Will Machines Tell Better Stories than Humans?

A manager at PayPal once handed his team a sheet listing emotions. Before any presentation, he asked them to circle two things they wanted their audience to feel. Mike Todasco initially found the exercise excessive. Surely a routine product update did not need that much thought?

Then it changed how he approached the meeting. Was the aim simply to inform people, or should they leave excited, concerned or ready to act? The facts might be the same, but the reason for communicating them became clearer.

In this conversation with Dr. Christian Mühlroth, Todasco connects that experience to storytelling, creativity and the adoption of AI. Drawing on his work in innovation at PayPal and his subsequent exploration of AI and writing, he makes a practical case: helping people understand and care about a new possibility is part of the work of innovation, not a presentation task added at the end.

Make the problem specific enough to remember

Todasco remembers working on a product for financially underserved customers. The team had research and customer interviews, but many employees had never personally used the services their prospective customers depended on.

He and a colleague decided to visit a check-cashing business and experience a short-term loan themselves. Todasco arrived with assumptions about what the experience would be like. Afterwards, the pair brought their account back to the team.

What stayed with colleagues was the fact that two people they knew had actually done it. The story gave their observations attention and credibility in a way another summary of research had not.

It was not a replacement for understanding many customers. It made an existing problem concrete. For a product or R&D team, that distinction matters: a personal account can help people connect with evidence without pretending that one experience represents everyone.

Todasco's advice on what makes a story memorable is direct:

“Being personal and being specific are the two things.”

Tell the story for the person listening

The emotions exercise is useful because it forces a decision about the audience. A potential customer, an investor and a prospective employee need different reasons to care about the same company.

Todasco argues that the story should change accordingly. What does this person need? What problem are they trying to solve? Which parts of the explanation will help them see a worthwhile next step?

He applies that reasoning even to a job interview. The candidate is trying to demonstrate how their experience can solve a problem for the employer. Relevant examples carry more weight than an undirected account of everything the candidate has done.

In an innovation project, the same discipline can improve a discussion with engineers, executives or customers. Keep the underlying facts consistent, but choose the detail and emphasis that help each group make its decision. A compelling vision does not remove the need to understand the listener's practical concerns.

Good writing happens after the first draft

Todasco returned to education to study writing. One of the most valuable parts was sharing a story he thought was excellent with classmates who were willing to explain what did not work.

That feedback changed the result. Writing involved revisiting earlier sections, recognizing missing connections and restructuring the story. The process was nonlinear and collaborative.

He contrasts that with asking an AI model to produce an entire story in one attempt. At the time of the conversation, he was testing AI-written books under a disclosed pen name. He was unimpressed by their quality, describing the results through an expression borrowed from his teenage son: painfully average.

The useful distinction is between generating a draft and developing a story. A fast first version still needs judgment about structure, relevance and what a reader will understand. Speed does not make those decisions disappear.

Use AI as someone to think with

Todasco sees more promise in AI as a brainstorming partner. While developing a science-based book idea, he used a voice interface during walks around a lake. He described the concept, asked questions and invited the system to ask questions back.

At the end, he requested a summary and an outline of the discussion. The result captured an exchange he could continue working with, rather than handing him a supposedly finished book.

He also questions whether technical performance alone determines what people value. His example is basketball: the appeal of watching an exceptional human player includes the person, the uncertainty and the story around the performance. A machine performing the task perfectly would offer a different experience.

For creative work, his point is that intention and human context still matter. Tools can expand what someone can produce while leaving important choices with the person directing the work.

Let small experiments tell the adoption story

When the discussion turns to corporate AI adoption, Todasco resists the idea that distributing tools across a large workforce will automatically produce value. Some employees will try once, get a disappointing answer and stop.

His alternative begins with small teams and specific opportunities. Find people in different functions who can identify mundane work they would like to improve. Help them experiment, observe what succeeds and share what fails. The first useful applications need not come from a technology department.

Then give those teams a platform to explain what happened. Todasco suggests a short slot at a company all-hands meeting where employees can tell their peers what they tried and learned.

This is a proposed approach, not a claim that every AI pilot will work. He explicitly acknowledges errors and bias. The aim is to build confidence through visible experience, including limitations. A colleague explaining a useful experiment gives the next team something more concrete than another announcement that the company is embracing AI.