Carrie Bickner

Or How Object-Orientation Could Control Chaos

This morning, as I was looking for two editions of Leaves of Grass, at my district public library. I needed to ask the librarian for help. She spoke energetically as she walked me over to the poetry section, verbally guiding herself through the collection of some 250,000 items: “Poetry is in the 800s, and American poetry is 811, then by author.”

The clarity of her process struck me. In just a few words, she invoked an entire system—a structure that organizes not just books, but ideas, authors, and genres. It was elegant and intuitive. But I also felt a twinge of embarrassment. I once knew this. I went to library school just 18 miles from where we were standing. And yet, here I was, needing to be reminded of the most basic organizing principle for locating a book of poetry.

This got me wondering: why did it take me so long to see object orientation as the framework I needed to understand AI? It’s a tool I’ve used before, but when it came to AI—especially generative AI—it seemed counterintuitive. Generative AI feels sprawling, chaotic, and messy, and object orientation is so structured. Yet, as I thought about it more, I realized that’s exactly why it works.

An Object-Oriented Way of Seeing the World

There’s an old in-joke in library circles about a patron who walks up to the reference desk to ask for a photo of Jesus. The librarian—polite, professional—has to suppress a smile as their brain flags the obvious dissonance. Photography, of course, begins in the 19th century, and Jesus lived long before that.

It’s one of those stories that sticks because it illustrates something bigger about how we think. We instinctively organize the world into categories—defining things, assigning attributes, and mapping relationships. It’s an object-oriented way of seeing, and it’s baked into how we make sense of the world.

This same framework can be applied to generative AI. It might seem like an intellectual clash at first, but it works because it mirrors the way we naturally think about complex systems.

Making Sense of Generative AI

Generative AI is a field that cries out for structure. It’s a rapidly evolving ecosystem of models, training datasets, and outputs, and it’s easy to lose track of how all the pieces fit together. Object orientation gives us a way to map this landscape.

• Chatbots can be understood as objects tied to specific large language models.
• LLMs (large language models) can be categorized by their versions and iterations—GPT-3, GPT-4, Claude, and so on.
• Training data serves as a foundational object type, with attributes that influence the behavior of the models they train.

When you start to see generative AI as a network of objects with defined attributes and relationships, the field begins to feel more comprehensible.

A Framework for Interdisciplinary Collaboration

This approach isn’t just useful for AI. If a field can be modeled, it can be shared. Object orientation supports collaboration across disciplines because it provides a shared language for defining, comparing, and connecting complex systems.

• In biology, genes can be treated as objects with traits and interactions, allowing researchers to share insights across teams.
• In literature, a text like Leaves of Grass can be treated as an intellectual unit, with its various editions and revisions mapped as instances of a broader object.
• In art, static images—whether human- or AI-created—can be framed as objects, enabling artists and technologists to collaborate using shared models.

Object orientation provides a structure that bridges these gaps, fostering new insights and innovations.

Why It Matters for AI

Generative AI is both a creative and technical domain, and preserving the relationships between its many components is critical. Object orientation offers a framework for doing this.

1. A human-created static image and an AI-generated image might belong to the same object category, but their origins, contexts, and purposes form distinct attributes.
2. A text like Leaves of Grass exists both as literature and as training material for large language models, with each role carrying its own context.

By embedding metadata into AI-generated materials, we can create a system where every object carries its history and relationships with it. This has the potential to make AI more transparent, accountable, and ethical.

Looking Ahead: Building an Understanding of Object Orientation

This idea of object orientation isn’t limited to AI. Its principles underlie much of how the modern world works, from automakers to auto parts manufacturers, medical record systems, and enterprise solutions like those used in Human Resources. It’s the same concept that powers integrated systems in industries as diverse as healthcare and logistics.

I’m currently developing an online course to explore these ideas further. The course will cover the history of object orientation, how it functions across industries, and how it shapes the systems we interact with daily. We’ll look at examples ranging from enterprise HR systems to automotive supply chains, exploring how object orientation connects the dots in complex networks.

Conclusion: The Utility of Frameworks

Object orientation is not the perfect solution to organizing generative AI, but it is a powerful one. It’s a way to find order in the chaos, to preserve context and relationships, and to create a shared framework for collaboration across disciplines.

Like the Dewey Decimal System, which places poetry in the 800s, object orientation is a tool. It’s not an answer in itself, but it helps us ask better questions—and that’s where understanding begins.