ChatGPT and Me

A First-Hand User Experience

There are only 400 recorded cases of people who have contracted Urbach-Wiethe Syndrome. Symptoms include a hoarse voice, small bumps around the eyes, and calcium deposits on the brain. These are fearless people.

Like most of the remaining 8,085,376,029 billion, I deal with varying degrees of trepidation when it involves developing new skills in unfamiliar territory. A prime example is adopting new technology. I wasn't born with a native sense of how to use software tools. Promises of technology making my life easier have mostly gone unfulfilled.

I'd like to tell you a story about one consumer software application that has fulfilled its promise.

ChatGPT.

In February 2023 I cleared a weeks-long waiting list and visited Open AI's website to look around. But it wasn't until I downloaded the OpenAI app in May 2023 that I started to understand ChatGPT's full potential.

The chatbot's legal name is Chat Generative Pre-Trained Transformer. It was conceived in Toronto and born on November 30, 2022, in San Francisco. The chatbot gave birth to fraternal twins, GPT-4 and GPT-3.5. I have nothing against GPT-4, but GPT-3.5 is OpenAI's self-described "fastest model, great for most everyday tasks" and is free.

Like every new generation, ChatGPT uses different terminology. The user experience is formed on prompts and replies which is different than a search engine's queries and returns.

Think of prompts as a conversation starter and queries as a question and an answer.

Successive prompts, known as prompt engineering, are considered at each stage as a context. In other words, a prompt can be the beginning of a conversation on a subject that could include several prompts leading to a longer conversation with GPT-3.5.

I'll share an actual example from my prompt library:

KIRK: Quantify the most effective manner in which most of us learn.

CHATGPT: The most effective learning often involves a combination of visual, auditory, and hands-on experiences, catering to diverse learning styles and preferences.

KIRK: Apologies, my previous prompt was too broad. I'll break my question down into two prompts. First, provide a list of the diverse learning styles you referred to in your reply. Then quantify the learning effectiveness of the top three styles. Maybe a better prompt is to quantify the effectiveness of different ways to learn based on retained knowledge.

CHATGPT: Ways to acquire knowledge include reading, hands-on experience, lectures, discussions, videos, simulations, and online courses.

Quantifying effectiveness is challenging as it varies among individuals and subjects. Generally, hands-on learning often leads to higher retention, followed by discussions and interactive experiences. Reading and lectures might have lower retention rates but are still valuable for certain types of information. Individual preferences and the nature of the content play crucial roles in determining effectiveness.

The elapsed time of GPT-3.5's replies was less than 40 seconds.

I validated the reply when I reached way back in time to a grade school conversation with my now millennial daughter when she informed me she was a "visual learner." Baby Boomers were raised on a learning diet of the basics. I was confused and concerned. Until I read Temple Grandin's most recent book titled Visual Thinking nearly two decades later.

"The first step to understanding that people think differently is to accept that there are different ways of thinking," said Grandin.

My earlier example of prompt engineering doubles as an explanation for use cases. Based on my prompt history, 50 percent of my prompts are research-related, 30 percent are book summaries and 20 percent are "how to" prompts. I've also had GPT-3.5 write a wedding toast and a newsletter that appeared under my earlier newsletters. I wasn't impressed by GPT-3.5's efforts in either of the last two examples.

A friend recently recommended that I read Zen and the Art of Motorcycle Management. Before I purchased the book, I asked GPT-3.5 to summarize it for me. It's worth sharing this single prompt and reply (which I've edited for length):

KIRK: Provide a 1000-word review of Zen and the Art of Motorcycle Maintenance.

CHATGPT: Certainly! Here's a 1000-word review of "Zen and the Art of Motorcycle Maintenance" by Robert M. Pirsig.

Genre: Philosophy, Autobiography

Published: 1974

"Zen and the Art of Motorcycle Maintenance" is a philosophical novel, first published in 1974. This book is not just about motorcycles; it's a journey into the realms of philosophy, the human mind, and the art of living. Blending autobiography, philosophy, and captivating narrative, Pirsig explores the intricate connections between rationality, the pursuit of quality, and the mystique of Zen Buddhism.

The book is presented as a first-person narrative, where the author embarks on a cross-country motorcycle trip with his son, Chris. This journey becomes a backdrop for Pirsig's philosophical musings and a quest to discover the nature of "Quality" in life.

In conclusion, "Zen and the Art of Motorcycle Management" is a unique and thought-provoking work that defies easy categorization. It seamlessly blends autobiography, philosophy, and travelogue into a narrative that challenges readers to rethink their approach to life and the pursuit of Quality.

The elapsed time of GPT-3.5's 1000-word reply was less than 30 seconds.

On average I prompt GPT-3.5 60-80 times a month. The average elapsed time for replies to 60 prompts is less than 60 minutes. It would have taken me 470 minutes to acquire the same information using traditional methods. GPT-3.5 is saving me 28 hours a month of work-related time.

It was time for a final prompt as it relates to this newsletter.

KIRK: You are saving me 28 hours a month in work-related time. Give me a few suggestions on how to spend it.

CHATGPT: That's great to hear! With your extra time, the possibilities are vast -- it depends on your interests and your goals.