/ The Problem
Most AI products assume users know what they want.
In reality, people often know how they feel before they know what they need.
AI is excellent at optimizing solutions.
But optimization only works when the goal is clear.
The real challenge is helping people discover and articulate their intent.
/ The Missing Layer
The AI products follow a familiar pattern:
User Request → AI Optimization → Recommendation
...
This model works well when goals are clear.
But many real-world decisions begin with uncertainty.
People often know how they feel before they know what they need.
eh..
em..
?
A parent planning a road trip may know they want the journey to feel easier, but may not immediately recognize that motion sickness, bathroom access, and driving fatigue are the factors that matter most.
Someone managing a long to-do list may know they feel overwhelmed, but may not realize that energy level, anxiety level, motivation, not only urgency, is the variable influencing their decisions.
Today’s to-do:
finish to-do list
narrow it down to the 10 most important tasks
reorder the tasks by urgency
don’t forget the unfinished task from yesterday
take a break
In these moments, optimization is not the problem.
I believe there is a missing layer in many AI experiences:
Human Context → Intent Discovery → Intent Translation → Optimization
Before helping users solve problems, AI should help them define them.
/ Three Principle
Principle 1: AI Should Reduce Effort, Not Agency
Many AI products are designed to make decisions faster.
Food delivery apps recommend meals.
Streaming platforms recommend content.
Shopping platforms recommend purchases.
These systems are often optimized around efficiency.
But efficiency and autonomy are not always the same thing.
Sometimes users do not want a recommendation.
They want exploration. They want surprise. They want the freedom to browse without being directed toward an outcome.
The goal of AI should not be to remove decisions from people.
The goal should be to reduce unnecessary effort while preserving meaningful choice.
Good AI does not replace agency.
It supports it.
Principle 2: AI Should Understand Context, Not Just Data
This principle became particularly clear to me while thinking about navigation systems.
Most routing algorithms optimize measurable variables:
Travel time
Distance
Traffic
Fuel consumption
Toll costs
These are useful metrics. But they are not necessarily the metrics people care about most.
Imagine a family preparing for a six-hour road trip.
The fastest route may save eighteen minutes.
But it may also include long stretches of winding mountain roads.
For a child prone to motion sickness, that route may create a significantly worse experience.
Time
Sensitive
Scenery
optimized
Comfort
needs
The navigation system optimized for Time.
The traveler optimized for View.
The child optimized for Not Getting Sick.
Three different definitions of success.
Only one appears in the data.
This observation led me to explore a concept called Intent-Aware Navigation.
Instead of asking users to choose between predefined route types such as "Fastest" or "Avoid Tolls," the system begins by understanding the goals behind the journey.
Instead of:
"Which route would you like?"
Ask:
"What matters most about this trip?"
The system then helps users discover and prioritize factors they may not have initially considered, translating human needs into routing decisions.
The goal is not simply better route optimization.
It is better understanding of what should be optimized in the first place.
Principle 3: AI Should Adapt to Changing Intent
One of the assumptions built into many digital products is that user goals remain fixed.
In reality, human needs constantly evolve.
A family may begin a road trip hoping to take a scenic route. Two hours later, a tired child changes the situation.The original goal no longer applies.
The most helpful AI system is not the one that follows a plan perfectly. It is the one that recognizes when the plan itself should change.
This idea extends far beyond navigation.
People shift priorities throughout the day. They gain new information. They become tired. They change their minds.
Designing for human intent means acknowledging that intent is dynamic.
AI should not simply optimize for ONE goal. It should help people continuously redefine their goal at the moment as circumstances evolve. And be flexible and easy enough to adapt into the latest context to generate better solutions.
/ Designing the Boundaries of AI
As AI systems become increasingly powerful, the central challenge is no longer capability.
The most important design decisions of the next decade will not be about what AI can do.
They will be about what AI should do.
And perhaps more importantly:
What AI should leave for humans to decide.
I am less interested in designing systems that generate better answers.
I am interested in designing systems that help people ask better questions.
Because the future of AI is not simply optimization.
It is understanding.
And understanding begins long before a solution appears.
/ Reflections
This project changed the way I think about AI.
Originally, I believed the challenge was helping AI understand people. Through this project, I realized sometimes people themselves struggle to articulate what they need.
Perhaps the role of design is not simply to improve AI outputs, and our responsibility is to help people negotiate their own goals before optimization begins.
A system approach to help people navigate, decide, and act in AI-shaped world .
This raises questions beyond navigation.
What if AI products stopped asking users to choose solutions, and instead helped them discover what success actually looks like?
This question extends far beyond travel.
It changes how I think about planning, productivity, commerce, and the future of AI agents.
/ Next Step
A long-term exploration of how design can help people navigate increasingly intelligent systems.