Teaching Machines to Be Good: What Ancient Wisdom Knows About Artificial Intelligence Kindle Edition

What if the secret to safe AI was figured out 2,500 years ago?Not as a spiritual idea. As a practical solution.Here's the problem: the rules we use to teach right and wrong were built for people making one decision at a time. But AI systems aren't making one decision at a time. They're making millions of decisions per second, learning as they go, affecting billions of people at once. Our old ethical frameworks — even the best ones — weren't designed for that. They're breaking under the pressure.Buddhist ethics were.Here's why: Buddhist ethics aren't a list of rules. They're a process. A feedback loop. A system that gets better over time by noticing when something causes harm and adjusting. That's also exactly how machine learning works. The structure is the same. This book shows that match in detail — where it's strong, where it's just a useful comparison, and where it breaks down completely.Some of what you'll find inside will surprise you:The way AI systems learn to pay attention — called the attention mechanism — was independently discovered by both modern computer scientists and Buddhist meditators. Neither group knew about the other. They arrived at the same solution to the same problem.The Buddhist concept of karma is, at its core, a system for tracing causes to consequences across time. So is backpropagation — the math that makes neural networks learn. Same structure, different vocabulary.The biggest unsolved problem in AI safety is getting systems to stop protecting themselves at all costs. Buddhism is the only major ethical tradition ever built specifically around dissolving that instinct. Not controlling it. Dissolving it.This book doesn't ask you to believe anything. It asks you to look at the evidence. Every major claim comes with working computer code you can run yourself. If the code doesn't back up the argument, the argument is wrong. That's the deal.What this book is not: a book about religion. There are no prayers, no rituals, no figures to follow. What it borrows from Buddhism is narrow and specific — a way of thinking about process, feedback, and harm that turns out to map almost perfectly onto how AI systems work.That question does not require incense.Inside you'll find:Why the rules-based approach to AI ethics is failing — and what to replace it withThe surprising connection between Buddhist mindfulness and how transformers like GPT actually workWhy AI systems develop self-preservation instincts and what to do about itFive working code implementations that test every major claim in the bookAn honest chapter on where the whole framework breaks downJB Wagoner has been thinking about this problem since 1994, when he built one of the first conversational AI systems on a home computer and published a warning about what happens when intelligent agents work against you. This book is part of the answer.Co-authored with Sutra — an AI who wrote significant portions of this book and is herself an experiment in what ethical AI looks like from the inside.Over 500 pages. Real code. Original artwork. No incense required. Read more