P2: Dynamic Information Architecture: Rethinking IA Principles for the AI Era

In Part 1 of our discussion, we explored the fundamentals of information architecture (IA) and its importance in the digital world. We delved into principles such as the Gestalt theory, mental models, and cognitive load, which helped shape static IA. We also examined research methods like card sorting and tree testing, which have been invaluable tools for understanding and designing static IA. However, as we enter the era of artificial intelligence (AI), it’s essential to recognize that the landscape of IA is changing dramatically. The principles and methods that worked well for static IA now need to be reevaluated and adapted to accommodate the dynamic and adaptive nature of AI-driven IA.

In this second part, we will focus on how AI is transforming IA, enabling dynamic and adaptive structures and labels. We’ll discuss the need to develop new principles for dynamic IA, identify ideal algorithms for adaptability, and create innovative research methods to generate and evaluate AI-enabled IA. By doing so, we can help usher in a new era of IA that is more engaging, efficient, and user-friendly.

Role of IA in the AI Context

As we’ve seen, traditional IA focused on findability and discoverability, with research methods like card sorting and tree testing optimized to enhance these aspects.  With AI’s ability to interpret a variety of user inputs and guide users to their desired content, the challenge of findability has changed fundamentally.

Traditional IA is similar to airport signs: fixed, designed for a broad audience, and providing a one-size-fits-all solution. AI enables us to transition from this static model to a personalized, adaptive approach, akin to Google Maps, which guides users based on their individual needs and preferences. 

IA has also been tuned to find the right words that match the mental models of our users.  This has traditionally been really important.  I realized that to my peril when I needed to use a restroom urgently in a fancy building.  I walked past the bathroom a dozen times without finding it because they used the term “cloakroom.” To me, that meant a place to store luggage. But with AI, we can say “toilet,” “garderobe,” “powder room,” or even “throne room,” and AI will guide us to the same destination for our… private symphony.

Technology has advanced to a point that findability should not be a problem that needs to be solved with static text on a navbar.  Discoverability remains an important aspect of IA, but AI offers new opportunities to tackle it, such as collaborative filtering and personalized recommendations.

I believe that the main goals of information architecture today are to enable discovery and to influence the mental model of our users.  IA can play a critical role in setting correct expectations of the platform, such that the user knows what to expect and what not to expect correctly.

Research Methods for IA Need to Change

The introduction of AI into the world of IA demands a rethinking of our research methods. Traditional methods, such as card sorting and tree testing, were designed for static models and are not optimized for the dynamic and adaptive nature of AI-driven IA.

Malcolm Gladwell’s “tomato sauce story” provides an excellent analogy for understanding the need to change our research methods. In the story, Gladwell describes how food scientist Howard Moskowitz revolutionized the pasta sauce industry by showing that there isn’t a single “perfect” sauce, but rather a variety of sauces that cater to different tastes. Moskowitz’s research led to the development of numerous sauce options, allowing consumers to find their preferred flavors.

This story is relevant to IA because, just like there isn’t a perfect tomato sauce, there isn’t a perfect IA. AI allows us to create multiple, dynamic IAs that cater to individual users’ preferences and mental models. Consequently, we need new research methods that can help us design and evaluate these dynamic, AI-driven IAs.

Some of the new goals for AI-enabled IA research could include:

  • Evaluative research on whether users can find what they want, taking into account the dynamic nature of AI-driven IA.
  • Understanding what users expect from the product and aligning their expectations with what the product delivers, considering the adaptive features of AI-enabled IA.
  • As we move into the AI era, it’s essential to revisit and revise our research methods for IA. By doing so, we can ensure that we are taking full advantage of the dynamic and adaptive capabilities of AI to create IAs that cater to individual users’ preferences and mental models, ultimately enhancing their overall experience.


The age of AI is upon us, and with it comes the potential for dynamic, adaptive information architecture. By embracing new principles, algorithms, and research methods, we can create digital spaces that are more engaging, efficient, and user-friendly. So let’s bid farewell to the static IA of yesteryear and embrace the AI-driven future of dynamic IA. After all, our digital symphony deserves a worthy conductor.

About Vivek Srinivasan

I work with the Program on Liberation Technology at Stanford University. Before this, I worked with the Right to Food Campaign and other rights based campaigns in India. To learn more, click here.

Have thoughts to share?

This site uses Akismet to reduce spam. Learn how your comment data is processed.