Date: February 10, 2023
Author: Bob Royce
Reading Time: 5 min 29 sec

Early in my career, a housemate told me about his adventures teaching a robot arm to build a tower of blocks using an AI framework for developing systems that exhibit intelligent behavior. Define a block. Define a tower of blocks. Find the block. Pick the block up. Place the block. Place another block, and so on to make a tower. Oh wait, the arm doesn't know about gravity so it tries to start at the top and work downward. Teach the arm about gravity. Oh, that is a glass table top, so learn to start there instead of crashing through to start on the ground. And so on. It's hard work to teach a machine about the world. But if you want general intelligence, you must provide the context for the world the machine will operate within.

This is why chatGPT, however powerful it might seem, is ultimately just a pastiche of words, as Gary Marcus refers to it in his discussion with Ezra Klein — it has no knowledge of the world, no frame of reference for truth. It merely synthesizes what it ingests, algorithmically processed to find the next best word, and behold the magician pulls what appears to be a beautifully formed answer from its hat. Is that really a rabbit? Maybe not. ChatGPT has no concept of truth or morality. Plato would call it the first sophist, or as modern philosopher Harry Frankfurt might say—chatGPT is the ultimate bullshitter. It speaks with zero regard for the factuality of what it is saying.

So we’ve been thinking about how we might fill that gap between helpful moral philosopher and ultimate bullshitter through the practice of information architecture. Truth in the context of language is largely a human construct. Machines must be taught frameworks for shaping their linguistic world toward truth—as that is defined by the human context. The machine will remain a bullshitter, but people have the capacity to shape the way it responds so that answers can be counted on to be true within the context of use.

This is basically what we do at TUG. We build frameworks of language, centered around some particular context, that “mediate our experience with things,” as Dan Klyn noted recently to his class at the University of Michigan School of Information. Today the experiences needing mediation are mostly digital places like websites, apps and intranet sites, but more and more, people will interact with other complex things like cars and their home. Here, too, information architecture has the power to foster resilient experiences where people enjoy doing things.

How? Fundamental to any good experience is a sense of safety and trust. The goal is to create an environment where people are comfortable dwelling, stopping to take it all in here, then moving from place to place freely, never afraid of getting lost, and confident that where they go fits within the whole in the way they expect. A place where the world as they experience it is truthful and sensical in context, and where the lived experience corresponds well with the concepts, labels, and ideas that designers make available in the product or service. Doors to rooms are really doors to the next room, not some other place. Exits are clearly marked. There is a sense that the people running this place are friendly and here to help you.

In the digital world, creating this sense of trust starts with designing navigation systems and using labels and systems of language that make sense to people, and that are applied coherently and consistently throughout the experience. This is challenging for complex places and requires discipline and good governance.

For the contents of a place to really ring true to people, you must go deeper into what is true about the world being mediated, starting with your core brand identity and presence in the world. Your brand identity is your most foundational expression of truth. Why do you exist? What is your promise? Who are you talking to? How do you measure success? Many companies invest heavily in defining and cultivating their brand. They know the power of clarity and of consistently communicating their brand story. It takes an investment to define and structure how a brand is expressed. But how does that brand identity show up in more complex contexts like a website? What kind of place looks and feels and performs like the brand’s promise?

This is why identifying and unpacking what we mean by “truth” for a given place is so important. We don’t trust places that we suspect might not be truthful. So one of the first things we work to do when crafting the design of a website is to understand the truth of that place. Why does this place exist? What can I expect to do here? What are its promises? Who is this place for? What does a good experience look like? How do we talk about success? All of these questions get at the reality of the world behind the site and the assumptions about how that world operates that shape how people understand and engage with digital places.

To accomplish this requires work up front that is often seen as beside the point and subordinate to staying on time and on budget. While it’s true that research and modeling and alignment take time and cost money, the upfront investment leads to more efficient teams, and significantly reduced risk as teams develop a shared language talking about problems and solutions.

This is necessary to do because everyone brings their own unique frame of reference for what words mean and what is implied by different combinations of words. As Jesse James Garrett writes in his forward to Advances in Information Architecture titled Information Architecture: The First 300,000 Years:

“Every fracture in our culture wars is the result of clashing, incompatible information architectures: one individual drawing a very different meaning from the world than another, based on a very different internal understanding of the world. Every individual human consciousness shapes, and is shaped by, its own information architecture."

To avoid clashing with people, then, our apps, platforms, and websites must make sense to a wide variety of people coming for various reasons to work and engage and transact with other people. Our job as information architects is to design places made of information that make sense to people and ring true within the context of use.

Getting to this shared way of talking about the world has benefits right from the start of the journey. In the face of complex problems and complicated digital systems, teams need to gain a shared understanding of the problem domain and a shared way to talk about it and possible solutions. Misunderstanding leads to confusion, leads to delays, disruptions, and doom. If there is one or more departments or groups involved that need to work together, you can assume from the start that they will each have unique ways they talk about the world, unique information architectures, that are different and likely confusing to other groups. Finding those and resolving them so groups can communicate clearly with each other saves untold woe and friction and dollars later.

This is why we start an IA project by looking for a shared way to represent how the core actors and systems relate to each other given the focus of our particular domain. Depending on the situatedness and nature of things in the place we are designing, there may be particular kinds of relationships that matter far more than others. In this way, we help teams develop a common understanding of the problems they are trying to solve, the objectives they are trying to achieve, and outcomes they want to produce. Equally important, we expose tensions and fragile dependencies so you can resolve conflict in the design phase and avoid costly over-runs.

So as we come into 2023 we are thinking more and more about how we can apply information architecture to bring about change in the world for good.

May your 2023 find increasing clarity in the face of complexity through the power of information architecture.

Bob

Related Article:
Setting Artificial Intelligence Projects Up For Success - Bob Royce