Time
AI Strategy & Human–AI Systems
Project
AI is not simply giving us a new interface. It is changing where intelligence lives — and that changes what designers are responsible for.
I have spent much of my design career working inside complex systems. The deeper I go into systems shaped by algorithms, optimization, automation, and artificial intelligence, the more strongly I return to one simple principle: Design for Human.
Human attention is finite. Memory is fragile. Judgment is contextual. We make decisions with incomplete information, under time pressure, while carrying assumptions we may not even recognize. AI is becoming extraordinarily good at compensating for some of these limitations.
How should intelligence be distributed between the human and the system?
AI is not a new UI problem
We often describe AI as a better assistant, a smarter search box, or a conversational interface. But these descriptions are too small. Like electrification, AI will change the structure surrounding the capability itself. This time, the capability being reorganized is intelligence.
Intelligent systems can now observe, interpret, prepare context, generate options, recommend, challenge, coordinate, and act. We are no longer designing only interfaces between a person and a tool. We are designing an operating model between human and machine intelligence.
The scarce resource is no longer information
Herbert Simon’s work on bounded rationality remains foundational here. Humans do not have unlimited time, attention, memory, or computational ability. AI reverses part of the information problem: we may soon have more generated analysis, recommendations, explanations, and possible actions than a person can meaningfully inspect.
Intelligence is not only the production of answers. It is the structuring of attention under constraint.
A good decision system understands which insights matter now, which differences require human judgment, and what a person needs to know before acting.
Design for Human → Enter the Complexity → Restructure Intelligence
1. Design for Human
Human cognition is not scalable. We can increase compute, storage, and model capability, but not a person’s attention, working memory, or capacity to reason across hundreds of interacting variables. Designing for humans is clarity within cognitive limitation — deciding what must remain visible, what the system can prepare, and what actually needs a person’s judgment.
2. Enter the Complexity
Before simplification, enter the complexity. Ask where it actually comes from: the interface, business model, regulation, system dependency, conflicting objective, historical process, organizational boundary, incomplete data, or a decision that genuinely needs several perspectives.
Complexity is not something to remove. It is something to structure.
3. Restructure Intelligence
Only after understanding the human and the system can we ask what AI should do. Where should intelligence live? What should the system observe, interpret, construct, surface, recommend, execute, and keep explainable, reversible, and accountable?
Human, Agent, System
The human layer is about attention, mental models, uncertainty, trust, cognitive load, and judgment. The agent layer is about intent, context, evidence, alternatives, recommendations, and action. The system layer anchors intelligence in signals, rules, sources, dependencies, business objects, actions, and outcomes.
A traceable chain matters: Signal → Evidence → Interpretation → Recommendation → Human Decision → Action → Outcome. Without it, intelligence risks becoming abstraction. With it, intelligence can remain accountable to consequences.
Designing the conditions for thinking together
The real work is deciding what the human should understand, what the machine should understand, what each should do, and how responsibility moves between them. The next generation of UX will be about more than screens: we will be designing the conditions under which human and machine intelligence think together.