STEAM Spark

Generative AI: My Ongoing Journey

By C. Clark

Although I’m writing about my AI journey in 2025, it actually began in 2023. ChatGPT launched on November 30, 2022, but I didn’t dive in right away. I usually like to wait a bit when something new hits the scene. But by August 2023, after hearing about it all year, curiosity got the better of me—I created a ChatGPT account just to see what everyone was talking about.

I didn’t take a course or read up on it first. I just started using it—hands-on. My first impression? It felt like a more advanced version of Google Search. But as I began learning more, I realized it was so much more than that.

Exploring AI Through Learning

To go deeper, I took several LinkedIn Learning courses:

• What is Generative AI?

• OpenAI ChatGPT: Creating Custom GPTs

• Integrating Generative AI into the Creative Process

It was that last course that shifted how I viewed AI. A coworker casually mentioned they used AI to help with the repetitive parts of their creative work, and that stuck with me.

At the time, I was editing my novel and started thinking about how to incorporate visuals. I briefly considered turning it into a graphic novel—but I don’t draw, and realistically, it would have taken too long. And knowing myself, I probably would’ve lost interest halfway through.

That’s when I remembered DALL·E—a tool I had signed up for earlier just to explore. To my surprise, I discovered DALL·E and ChatGPT had since been integrated. I logged in, started experimenting, and created an image of a young African American woman sitting beside a fireplace. It was perfect. That image sparked something in me. Suddenly, editing my novel became exciting again—especially as I began envisioning it as a weekly streaming-style visual story.

Some of those early images are now featured on my website.

Creating My Own AI Chatbot

While taking the AI courses, I was introduced to vector databases—and that caught my attention. I’ve always enjoyed working with vectors, especially when I studied Linear Algebra, so I followed that curiosity and took another course:

• LLM Foundations: Vector Databases for Caching and Retrieval-Augmented Generation

That was the last course I took on LinkedIn Learning—for now. From there, I shifted into more hands-on, independent learning, setting a goal: to create my own custom AI chatbot.

I found a great article on Medium.com by Woyera. It was the kind of guide I love—clear and to the point. It outlined the 9 core components needed to build your own AI chatbot. I was moving through the list smoothly, but when I got to #5: Vector Database, I paused and decided to explore that part more deeply.

Where I Am Now

Here’s my current understanding (which I’ll continue to update as I learn more):

1. Start with unstructured data (like text, documents, notes)

2. Use a Transformer Model to turn that data into a vector embedding—a numerical representation of the original data

3. Store those vectors in a Vector Database

4. Use a similarity search to retrieve relevant information when a user inputs a prompt

That’s where I am right now. I’ve focused my learning on how the vector database fits into the larger chatbot system. I even sketched out a diagram to help visualize the whole architecture.

It’s been a little while since I looked at that diagram, but I’ll revisit it and go into more technical detail in my next article—coming this Saturday.

STEAM Takeaway:

 It’s okay to start slow, explore out of curiosity, and pause when you hit a bump. That’s how innovation begins—especially when art and science collide.