Using qualitative research methods and a human-centered design approach, I designed a text-based user interface (TUI) that encourages user trust when interfacing with artificial intelligence.
natural language processing
mobile
passion project
Context
For people with ADHD, keeping track of the little things can be tough: checking the mail, remembering to eat dinner, making doctor appointments. I quickly realized that most reminder apps were just too convoluted, requiring a multitude of steps to log a single task.
There had to be a better way. So I challenged myself to design an interactive solution that would combine my love for artificial intelligence with my passion for design into a lightning-fast text-based interface that could enable users to quickly and reliably log information to be remember later.
Tools
Figma
FigJam
Google Forms
Notion
Team
Me • Product Designer
Antek Olesik • Full-Stack Engineer
Timeline
May 2024 - Sep 2024
Research
I set out to discover answers by conducting a UX Research survey with 17 participants.
With a 20-question survey featuring both multiple choice and short-response questions, I was able to generate 340-responses of both quantitative and qualitative data to inform the direction of my design.
Morphing anonymous data points in to detailed user personas allowed me to better empathize with what were now tangible humans who could guide my problem solving process. To ensure accessibility, I created two distinct personas based on average responses and outlier responses, respectively.
To my surprise, most respondents didn't experience the same level of clunkiness that I had, and were generally able to log reminders with their software of choice within 1-minute. A common concern, however, was system reliability. Accuracy could vary, especially when using AI assistants like Siri. This encouraged to shift my north star from speed to reliability, accuracy, and transparency.
Design
Utilizing clear, digestible indicators of system-status was essential for facilitating reliability among users.
For older (or less tech-literate) users, interfacing with artificial intelligence can feel alien. By emulating UX patterns found in chat interfaces, I hoped to anthropomorphize the experience, creating a familiar digital environment that could build trust.
To provide an extra layer of user control and freedom, I designed tools allowing users to quickly visualize, edit, and revert reminders, making them feel more comfortable with trusting an AI system. I created a multi-property UI component called 'Preview', which efficiently summarizes processed requests, displaying core elements of reminders in a clear, scannable format.
The AI continuously adapts to user habits, slowly building a relationship users could rely on. As patterns are recognized, the software predicts and suggests additional reminders to help ease cognitive load. User may choose to add suggested reminders, or swipe up to dismiss them.