This is a short piece I wrote using Penelope AI, followed by quick grammatical and syntax checks on the free tier of Grammarly. For the cover image, I used a basic prompt in Prompt Extend to make it more detailed and created several options in Karlo in about 17 seconds.

I was inspired by two sources to write this narrative:

• Investments Unlimited by Beal et.al. - This was an excellent and relatively short book providing the importance of standardized DevSecOps practices in the form of a narrative much like The Phoenix Project

• What 2026 looks like by Daniel Kokotajlo

I think stories and narratives have a place in tech writing and can convey a point or paradigm more powerfully than any kind of more objective writing. I hope this gives you an idea of where I think AI could get implemented into our lives in the near future.

I hope you all enjoy it!

---

"Emma, why don't you tell the class a little bit about your narrative piece and how you wrote it?"

Emma looked up from her computer in response to Professor Smith's question. Admittedly, she was distracted. She was reading up on the latest LitGPT4 model coming out of OpenAI and wondering if it was worth swapping out the model in her current workflow from Google's LitGen.

"Of course Professor! Give me a second to share my screen and I'll walk through my process" she exclaimed.

Emma quickly proceeded to the front of the class and shared her presentation to the projector through the school's casting service. She began to explain her narrative piece, taking the class step by step through her writing process. Her writing tools were pretty simple: Microsoft Word was up on the left of her screen but to the right were a few AI-enabled autocompletion tools with various toggles and prompt windows.

"So, for this piece, I knew I wanted to write something from the perspective of a young kid facing adversity. New York City has been a major American literary and cultural Mecca so I decided to start with a LitGen prompt of 'young kid living in 1970s New York City has to find a job, Ernest Hemingway, 20 pages.' What came out was pretty impressive but the tones mismatched the period since Hemingway was really writing a majority of his work about 50 years earlier."

Emma was a Senior at Carnegie Mellon University's English school, class of 2026. Like all the other students, she was eager to learn and succeed. She had heard of the new technology sweeping the world - Artificial Intelligence (AI) - and was determined to learn how to use it.

At first, Emma thought it would be an easy task. After all, she had been studying English for years; surely AI would just be a few more tools in her toolbox. She soon discovered that wasn't quite the case.

"What really stood out to you about the story produced by the AI?" Professor Smith asked.

"Well," Emma pondered, "I suppose the most interesting thing to me was that Litgen defaulted to a rich white Christian kid living in Manhattan as the subject of the narrative. The subject also spoke in a tone that lacked any kind of vernacular or colloquial language for the time and era. I thought this likely stemmed from a heavy initial model weight for Hemingway as the author and my model output analysis basically confirmed that."

She highlighted a selection of text referencing the character and a visualization tool opened up to her right. At the center of the visualization was her highlighted text, but extending out from it were various other terms and phrases. The term "Ernest Hemingway" was very bold and big relative to many other words.

She continued, "this was a narrative element I didn't necessarily agree with. Luckily with the context-agreement tool, I just had to tune a few things in the first few paragraphs and I had a more historically agreeable subject consistently applied through the rest of the narrative."

The AI-driven writing software that was emerging was far more sophisticated than anything she had ever seen before. It could analyze complex data, detect subtle nuances of language, and even generate its own creative ideas! Emma quickly realized that if she wanted to stay competitive in this new world of technology, she would have to learn how to use AI for writing.

"Another thing I wanted to tune was the ending. The current implementation of LitGen overwhelmingly outputs positive endings both due to the bias of the dataset as well as the bias of its developers at OpenAI to prevent it from being used for negative misinformation or poor user reactions. As a result, my narrative had a cheery and positive ending where the subject got a very well-paid job quite easily. I didn't think this fit the overall tone and theme I was trying to convey." Emma stated.

"In your final draft I read that didn't seem to be the case, what did you do to fix that?" asked Smith.

When AI-driven writing first started gaining popularity around 2022, professors across the United States initially did not like the usage of AI in their classrooms. Students who were motivated to just get a passing grade were copying their writing prompts directly into tools like GPT3 or ChatGPT and turned in entire AI-written essays. Some companies emerged producing software to find the essays that were written by AI, but open-sourced AI emerged that would just re-write the text based on a randomized number, resulting in a constant game of cat and mouse.

However, as time went on and students began to understand the nuances of AI-driven writing, these same professors began to see the potential of this technology and many of them began to actively encourage its usage in their classrooms. Professor Smith became a major trailblazer among his colleagues in English Literature academia when he introduced "AI-Enabled Fiction Writing", a 101 writing course produced in coordination with the Computer Science department.

Like Data Science classes that were popular as early as 2016, the class demonstrated to students that AI could be tightly integrated into their writing process without much coding or technical knowledge. Thanks to it's popularity, Emma has been able to attend the inaugural follow-on "Advanced AI-Enabled Fiction Writing" in her final semester at CMU.

"Good question, Professor." Emma started excitedly, "I thought this was going to be difficult and I would have to hand-correct it. It turns out that, luckily, Project Gutenberg conducted a huge initiative with Snorkel AI to programmatically label the outcome of all of their books by a selection of outcomes like "good" or "bad". More importantly, they also labeled those outcomes by various sentiments. I just went to their website and pulled all the collated text for all the books labeled 'Bittersweet'."

She began to get excited. She knew what she was about to say next was going to excite the class.

Emma smiled and continued, "I used the sentiment-tagged text from Project Gutenberg and enlisted the help of a friend from the Computer Science department to help me with this last part. OpenAI provides guidance for a method they call transfer learning of LitGen. That was a little out of my technical wheelhouse, but my CS friend was able to use my Bittersweet dataset to produce a new and improved LitGen model I've called 'LitGen_Bittersweet.' Re-running the text through the model and prompting for an ending was able to craft something more appropriate to my goals for my narrative. Originally, the subject inexplicably got a job as a successful banker. Now, the subject gets a more basic job as a dishwasher but still pines for the cushier jobs he has observed around the city."

Astonished murmurs rose from those around the class. Professor Smith was impressed with Emma's work and nodded in agreement as she finished her explanation.

"By the way," Emma continued "LitGen_Bittersweet is now available for everyone here through our private Academic OpenAI account. You should all be able to use it for story-tuning going forward if you'd like to incorporate more bittersweet elements into your writing."

"Very impressive Emma! It's amazing to see the progress Emma has made in using AI to enhance her writing," Professor Smith said, "I'm confident that this skill will prove invaluable as she begins her new job after graduation. I think we could all learn a little bit about going the extra mile by tuning our generative models with transfer learning, very impressive collaboration."

Emma grabbed her laptop and sat down back at her desk, beaming with pride.

Professor Smith continued, "Now, let's move on to the next student and see what kind of creative stories we came up with today..."