AI Storytelling – Is it any good? I entered a simple prompt into ChatGPT. Yes, I could have been more specific about what I wanted—better prompts make for better output from the AI—but my intention was to analyze the writing and storytelling. Given few constraints, how good is AI storytelling?

 

Write a short story about someone who uses artificial intelligence to write an important message to a loved one.

 

The AI returned the following result:

 

Mark had been thinking about his late wife, Anna, a lot lately. It had been three years since she passed away, and he missed her every day. They had been married for 30 years, and he had never loved anyone as much as he loved her.

One day, Mark decided to write a letter to Anna, but he didn’t know how to express his feelings. He wanted to say so many things, but he didn’t know where to start. That’s when he had an idea: he could use artificial intelligence to help him write the letter.

Mark sat down in front of his computer and started typing. He wrote about how much he missed her, how he still remembered the sound of her voice and the way she smelled. He wrote about the good times they had shared and how much he regretted not being able to spend more time with her.

As he wrote, an AI program analyzed his text and suggested changes to improve the message’s impact. It provided synonyms to enhance the vocabulary and helped Mark structure the message in a more cohesive and coherent way.

Mark continued to write, pouring his heart out onto the screen. The AI program kept analyzing and suggesting changes, making the letter more heartfelt and emotional with each iteration.

After a few hours, Mark had a beautiful letter that captured all of his feelings for Anna. He printed it out and placed it on her bedside table, just like he used to do when she was alive.

As he looked at the letter, Mark felt a sense of closure. He knew that Anna couldn’t read it, but writing it had helped him come to terms with his loss. He knew that he would never forget her, but he could now move on with his life, knowing that he had said everything he needed to say.

 

The premise is interesting. Of all the plotlines the AI could have come up, one has to wonder how the algorithm chose writing a letter to a deceased spouse as the best example of “an important message.” It’s not a bad choice—just an interesting one. What are ChatGPT’s criteria for “importance” in AI storytelling?

The style of the AI storytelling is very much “on the nose” a list of functional factoids. Each sentence makes a statement, adds detail, advances the plot, and moves toward the conclusion. Below are the stripped-down sentences. Because there’s little variety in the syntax, the prose tends to be hypnotically repetitive.

 

Mark had been thinking…

It had been three years…

He missed her…

They had been married…

He had never loved anyone…

Mark decided…

He didn’t know…

He wanted to say…

He didn’t know…

(That’s when) he had…

He could use…

 

Though the story structure (conflict to transformation) is there, the prose reads more like a list than a narrative. Almost every sentence starts with “He…” and continues on with not-very-well-chosen verbs. Verbs are the spices in the writer’s literary kitchen. Verbs of being (is, has, was, had, have, etc.) merely assert existence. Verbs of doing (explore, inspire, discover, realize, etc.) carry emotional energy. None of the underlined verbs below are especially potent or evocative.

 

Mark sat down in front of his computer and started typing. He wrote about how much he missed her, how he still remembered the sound of her voice and the way she smelled. He wrote about the good times they had shared and how much he regretted not being able to spend more time with her.

 

Yawn! Here’s my quick take on the same content:

 

Mark had read about AIs that could create content, but could a computer really help him express his feelings? If Anna were to read my letter, would she be touched by my grief and appreciate my affection … or would she think I’d failed her by asking a robot to speak for my heart?

“Might as well give it a go,” said Mark out loud. He began to type, not worrying about whether his prose was eloquent or his grammar correct. Happy memories and deep sorrow flowed through his fingers onto the screen. He knuckled away a tear and kept writing.

 

Not all the verbs are aspirational but the syntax varies. Spoken and unspoken dialogue allows the character to tell the story. And instead of saying “Mark felt sad” as the AI would have done, Mark “knuckles away a tear” to let the reader know that without spelling it out. In this updated version, we experience Mark’s emotions—his skepticism and his grief—instead of being told what he felt.

Though it’s impressive that an artificial intelligence algorithm can aggregate content into a cohesive narrative that dovetails with human intentions, it’s a story written by a robot and it sounds like one—even if the poignant outcome might distract some readers from the mediocrity of the prose.

And though it might make commercial sense to ask a machine to write for us in some instances, it’s worth it for all of us to embark on the writer’s journey, master the art of the message, and discover our power to express ourselves.

Among the greatest of human callings is our desire to be listened to and understood. Writers driven by the search to find and share meaning will discover the eloquence and authenticity to achieve connection. Unlike Mark in our fictional, AI storytelling narrative, those who ask a machine to speak for their heart are likely to disappoint their readers.