How Does AI Fit Into a Space That Behaves Well?
What Should AI Be Allowed To Do?
Working at Microsoft, I get asked about artificial intelligence almost everywhere I go. Sometimes the questions come from business leaders trying to understand the future of work. Sometimes they come from homeowners wondering what intelligent homes will eventually become. Sometimes they come from architects, designers, or operators thinking about the environments they create and care for every day. What's interesting is that the conversations rarely begin with technology itself. They usually begin with curiosity. Sometimes uncertainty. Occasionally even a little anxiety. People aren't really asking about algorithms or models. They're trying to understand where AI fits into the parts of life they care about most.
A few nights ago, I found myself having one of those conversations in a beautiful showroom filled with carefully designed office environments, hospitality-inspired spaces, luxury kitchens, and gathering areas. The room was filled with people who spend their professional lives imagining what environments might become. Architects were there. Designers were there. Business leaders were there. Everyone brought a slightly different perspective, but the conversation eventually converged on the question that seems to emerge whenever AI enters the discussion.
"What is AI going to do to all of this?"
It's a reasonable question. AI is already changing how we work, communicate, create, analyze information, and make decisions. Every week seems to introduce another capability that would have felt impossible a few years ago. Yet as I sat there listening to the discussion unfold, I realized I wasn't particularly interested in answering the question the way most people expected.
I wasn't especially interested in what AI would do to the environment.
I was much more interested in what the environment should be allowed to do with AI.
That may sound like a subtle distinction, but I've come to believe it changes the entire conversation.
Over the past several years, I've noticed that discussions about AI almost always drift toward capability. What can it automate? What can it predict? What can it optimize? What can it generate? Those questions make sense because the capabilities are impressive. AI can summarize information in seconds. It can recognize patterns hidden inside enormous amounts of data. It can communicate naturally and help coordinate activities that would have required significant human effort only a few years ago.
The challenge is that environments are not collections of capabilities.
They're places where people live, work, gather, learn, heal, celebrate, and build relationships. A hotel exists for guests. A workplace exists for the people doing the work. An assisted-living community exists for residents and caretakers. A retail environment exists for customers. Long before technology enters the conversation, the environment already has a purpose. Technology matters because it helps fulfill that purpose. The purpose doesn't exist to justify the technology.
That's one reason I often find myself returning to a simple hotel scenario.
Imagine arriving at a hotel shortly before midnight after a long day of travel. Your flight was delayed. You missed a connection. You're tired, frustrated, and ready to be done with airports for a while. At that moment, you're not thinking about artificial intelligence. You're trying to figure out where your room is. You're hoping check-in goes smoothly. You want help when you need it and as little friction as possible.
An AI assistant could be incredibly useful in that situation. It could answer questions conversationally. It could guide you through the property. It could recommend a late-night meal. It could help you extend your stay, navigate hotel amenities, or solve a problem without requiring a trip back to the front desk.
All of that sounds wonderful.
But none of it addresses the question that interests me most.
How does the environment know what is appropriate?
The assistant may know your name. It may know your reservation details. It may even remember your preferences from previous visits. But does it understand which services are included in your stay? Does it understand what information should remain private? Does it know when personalization becomes intrusive? Does it know the difference between a first-time guest and someone with privileges that come from a different relationship with the property?
And if it makes a mistake, who is accountable?
That is the moment where the conversation stops being about AI and starts becoming a conversation about the environment itself.
Because guests don't experience language models.
Guests experience hotels.
Residents experience communities.
Visitors experience buildings and spaces.
Whatever technology participates in creating that experience, the environment remains responsible for how it behaves.
One advantage of spending a career around technology is that eventually you begin recognizing patterns. Long before AI, organizations reorganized themselves around enterprise software. Then came the web. Then mobile devices. Then cloud computing. Then connected devices and IoT. Every technology wave introduced genuine value, genuine innovation, and real opportunities. Yet every wave also produced the same temptation. People began treating a capability as though it were a strategy.
AI feels remarkably familiar in that regard.
It's an extraordinary capability. It will almost certainly become embedded within countless environments. Hotels will use it. Buildings will use it. Retailers will use it. Membership organizations, hospitals, schools, and communities will use it. None of that concerns me.
What concerns me is when people begin assuming AI replaces the responsibilities of the environment.
It doesn't.
The environment still needs to understand what is true. It still needs to understand who it is interacting with. It still needs governance over what should be remembered, what should be forgotten, what is appropriate, what is explainable, and who remains accountable for the outcomes people experience.
AI may help fulfill those responsibilities.
It doesn't become those responsibilities.
Perhaps nowhere is that more visible than in the growing conversation around memory.
Modern AI systems make it possible for environments to retain context in ways that would have seemed extraordinary not long ago. In many cases, that continuity can be genuinely beneficial. A returning guest may appreciate not having to repeat an accessibility accommodation. A resident may value remembered preferences. A community member may benefit from continuity across recurring activities and interactions.
Yet every remembered detail creates a new question.
Not whether the environment can remember.
Whether it should.
What information continues serving a human purpose? What information has outlived its usefulness? Who controls that information? How long should it be retained? What obligations accompany its retention?
Those questions aren't really about AI.
They're questions about stewardship.
AI simply makes them impossible to ignore.
The same thing is true for explainability. I suspect we're going to hear the phrase "the AI decided" quite often in the coming years. Unfortunately, that phrase isn't actually an explanation. If an environment denies access, recommends an action, alerts a caretaker, changes conditions, or intervenes in some way, people naturally want to understand why. They don't need source code or mathematical models. They simply need an explanation that makes sense within the context of their experience.
The outcome matters.
But understanding the outcome matters too.
The longer I study intelligent environments, the more convinced I become that explainability, accountability, and trust will become more important as AI becomes more capable, not less.
Which brings me back to the question from that showroom.
"What is AI going to do to all of this?"
I still think it's an interesting question.
I just don't think it's the most important one.
The question that captures my attention now is different.
What should this space do for the people within it, and where can AI responsibly help?
That shift changes everything. When technology becomes the starting point, people are forced to adapt to technological capabilities. When human outcomes become the starting point, technology must earn its place by helping the environment fulfill its purpose.
That's where I believe AI belongs.
Not at the center of the architecture.
Not as a replacement for governance.
Not as a substitute for accountability.
Instead, AI becomes one capability among many, helping environments communicate more naturally, coordinate more effectively, and serve people more thoughtfully.
The environment remains responsible for the experience.
The people remain responsible for the decisions.
And Human Trust remains the outcome we're trying to achieve.
Artificial intelligence is not the architecture. It is one capability operating within it. A space earns Human Trust because it behaves appropriately toward the people it serves. AI may help an environment understand, communicate, coordinate, and assist, but responsibility for the experience still belongs to the environment and the people who choose how it behaves.
