# AI Assistance for UX: A Literature Review Through Human-Centered AI

**Authors:** Yuwen Lu, Yuewen Yang, Qinyi Zhao, Chengzhi Zhang, Toby Jia-Jun Li  
**arXiv:** [2402.06089](https://arxiv.org/abs/2402.06089)  
**Date:** 2024-02-08  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2402.06089v2/

## Key points

- Growing interest in the intersection of Human-Computer Interaction (HCI) and Artificial Intelligence (AI) to enhance user experience (UX) through AI-enabled tools
- Limited widespread adoption of AI-enabled tools despite potential benefits
- Systematic literature review conducted on 359 papers to synthesize current landscape, identify trends, and uncover unmet needs of UX practitioners in AI support
- Analysis revealed potential hindrance of empathy-building process by AI automation in UX design
- Importance of considering interactions and user flows rather than focusing solely on individual UI screens for practical value of AI tools for designers
- Need for understanding UX practitioners' unique perspectives and incorporating more designer-centric datasets and evaluation metrics in HCI and AI research
- Key aspects for further exploration in AI support for UX design include understanding user needs deeply within different phases of design processes
- Importance of holistic approach integrating technical feasibility with human-centered design principles to create effective AI tools supporting UX practitioners

## Summary

In recent years, there has been a growing interest in the intersection of Human-Computer Interaction (HCI) and Artificial Intelligence (AI) to enhance user experience (UX) through AI-enabled tools. Despite the potential benefits of emerging models and interaction mechanisms, the widespread adoption of such tools remains limited. To address this gap, a systematic literature review was conducted on 359 papers to synthesize the current landscape, identify trends, and uncover unmet needs of UX practitioners in AI support. The analysis, guided by the Double Diamond design framework, revealed that while AI automation can offer efficiency, it may hinder the crucial empathy-building process that is central to UX design. Additionally, focusing solely on individual UI screens without considering interactions and user flows may limit the practical value of AI tools for designers. This highlights the importance of understanding UX practitioners' unique perspectives and incorporating more designer-centric datasets and evaluation metrics in HCI and AI research. Through iterative sampling processes involving backward and forward sampling of relevant papers, a comprehensive dataset was compiled representing existing work in the domain. The analysis categorized papers into phases of the Double Diamond framework, with additional categories for UX-related datasets and general AI models. Each paper was evaluated based on research contribution type, target problem/task addressed, consideration of user needs, empirical evidence from previous work,

AI model architecture and data modality used

important model aspects like user control and explainability

as well as UX artifacts involved. The study identified key aspects for further exploration in AI support for UX design including understanding user needs more deeply within different phases of design processes. The findings underscored the need for a holistic approach that integrates technical feasibility with human-centered design principles to create effective AI tools that truly support UX practitioners. This comprehensive review represents a significant repository of literature on AI for UX support compared to past reviews in this field and sets a foundation for future research collaborations between HCI and AI communities towards enhancing UX through innovative technologies.