As artificial intelligence moves deeper into everyday products, designers face a growing challenge: creating interfaces that people can trust. Adaptive systems now learn from user behavior, anticipate intent, and sometimes act without being asked. For users, that can feel like convenience. For designers, it introduces a new set of questions about clarity, control, and accountability.
“The hardest part of designing with AI isn’t making it intelligent,” says Osman Gunes Cizmeci, a New York-based UX and UI designer known for his writing on emerging design trends. “It’s making it understandable.”
From Usability to Accountability
For much of its history, user experience design has been about simplicity and ease of use. Designers focused on reducing friction and helping users achieve goals efficiently. But as algorithms begin to make decisions on behalf of people, clarity and accessibility are no longer enough.
Products such as Microsoft Copilot, ChatGPT, and Google’s Gemini now adapt in real time to user data, learning from behavior and adjusting content, layout, and tone. “We’ve entered a phase where users aren’t just interacting with software,” Cizmeci explains. “They’re forming relationships with systems that act on their behalf.”
That shift changes what UX is responsible for. “Designers used to focus on what users could do,” he says. “Now we have to focus on what the system should do, and how it explains itself along the way.”
The Transparency Problem
Adaptive systems can make technology feel responsive and personal, but they can also introduce confusion. When an app rearranges itself or offers a suggestion without context, users can lose confidence in the system.
“When interfaces start adapting invisibly, they stop feeling intelligent and start feeling unpredictable,” Cizmeci says. “Trust breaks the moment a user asks, ‘Why did this happen, and who decided it?’”
He believes that feedback and visibility should now be considered essential design principles. “If an AI system learns something or makes a change, the user should see that,” he adds. “Even a small notification that says, ‘We reorganized this based on your recent activity,’ helps build understanding.”
Cizmeci calls this new direction “shared agency,” a design approach where humans and systems cooperate rather than compete for control. The goal is to design interfaces that show where automation begins and where human choice still matters.
“When I design adaptive features, I map decision points early,” he explains. “I outline where the system should suggest, where it can act automatically, and where the user needs to confirm or override. That structure keeps the experience transparent.”
He sees trust as the central challenge of modern UX. “Automation can be powerful, but if users feel sidelined, they disengage. The goal isn’t to build systems that act faster; it’s to build systems that act respectfully.”
Trust as the New Usability
A growing number of designers agree. The Nielsen Norman Group recently identified trust, transparency, and user control as defining UX themes for 2025. In this new era, the measure of a good interface is not only how easy it is to use but how reliably it explains its own behavior.
“The new question isn’t ‘Can users do this?’” Cizmeci says. “It’s ‘Do they believe the system is acting in their best interest?’”
He argues that this evolution requires tighter collaboration between designers, engineers, and data scientists. “We can’t treat algorithmic systems as black boxes,” he says. “Designers have to understand how the model behaves, what data it uses, and what boundaries it needs.”
Human-Centered, Not Data-Defined
Cizmeci warns against the growing tendency to let data fully dictate design. “Analytics are useful, but they can’t replace human judgment,” he says. “Designers exist to interpret context, not just measure it.”
He uses the term “data-informed, not data-defined” to describe his approach. Metrics help guide decisions, but empathy determines direction. “If a user says they feel confused or manipulated, that matters more than any engagement score,” he explains. “People don’t trust products that feel like they’re optimizing at their expense.”
The Ethics of Adaptation
As companies race to deploy AI across products, Cizmeci believes ethical design will become a competitive advantage. “Every adaptive system makes choices about what to show and what to hide,” he says. “Those choices reflect values, whether we admit it or not.”
He encourages design teams to document these decisions explicitly, the same way they document accessibility or privacy practices. “Transparency is part of accountability,” he says. “If you can’t explain how your product makes decisions, you shouldn’t automate them.”
Bias and fairness are also part of the equation. Adaptive systems trained on narrow datasets can amplify existing inequalities, optimizing for the majority while excluding others. “We need to design for diversity in behavior, not just averages,” he says. “The system should adapt to people, not the other way around.”
The Future of Trust
Cizmeci believes that as AI becomes more integrated into daily life, design’s role will expand from usability to stewardship. “We’re the ones translating complexity into something people can understand,” he says. “That makes trust a design deliverable, not a byproduct.”
He also sees this as a moment of opportunity for designers. “The more autonomous our systems become, the more human our work has to be,” he says. “Trust is no longer a bonus feature. It’s the foundation of the experience.”
