How Growing Up with Librarian Parents Helped Me Learn the Art of Prompt Engineering
If there’s one thing I’m deeply passionate about, it’s finding information—especially the kind that’s hard to find.
In today’s world, where we’re flooded with endless content from every corner of the internet, the biggest challenge we face isn’t access—it’s quality. The ability to sift through the noise, follow digital breadcrumbs, and uncover valuable insights has become a rare art, and for me, it’s more than just a skill—it’s part of who I am.
Born Into the World of Information
People often ask me, “How did you get so good at searching and finding things online?” My answer is simple: I was born into it. Both of my parents were academic librarians.
Growing up, I was surrounded by books—not just as tools for learning but as objects woven into the fabric of my childhood. I built make-believe castles out of books when I wasn’t reading them. Information wasn’t just something I consumed; it was something I played with, navigated, and explored.
I still remember visiting my parents at work—academic libraries filled with towering shelves and endless rows of books. As a kid, I couldn’t just wander freely. Academic libraries aren’t exactly playgrounds. Most kids aren’t casually flipping through books in university archives unless it’s on a field trip. But every now and then, I’d plead with my parents:
“Please, can I just look at a book?”
When they finally gave me the green light—usually right before closing time—I knew I had less than a minute to find something. No knowledge of the Dewey Decimal System. No understanding of library classification codes. Just instinct. I’d dash between shelves, my eyes scanning titles, hoping to find something like Winnie the Pooh (spoiler: I never did). But I stumbled across other treasures—books about Charlie Brown's Charles Schulz or the Muppets' Jim Henson, autobiographies, sports books about Ricky Henderson, explanations of how baseball card price guides were written, and many other stories I never would have discovered otherwise.
Those frantic 60-second sprints taught me how to feel my way through information and to trust my intuition even when I didn’t know exactly what I was looking for. That was the beginning of how I became skilled at finding information with limited knowledge or resources.
Similarly, that’s how I approach UC Berkeley’s collection of government documents. With thousands of records—and more added all the time—the challenge is covering as much ground as possible with limited time. That’s why I sue simple tools like ChatGPT to navigate and extract insights efficiently. I dive deeper into that process here in an article about using AI to search large documents or publications quickly.
From Libraries to Large Language Models
Fast forward to today, and that same curiosity drives my work in prompt engineering. If large language models (LLMs) are like vast digital libraries, then ChatGPT is the tool that helps us search them. But here’s the thing:
A large language model is just a repository. It holds information from countless sources, like an academic library filled with data instead of books.
ChatGPT is the interface. It’s how we ask questions, give instructions, and pull relevant information from that vast repository of information.
This is where prompt engineering comes in. To get the best results, you need to know what questions to ask and how to phrase them correctly in ChatGPT to extract quality information from its large language model—just like I had to learn how to find a book on Jim Henson without knowing the library's system. This is called prompt engineering. However, a good prompt isn't just a question; it's a carefully crafted set of instructions designed to retrieve meaningful, high-quality information.
The Joy of the Search
What I love most about this work isn’t just finding the answer. It’s the process—the thrill of chasing down leads, refining queries, and uncovering hidden gems. In an age where most people settle for the first search result, I find joy in going deeper, following digital bunny trails that others might overlook.
But there’s a second dynamic that makes prompt engineering even more exciting for me. Unlike traditional searching, where you’re directly interacting with a search engine or database, prompt engineering is about designing instructions that guide a machine to do the searching for you. It’s not just about finding information; it’s about training the tool to think the way you would—or even better.
And it doesn’t stop there. These prompts aren’t just for me. They’re designed to be used by other people, which adds an entirely new layer of complexity. Therefore, a good prompt has to be:
Useful: It needs to pull accurate, relevant information every time. Accessible: People with different levels of expertise should be able to use it effectively. Nuanced: It must account for variations in language, context, and user intent. This means I’m not just solving a puzzle for myself—I’m creating prompts that others can rely on, tools that can adapt to different needs, cultures, and situations.
Prompt Engineering is a blend of curiosity, strategy, and empathy, all rolled into one. And that’s what keeps me hooked—the challenge of not just finding answers, but building systems that help others find them too.
If you’d like help setting up ChatGPT for research, improving your prompts, or answering any questions, feel free to reach out!