Just Predicting Words: How ChatGPT, Claude, and Modern AI Actually Work

In November 2022, ChatGPT arrived and the world tilted. Three years later, AI is everywhere, the trillion-dollar companies have chosen sides, and most people still cannot explain in plain English what these systems actually do.This book is the calm explanation. How tokenization works. How attention works. Who OpenAI and Anthropic and Google and the Chinese labs really are, and how their choices shape the technology. Where the risks are real, where they are hype, and what to do with all of it.No buzzwords. No doom. No utopia. Just the machine, the companies that built it, and the world it is remaking — explained the way a smart friend would, over coffee.THE 2026 ANNUAL EDITIONUpdated yearly as the field evolves.---What you'll learn:- How language models actually work, step by step, in plain English- Why these systems make stuff up and what to do about it- The real players: OpenAI, Anthropic, Google, Meta, the Chinese labs- What AI can do well right now, and where it fails- The risks that are real, and the ones that are mostly hype- How regulation is shaping the future- Practical advice for using these tools at work, with your kids, in your lifeWritten by Ziad El Feghali, a technologist and educator who teaches at the Holy Spirit University of Kaslik. Designed for the intelligent reader who wants to understand AI without needing a computer science background. Read more