You Look Like a Thing and I Love You
by Janelle Shane
Janelle Shane's hilariously enlightening exploration of artificial intelligence through her AI Weirdness experiments
"AIs don't understand the problems you want them to solve".
Editorial Summary
Optics research scientist Janelle Shane demystifies artificial intelligence through her signature blend of humor and rigorous analysis in this accessible exploration of machine learning's quirks and limitations. Drawing from her popular AI Weirdness blog, Shane documents her experiments training neural networks to generate pickup lines, paint colors, recipes, and Halloween costumes, revealing how artificial intelligence actually works beneath the hype. The book examines real-world applications from Tesla Autopilot to Google Translate, while Shane's "Five Principles of AI Weirdness" expose the fundamental ways algorithms misunderstand human problems. Through cartoons and pop culture references, Shane shows how artificial intelligence takes shortcuts, fails at creative tasks, and reflects both human ingenuity and folly.
Perspective
"You Look Like a Thing and I Love You is the book that makes you laugh at AI and then realize, mid-laugh, that you've just understood something important — Shane's neural network experiments are genuinely funny and genuinely illuminating about how machine learning actually works. The distinctive contribution is the AI Weirdness method: by showing what AI produces when it fails, Shane reveals the underlying structure of how these systems represent the world more clearly than any success story could. Readers who want an honest, technically grounded, genuinely entertaining introduction to machine learning's actual capabilities and limitations will find this the most effective demystification available."
Matched by concept and theme
