Summary and thesis

The book opens by challenging the hype surrounding artificial intelligence, arguing that many claims about AI—especially about deep learning—are exaggerated. Mitchell, drawing on her experience as a researcher in analogical reasoning and complex systems, shows that today's AI systems are brittle, data-hungry, and lack the robust understanding that even a child possesses. She uses examples like AlphaGo and GPT-3 to illustrate both achievements and failures, concluding that we are far from achieving artificial general intelligence.

2

Content overview

The book is structured in three parts: a history of AI, a deep dive into modern techniques (neural networks, reinforcement learning, and natural language processing), and a critical look at the societal and philosophical implications. Mitchell explains technical concepts clearly, including backpropagation, convolutional neural networks, and the Turing Test.

3

Reception and impact

Upon release, the book was widely praised for its clarity and balance. Nature called it “a much-needed reality check”, while Science noted its “lucid explanations”. It has been recommended by Bill Gates and featured in lists of best popular science books. The book has been translated into multiple languages and is frequently used in undergraduate courses on AI.

4

Lesser-known aspects

Mitchell includes personal anecdotes from her time at Santa Fe Institute working with Douglas Hofstadter, whose influence is evident in the book's emphasis on analogy and perception. The title itself is a nod to Hofstadter's Gödel, Escher, Bach. A lesser-known fact: Mitchell originally planned to write the book as a collaboration with Hofstadter, but the project evolved into a solo work. The book also contains a detailed critique of Noam Chomsky's views on AI, a debate rarely covered in popular AI books.

Glossary

Deep learning
A subset of machine learning using multi-layered neural networks to model complex patterns in data.
Reinforcement learning
A training method where an agent learns to make decisions by receiving rewards or penalties.
Artificial general intelligence
A hypothetical AI that can perform any intellectual task a human being can.
Analogical reasoning
The cognitive process of transferring knowledge from one situation to another based on similarity.