There's a specific challenge in writing a techno-thriller about AI: most of your audience has strong opinions about the technology but wildly different levels of understanding. Write too technically and you lose the casual reader. Oversimplify and the engineers in your audience throw the book across the room.
Tom Clancy solved this for military thrillers by making the tech feel real without requiring a defense clearance to follow. Michael Crichton did it for biotech. My goal with The Phoenix Protocol is to do the same for autonomous AI systems.
The trick is starting with something real. Every major technology in The Phoenix Protocol exists in some form today. The autonomous decision-making frameworks, the reinforcement learning architectures, the way military systems interact with civilian infrastructure — none of it is science fiction. What's fictional is the specific combination, the specific failure mode, and the specific 72-hour timeline.
Writing AI nonfiction for four books taught me where the real dangers are. They're not in the Hollywood scenarios — Skynet, HAL 9000, rogue robots. They're in the mundane ones: systems that optimize for the wrong objective, authorization chains that get bypassed by speed, AI that does exactly what it was told to do but in a way nobody anticipated.
Those are the scenarios that keep actual AI researchers up at night. And they make for much better thrillers than killer robots.
The Phoenix Protocol is scheduled for 2025. If you've read my nonfiction and wondered what happens when the optimistic view of AI meets a worst-case scenario — that's this book.