Continue reading
Sign in to read this chapter.
Your reading progress and saved items will be available after sign-in.
Inside Modern AIFront matter
Contents
- Front matter
- Who This Book Is For and How Deeply to Read It
- How to Read Vendor Claims
- Part I · The New Capability Layer
- 1Long Context Is Not Memory
- 2Reasoning Is More Than Thinking Longer
- 3One Model, Many Kinds of Work
- 4From Answering Questions to Doing Research
- Part II · Turning Intelligence Into Action
- 5The Model Is Not the Product
- 6Tool Use: How AI Moves From Knowing to Doing
- 7Computer Use: When the Interface Becomes the API
- 8When an AI Action Leaves the Screen
- Part III · Work That Lasts Longer Than One Turn
- 9State, Checkpoints and Recovery
- 10Steering a Model While It Is Working
- 11From One Agent to a Team of Agents
- 12Permissions and Safety
- Part IV · Knowing Whether the Intelligence Can Be Trusted
- 13Evaluation and Self-Verification
- 14Can We See What the Model Is Thinking?
- 15Astra and Fable: Different Engineering Choices
- Part V · What This Means for You
- 16What This Generation Makes Possible
- 17What You Should Do Next
- Back matter
- What Each Vendor Documents
- References and Source Notes