Integrating AI Personas in UX Design Workflows: An Exploratory Study
Authors
Kronborg, Kasper Thorsø ; Bisbo, Kasper
Term
4. term
Education
Publication year
2024
Submitted on
2024-05-29
Pages
65
Abstract
This exploratory study investigates whether an AI persona—an AI assistant guided by a qualitative persona, that is, a user profile built from interview feedback—can strengthen UX design workflows. As a case, we used a wireframe of a meal-planning app for university students; a wireframe is an early, low-detail sketch of screens and functions. We created a qualitative persona from feedback from 13 Danish students and set up two GPT-based AI assistants: one informed by this persona and one without it. In a between-subjects user study, four UX professionals each used one of the two assistants and assessed effects on efficiency, creativity, and decision-making. The persona-informed AI provided more user-centered, practical feedback, helped reduce design biases, and kept attention on user needs. Participants also raised concerns about occasionally nonspecific answers and limited transparency about the AI’s knowledge sources. Overall, they were positive about using the informed AI persona and AI tools in their workflows. The study suggests that informed AI can support UX design with fast, user-focused feedback—especially when access to users is limited—and highlights the promise of combining UX and AI.
Dette udforskende studie undersøger, om en AI-persona – en AI-assistent guidet af en kvalitativ persona, dvs. en brugerprofil baseret på interviewfeedback – kan styrke UX-designarbejdet. Som case brugte vi en wireframe af en madplanlægningsapp til universitetsstuderende; en wireframe er en tidlig, enkel skitse af appens sider og funktioner. Vi skabte en kvalitativ persona ud fra feedback fra 13 danske studerende og opsatte to GPT-baserede AI-assistenter: én informeret af personaen og én uden denne information. I et mellemgruppestudie testede fire UX-professionelle én af de to assistenter og vurderede effekten på effektivitet, kreativitet og beslutningstagning. Den persona-informerede AI gav mere brugernær, praksisrelevant feedback, hjalp med at reducere forudindtagethed i designet og holdt fokus på brugernes behov. Deltagerne udtrykte dog bekymringer om, at svarene ikke altid var tilstrækkeligt specifikke, samt om manglende gennemsigtighed i AI’ens vidensgrundlag. Overordnet var holdningen positiv til at bruge en informeret AI-persona og AI-værktøjer i UX-arbejdet. Studiet peger på, at informeret AI kan understøtte UX-design med hurtig, brugerfokuseret feedback – især når direkte brugerinddragelse er begrænset – og tydeliggør potentialet i at kombinere UX og AI.
[This apstract has been rewritten with the help of AI based on the project's original abstract]