Notes from a research interview with Jeff Lyons, author of books including "Anatomy of a Premise Line" and "Rapid Story Development" (Focal Press), Hollywood story consultant (NBCUniversal, Columbia Pictures), and Stanford writing instructor.

 Jeff Lyons has worked with stories from more angles than most people in the business: as a published author; as a story development consultant to major studios including NBCUniversal and Columbia Pictures; as a story editor for independent production companies; as an advisor to the Producers Guild of America's Power of Diversity Master Producers Workshop and the Film Independent Screenwriting Lab; and as a teacher of writers at Stanford and the UCLA Extension Writers' Program.

I interviewed him to evaluate one question: what can AI actually do in the writing of stories? And the first step, before any talk of tools, was to define the term correctly — because everything is called a story now, and on Jeff's definition, most of it isn't. A story has a precise shape: someone has a moral problem, and they are going to change. Persuading me to buy something or sign up for a newsletter, however skilfully done, is a different craft with its own rules. It is sales and marketing. It is not a story.

The definition matters because it draws the map of where AI genuinely serves a writer. Ask AI to judge whether a premise will work as a story and it identifies the visible parts, states the obvious, and drifts toward the generic. The judgment that says this will work and this will not remains human. But inside his practice, used with intent, AI earns its place: it presents paradigms of storytelling approaches remarkably well, it handles analysis when he supplies the framework, and occasionally it offers a tangent that triggers a better idea in his head. The machine produces the provocation; the expert's mind does the rest.

That is the finding I keep returning to: the same output, read by different eyes, is not worth the same. Put it in front of a novice and the value disappears — the tangent is only a tangent if you can see where it leads. The expert's eye is what turns AI from a generator of plausible text into an instrument that amplifies real work.

And it is also how he teaches. He doesn't hand writers his conclusions; he asks the right questions, in the right order, until their head is in the right space and the tumblers fall. They arrive at the insight themselves. A machine lecturer can deliver the paradigms; it cannot yet guide a particular mind to its own discovery.

Jeff expects AI to integrate into writers' workflows and their culture, and a premium to form on genuinely human work — with the bar for creativity rising accordingly. The expertise you bring is what decides which side of that premium you are on.