Gabrielle Dolan Blind-Tested Her Stories Against AI's

Gabrielle Dolan writes stories for a living and teaches other people to tell them, so she ran an experiment to test her own advice: she wrote a story for each of six subjects, gave the same six subjects to ChatGPT and a second AI tool, then had readers blind-rate all eighteen stories without knowing which were hers. Her stories beat the AI's in four of six match-ups. One came out even. AI won once. The scoreboard wasn't the interesting part of the results. Where the points landed was.

That experiment sits at the center of Dolan's latest book, Story Intelligence: The Craft of Authentic Storytelling Made Smarter with AI, and of a conversation she had with Students Incorporated co-host Josiah Ann Proud about what more than two decades of teaching storytelling has taught her that AI still hasn't caught up to.

Eighteen Stories, One Missing Ingredient

Dolan had readers score every story on categories like clarity (did the story make sense) and authenticity (did it feel real). For her own stories, the two scores landed close together: a story that made sense also felt true. For the AI-written stories, clarity consistently scored well above authenticity. The stories tracked logically; they just didn't land as real, and readers kept reaching for the same word to explain why, even before they knew which stories were AI-generated. "It just felt cliché," one told her. Another said it "didn't seem real." When Dolan later revealed which stories had been written by AI, one reader admitted rating a story generously because "I didn't want to be mean to the person," not realizing there wasn't one.

"It's a human connection. It's the energy."

That's how Dolan describes the quality the AI stories kept losing points on. She's clear that this isn't an anti-AI position. She used AI as a creative partner while writing the book itself, including to brainstorm the title. But the test convinced her that AI can produce a story that reads fine and still be missing the one thing a story is supposed to do.

Dolan's blind test isn't an isolated finding. A growing body of academic research on what some researchers call the AI-authorship effect has found something close to a mirror image of her result: when readers don't know a piece of writing came from a machine, they often rate it just as authentic as human work, sometimes more so, but the moment the source is disclosed, ratings for the AI version drop and the human-written version pulls ahead, an effect one 2025 study tied to something close to moral disgust at the idea of an emotional message coming from something that never actually felt anything. Dolan's experiment ran that same logic in reverse. Her readers didn't know which stories were AI's until after they'd already scored the authenticity gap themselves, meaning the drop-off showed up before anyone knew there was a machine to distrust.

Twenty-One Years Making a Business Out of It

Dolan spent years in corporate Australia in senior leadership and change-management roles before she noticed something: when she explained the reasoning behind organizational change through stories instead of logic, people actually understood the message. Twenty-one years ago, with two children at home, she left to teach storytelling full time, telling herself she could always go back to a regular job if it didn't work out. She's since published eight books on the subject, a fact she finds funny given that she failed English in her final year of school.

The Story That Got a Risk Team to Listen

To show what she means by a story doing real work, Dolan told the one she considers her best example: a client named Rosemary, head of a corporate risk team, had spent months failing to convince business units that managing risk was their job, not just hers. Case studies didn't work, and neither did business examples. So Rosemary tried something else: she told her audience about a hot day on the farm where she grew up, when her mother sent her to grab her bike from the front gate and she froze because a massive copperhead snake was lying in front of it. She remembered exactly what her mother had taught her: freeze, stay still, back away slowly. She made it back to the house safely, then connected it to the point: "All I can do is give you the skills, knowledge, and advice, so when you come across your own copperhead snake, regardless of what that looks like, you will know what to do."

Dolan tests every story against three questions, and she ran them live on air: did it help you understand the message? Yes. Will you remember it? Yes. Could you retell it to someone else without losing the meaning? Yes. That, she says, is the whole case for storytelling in one exchange: a good story doesn't just explain a message, it makes the message sticky enough to repeat.

The Mechanics: Short, Named, Felt

Dolan's tactical rules are specific. Keep a story to 60, maybe 90 seconds; beyond two minutes, an audience starts silently thinking "get to the point." Skip the throat-clearing: don't announce "let me tell you a story," since the phrase itself makes people brace to disengage. Open with time and place instead, like "when I was a kid, I grew up on a farm," which signals a story is coming without the warning label, because people are wired to lean into a story once they recognize one starting. Name every character the first time they appear (her daughters are "Alex and Jess," not "my daughters"), with one exception: parents and grandparents keep their family titles. And say how you felt, not just what happened, the difference, she says, between a flat sequence of events and a story that actually taps into emotion.

Four Kinds of Stories, and a Starfish on a Beach

Dolan's framework sorts usable stories into four types: personal stories (pulling a non-work moment into a business message, which she considers the most underused and most powerful), professional stories (the classic work anecdotes people reach for in interviews), public stories (borrowed case studies, like the Steve Jobs example everyone already half-knows), and parables, the Aesop's-fables category that can carry a message without any of it being personally true. She offered one of her own: walking on a beach with her daughter Alex, then in her early twenties and adrift about not having found her "purpose" yet, Dolan retold the parable of a man throwing stranded starfish back into the ocean while a fisherman tells him it's pointless, there are too many to save. He picks up one more and answers, "I made a difference to that one." Her advice to Alex distilled the same idea: figure out, every day, how to make a difference to one person, and it doesn't have to be bigger than offering to grab someone a coffee.

Using AI as a Coach, Not a Ghostwriter

Dolan's guidance for using AI in storytelling is narrower than "use it to write." She recommends prompting it to act as a storytelling coach, asking it to interview you with questions the way a person across a coffee table would, rather than asking it to generate a finished story from a few bullet points. If AI does hand back a full story, the job is to fact-check it against your own memory and then read it out loud, because a story can look right on the page and still not sound like you.

Structurally, she teaches a simple shape: get clear on one single message before anything else, since trying to carry more than one is a common mistake, open with time and place, and use the middle to decide, deliberately, what stays and what gets cut. Avoid ending on "the moral of the story is...," which she considers too directive; better to close with something like "imagine what we could achieve if..." and let the audience arrive at the point themselves.

Her closing advice: treat storytelling like a skill, because it is one. People who try it once without preparing and have it fall flat often conclude it "doesn't work" for them, when the real lesson, she says, is closer to learning golf. You don't get better by taking lessons forever; you have to go out and actually play.

Headlines: Likeness, Trust, and the Human Core

The episode's news roundup circled the same tension from three directions: YouTube's 2026 creator-likeness tools, which let creators authorize AI to re-render their videos in other languages using their own face and voice, with mandatory "altered content" labels and CEO Neil Mohan calling it a "creative accelerator" rather than a replacement for human imagination; Thai news director Dr. Natha Kamolwatin of The Standard, describing a shift in local newsrooms toward "signature storytelling," the deep, ethical reporting AI can't replicate, as more than 43 percent of Thai audiences move toward video-first news; and marketing executive Matt Salvedo, CEO of Amplify, arguing that even law is being reshaped by the same principle: AI can draft a brief, but it can't replicate what actually persuades a courtroom.

← Back to all posts