AI-Augmented UX Design Practices: Investigating how AI is applied as a Designerly Tool in Design Practices: An inquiry of UX practitioners' experiences and practices of applying AI in the design process
Translated title
AI-Augmented UX Design Practices: Investigating how AI is applied as a Designerly Tool in Design Practices
Authors
Ager, Andreas Harboe Salskov ; Konecná, Martina
Term
4. term
Publication year
2023
Submitted on
2023-06-01
Pages
105
Abstract
The rollout of large AI models in 2022–2023 has made generative AI widely available for creating content. Because there is no single standard definition of AI, this thesis treats AI as intelligent systems that can pursue complex goals. These developments affect human-computer interaction and the design of information technology. A human-centered approach helps UX practitioners (designers and researchers focused on user needs) create systems that fit people. Prior work points to two directions: designing AI systems and designing with AI. This thesis focuses on the latter, where practical knowledge about applying AI in design is limited. Methodologically, it uses an explanatory, sequential mixed-methods design: a survey (N=64) mapped AI use and informed selection for the qualitative phase. The qualitative study (N=6) included contextual inquiry (workplace observation) and semi-structured interviews to understand practitioners’ experiences. Data were analyzed with inductive thematic analysis. About half of the UX practitioners use AI across the design process. AI serves as a design aid for inspiration, a starting point, and a brainstorming partner, especially in problem framing and concept development. Generative AI is the most common application and helps with routine, tedious tasks, freeing time for higher-level UX work. Outputs from generative AI are mainly used as inspiration rather than adopted directly, due to ethical concerns about trust, bias, and lack of transparency; they typically require validation or editing. Visual generative AI is used less than text-based tools because the output does not account for the overall user experience. Based on perceived challenges, AI is not seen as able to design autonomously in a human-centered approach; its output must be balanced with human creativity, intuition, presence, and empathy. This suggests AI can support, rather than replace, activities in human-centered design (HCD). Some challenges may stem from limited AI literacy or from using tools that do not fit the task. The discussion highlights that AI’s efficiency can make design more agile and iterative and reduce time, budget, and effort, but may also devalue certain UX competencies. Many practitioners also feel that human-centered design without real users is not truly human-centered because AI cannot capture contextual and intangible aspects of real-life situations. Overall, AI can support existing UX practices, methods, and tools despite its advantages and challenges. Further research is needed on AI’s capacity for creativity, innovation, and empathy within a human-centered approach to assess whether these concerns are justified.
Udrulningen af store AI-modeller i 2022–2023 har gjort generativ AI udbredt i arbejdet med at skabe indhold. Da der ikke findes én standarddefinition af AI, forstås AI her som intelligente systemer, der kan nå komplekse mål. Dette påvirker menneske-computer-interaktion og design af informationsteknologi. En menneskecentreret tilgang hjælper UX-praktikere (designere og forskere, der fokuserer på brugerbehov) med at designe systemer, der passer til mennesker. Litteraturen peger på to spor: at designe AI-systemer og at designe med AI. Dette speciale fokuserer på det sidste, hvor der mangler viden om, hvordan AI anvendes i praksis i designarbejde. Metodisk anvendes et forklarende, sekventielt mixed-methods design: et spørgeskema (N=64) kortlagde brugen af AI og informerede udvælgelsen til den kvalitative del. Den kvalitative del (N=6) bestod af kontekstuel undersøgelse (observation i arbejdet) og semistrukturerede interviews for at forstå UX-praktikernes erfaringer. Data blev analyseret ved induktiv tematisk analyse. Omtrent halvdelen af UX-praktikerne bruger AI på tværs af designprocessen. AI fungerer som et designer-værktøj til inspiration, som startpunkt og som sparringspartner, især i problemformulering og konceptudvikling. Generativ AI er mest udbredt og hjælper med trivielle og tidskrævende opgaver, så der frigøres tid til mere vigtige og abstrakte UX-opgaver. Output fra generativ AI bruges primært som inspiration fremfor direkte i designet, blandt andet på grund af etiske bekymringer om tillid, bias og manglende transparens; derfor skal output typisk valideres eller redigeres. Visuel generativ AI bruges mindre end tekstbaseret generativ AI, fordi outputtet ikke tager højde for den samlede brugeroplevelse. Baseret på de oplevede udfordringer vurderes AI ikke at kunne designe autonomt i en menneskecentreret tilgang; AI’s output skal balanceres med menneskelig kreativitet, intuition, tilstedeværelse og empati. Det peger på, at AI kan støtte aktiviteter i menneskecentreret design (HCD) snarere end erstatte dem. Samtidig kan nogle udfordringer bunde i manglende AI-læsefærdigheder eller i brug af AI-værktøjer, der ikke passer til opgaven. Diskussionen fremhæver, at AI’s effektivitet kan gøre designprocesser mere agile og iterative og reducere tid, budget og indsats, men også kan nedtone visse UX-kompetencer. Mange oplever desuden, at menneskecentreret design uden rigtige brugere ikke er reelt menneskecentreret, fordi AI ikke kan indfange kontekst og uhåndgribelige forhold i virkelige situationer. Overordnet kan AI støtte UX-praksisser, metoder og værktøjer trods fordele og udfordringer. Der er behov for mere forskning i AI’s evne til kreativitet, innovation og empati i en menneskecentreret tilgang for at vurdere, om de oplevede udfordringer er berettigede.
[This apstract has been rewritten with the help of AI based on the project's original abstract]
Keywords