When AI and information architecture work together, they drive success
Information architecture is important for creating a positive customer experience, but it’s also key for developing strong artificial intelligence (AI) systems. They work hand in hand.
AI systems drive how to structure and present information, while well structured information drives the success of the AI. They must marry and adapt to each other in order for either to work.
I’ll dive into why, but first let’s explain what we mean by information architecture.
What is information architecture?
A grocery store architecture
Let’s think about how a grocery store organizes its food. Each item in a store has a category. Apples belong with fruit; bread lives with the bakery; and eggs are categorized with dairy (at least at my store). Grocery stores organize their items based on a number of factors, including where users expect to find them.
My grocery store has a bulk section where you can buy coffee, dried fruit, grains, and spices in the quantities you want. Spices are organized on the shelf alphabetically.
Recently I went to the spice section to find pumpkin pie spice. I stood in front of the spice shelves looking for a long time. According to what I understood of the system, I should have found the pumpkin pie spice between the Prime Rib Rub and the Red Pepper. But I didn’t see anything.
Eventually I gave up and packed my other spices, but as I browsed in the bulk tea section, I spotted the pumpkin pie spice. It sat on a low shelf, off to the side — easy to miss. Someone likely forgot to include it with the others, so they put it in the closest available space. This low, labeled position was an after-thought.
Categorization of systems
The forgotten pumpkin pie spice is a great example of when an information architecture doesn’t accurately categorize and label each piece of information.
Information architecture is the structuring and labeling of information — any information — in a system. The system can be digital, like a website or app, but it can also be physical or even both. A library — where books are organized on shelves and digitally in a catalog search — is a great example of marrying physical and digital information architecture.
The most important feature of an information architecture is the ability to retrieve information from it. Information must be categorized in a way so users can find what they want.
How AI reshapes information
Personalization of content
One of the most exciting features of artificial intelligence is how it can enhance the user experience through personalized information delivery. We’ve already seen software and apps use algorithms to present personalized content based on user preferences and past interactions.
My Netflix account has learned what I like to watch based on my past viewing history and provides suggestions for me on what to watch next.
My Spotify account has a similar feature. The algorithm has learned from my listening habits. It not only offers listening suggestions but also gives me a personalized “Wrapped” every December where it tells me what I listened to that year and gives me a fun little interactive sequence that I can watch and share with others.
And because preferences and habits can change, the AI can adapt the information to the user over time. My “Wrapped” looks a lot different today than it did when my kids were smaller and weren’t introducing me to all of their music and favorite artists.
Contextual delivery of content
AI also enables contextual content prioritization. Human context is critical for strong persona development. The “me” that hangs out with my kids on a sunny weekend at the pool is a very different context than a Monday morning at the office where I’m facing a day jam-packed with meetings.
AI can deliver content based on the users’ context, like their location, the time of day or their device, so the user receives information at the right time.
AI enables a more powerful search, especially when it incorporates Natural Language Processing. When it does, the search functionalities can expand to understand semantics and user intent so search can return more accurate and relevant results.
All of these things require us to rethink how we shape the information architecture and how technology plays a role in that shift.
How to adapt information architecture for AI
Artificial intelligence is only as good as the source data that it learns from. And while information architecture is shaped by AI’s abilities, AI can only leverage carefully structured information.
When data has not been thoughtfully organized it makes it much more difficult to learn. Here’s an example.
Let’s imagine we have a huge database of books and we want to use AI to deliver content in a dynamic, personalized search. Many of the books are categorized according to multiple labels, like whether they are nonfiction or fiction, and to a specific category of interest, like biography, history, business. And they may even be labeled for the literary audience — adult, teens, children.
But let’s imagine that there are some books in the collection that were quickly entered into the database without any of these tags. They might only have a title. Lacking a complete labeling, these books are much harder to find — even for AI.
AI relies on a complete framework for the organization of data so it can be at its best. Information is organized data. AI must understand how data should be organized in order to be effective. And remember — the ability to retrieve information is the most important characteristic of all.
So information should be thoughtfully and thoroughly labeled so that AI can learn from it. In the case of lots of data, this may require rethinking previous systems and how metadata is structured so that AI can fully leverage it.
This is not a small task, and it may require hiring an information architect or expert agency to accomplish.
When it comes to AI and information architecture, the two are crucially linked. Information architecture is all about categorizing information so it can be retrieved. And thanks to AI, we now can enhance the user experience through personalized content delivered at the right time.
To do this, data must be organized thoroughly and thoughtfully so that AI can better learn from it. Without this step, AI systems will not reach their full potential.