How AI Thinks

Author: Nigel Toon Year: 2024 Genre: Artificial Intelligence

About This Book

How AI Thinks provides an insightful look into the cognitive processes and decision-making mechanisms of artificial intelligence systems. Nigel Toon explores the inner workings of AI and how these systems process information and make decisions.

Key Insights

• Thinking is computation, not consciousness: Toon explains how learning algorithms “think” through optimisation and representation rather than human-like awareness.
• From symbols to statistics: Why the field shifted from classical symbolic AI to data-driven, probabilistic approaches—and what was gained and lost.
• Neural networks as function approximators: How deep nets learn features and decision boundaries, and where they break.
• Hardware matters: The economics of compute and parallelism driving today’s breakthroughs.
• Limits and failure modes: Generalisation gaps, brittleness, and reasons models can appear intelligent yet lack understanding.

Why I Recommend It

It demystifies modern AI without hand-waving. Toon balances accessibility with enough depth to make the intuitions “stick,” making it ideal for product leaders, engineers, and curious readers wanting a solid mental model of how contemporary AI works.

You’ll come away with clearer language for discussing capabilities and limits—and better judgement about where to apply AI productively.