By the late 19th century, a convergence of economic trends and technical advances had transformed American society. Driven by new jobs and cultural opportunities, people abandoned their tight knit, rural communities and flocked instead to fast growing urban centers.
The skyline of these cities transformed in turn, as new tall buildings rose to meet the growing demand for office space amidst rising land prices. This created a quandry for architects and urban planners. While skyscrapers enabled the creation of new experiences and spaces, they were inherently utilitarian. How do you protect the human experience in a utilitarian world?
The architect Louis Henry Sullivan observed this collision. In his view, it fell to the designers of these buildings to bestow dignity on this new form of housing by instilling the cold pillars of concrete with the emotion and care that their grandeur was due to their scale and role.
However, he also advocated for constructing them intentinoally, designing them first and foremost around the needs of their occupants. Different parts of the buildings had different roles, different constraints, different impacts on the overall experience of the humans inside. Each component piece dictated the shape of the whole.
”Form follows function.”
Nearly 130 years later, we’re racing towards another economic inevitability, also spurred on by the incessant marching forwards of technology and markets.
AI is already disrupting how we communicate, work, and live, along with the value we place on other humans and the world around us.
The duality of the task before us is to retain the dignity of the human experience while intentionally shaping our digital spaces based on how they serve us, and not the other way around.
We must seek to understand the components of these spaces, how they differ and relate to each other, an the function they perform. As more of our lives are spent in digital interactions, the form of these interactions relies on the func
nstead of "freeing us from digital drudgery," what if AI's beauty and human value emerged precisely from its computational nature? Rather than trying to make AI feel more "human" or less "digital," what authentic forms of beauty could arise from honestly expressing what machine learning actually is and does?
Sullivan's approach to AI would more likely be: "These systems exist because of computational power, data processing, and optimization algorithms. How do we make that reality itself serve human dignity?"
The most surface level of AiUX contains the interfaces people use to interact with the AI explicitly. These also help the user build trust with the AI by being able to observe its logic, which opens up deeper modes of interaction through inferred or contextual inputs.
Direct inputs allow people to deliver instructions and constraints for the AI to follow, the most common of which is the open prompt box. Users can then verify the AI’s understanding through patterns like sample results, dashboards to monitor agent behavior, or through logical reasoning shared in real time.
Since these interactions don’t rely on the AI having a personal or deep knowledge of the user to be effective, this layer of experience is important for onboarding users into new services, or into new paradigms that might require a different level of trust and context. Once the user has developed a predictive relationship to the AI, these patterns remain important to allow users to observe and debug the AI program.
As the AI gains access to information about the user, it becomes more capable of anticipating the person’s needs without requiring them to directly or indirectly prompt the AI service. The AI is able to proactively anticipate the user’s needs and requires less interaction to continuously serve them. Contextual clues can be found through directly integrating into the user’s ecosystem through tools like RAG and referential artifacts, or indirectly through a common memory system of past exchanges. The deeper into the user’s ecosystem the AI can go, the better it can seamlessly serve the user. Earned trust becomes increasingly important.
The closest thing to what we think of as product design and strategy that we will find in this new world. Beyond just being a wrapper for different models, AI products opinionatedly mesh and apply the capabilities of different models to myriad human needs. Each of these services are delivered as a form of software that integrates with each other across the person’s digital footprint, learning from the user’s inferred and stated preferences and contextual clues to create increasingly personalized experiences. Agents and Agentive experiences extend these services further without requiring an equivalent amount of the user’s attention, while MCP capabilities multiply that augmentation across an unlimited number of related connections.