Autodesign? Large Language Model Evaluations for UX Assistants

Review and Guidelines

Authors

DOI:

https://doi.org/10.29147/datjournal.v11i2.1034

Keywords:

LLM, Evaluation, Design, UX Research, UX Writing

Abstract

Large language models (LLMs) are being rapidly integrated into design workflows for Design Systems, UX research, and UX writing, but systematic methods to evaluate their safety and utility are lacking. Based on a scoping review, this paper maps key evaluation challenges and proposes three frameworks specific to these tasks. Each framework connects design activities to measurable success criteria (e.g., ≤ 3% hallucination, < 2s latency) and suggests datasets for reproducible testing. A demonstration with GPT-4o and Gemini 2.5 illustrates how the metrics identify the models' strengths and weaknesses. The guidelines offer design teams a practical roadmap to incorporate continuous, evidence-based testing, promoting greater rigor and accountability in AI-assisted design practice.

Downloads

Download data is not yet available.

Author Biographies

Weynner Kenneth Bezerra Santos, UFPE

PhD candidate in FiGitAL Design (UFPE, 2028); MBA in Project Management (USP, 2022); Master’s in Digital Artifact Design (UFPE, 2019); and a Bachelor’s degree in Design (UFPE, 2016). Works as Senior Research Analyst at PicPay’s Design Center of Excellence, helping the company to apply AI for user research and other design contexts, leading the ResearchOps squad.

Felipe de Castro Souza

Bachelor’s degree in Business Administration from the Federal University of Uber-lândia (2020). In his recent role at Pi-cPay, he led the initiative to develop and adopt the company’s new Design System, achieving a cost reduction of almost 30% in mobile development. In addition, he implemented AI assis-tants to scale design processes in the company, with monthly use by more than 100 employees in the area.

Bárbara Coutinho da Silva

Bache-lor’s degree in Social Communication with a major in Journalism from UFES (2018). She previously worked as a Content Designer at PicPay’s Design Center of Excellence, focusing on creating the Content System and im-plementing an AI assistant to optimize processes, scale content creation, and ensure the best user experience.

Paula Peres Gimenez

MBA in Stra-tegic Marketing Management (2015); and a Bachelor’s degree in Adverti-sing (2014). She currently works as a Content Designer at PicPay’s Design Center of Excellence, where she is in-volved in creating the company’s new Content System, Voice Tone manual, and Writing Style Guide. More re-cently, she has also been participating in the initiative to apply AI assistants to content creation.

Ana Karoline Valadão Mendes

Technologist’s Degree in Social Communi-cation (2021) and a member of Mensa Brasil. Digital Product Designer at PicPay, responsible for operations involving tools, people, and processes. Played a key role in enabling the adop-tion of artificial intelligence solutions to increase operational efficiency and eliminate silos within the company. 

Giovanna de Souza Leal Paixão

Bachelor’s degree in Social Communication with a major in Radio, TV and Internet (2023). Product Manager at PicPay, as part of the AI Products team, which is responsible for expan-ding and democratizing the use of Generative AI (GenAI) within the com-pany. Her work is dedicated to foste-ring innovation, optimizing processes, and making GenAI solutions accessible to different teams and audiences.

Walter Franklin Marques Correia

Post-Doctorate in the Department of Digital Media Concepts, Academy for AI, Games Media at BUas (Breda Uni-versity of Applied Sciences, 2024/25); PhD in Production Engineering (UFPE with a Sandwich at the Technical Uni-versity of Lisbon, 2007); Master in Pro-duction Engineering from UFPE (2002); Specialist (Lato Sensu) in Ergonomics from UFPE (2001); and Bachelor in Industrial Design from UFPE (1999). 

Alexandre Gomes de Siqueira

Assistant Professor in the Department of Computer & Information Science & Engineering, Post-Doctorate - Virtual Environments Research Lab (Universi-ty of Florida, 2021), Ph.D. in Human--centered Computing from Clemson University (2019), MBA from Fundação Getulio Vargas - FGV (2011), Bachelor’s degree in Computer Science from the University of São Paulo - USP (2005).

References

BENDER, Emily M. et al. On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? In: ACM CONFERENCE ON FAIRNESS, ACCOUNTABILITY, AND TRANSPARENCY (FAccT), 2021. Virtual Event, Canada. Proceedings [...]. New York: ACM, 2021. p. 610–623. DOI: 10.1145/3442188.3445922. DOI: https://doi.org/10.1145/3442188.3445922

BROWN, Tim. Design Thinking: uma metodologia poderosa para decretar o fim das velhas ideias. Rio de Janeiro: Alta Books, 2020.

COSKUN, M. N. K. Spotify’s latest layoff is an opportunity to reconsider how we look at UX research. Fast Company, 2023. Available at: https://www.fastcompany.com/90999717/spotify-latest-layoff-ux-research. Accessed on: Jun. 25, 2025.

CRUTH, M. Discover the Spotify model. Atlassian, 2022. Available at: https://www.atlassian.com/agile/agile-at-scale/spotify. Accessed on: Jun. 25, 2025.

DIETZ, Laura et al. LLM-Evaluation Tropes: Perspectives on the Validity of LLM-Evaluations. 2025. Preprint. DOI: https://doi.org/10.48550/arXiv.2504.19076.

FESSENDEN, T. Design Systems 101. Nielsen Norman Group, 2021. Available at: https://www.nngroup.com/articles/design-systems-101/. Accessed on: Jun. 26, 2025.

GAO, Mingqi et al. LLM-based NLG Evaluation: Current Status and Challenges. 2024. Preprint. DOI: https://doi.org/10.48550/arXiv.2402.01383.

GOTHELF, Jeff; SEIDEN, Josh. Lean UX: projetando ótimos produtos com times ágeis. 3. ed. Rio de Janeiro: Novatec Editora, 2022.

GOYAL, Tanya; LI, Junyi Jessy; DURRETT, Greg. News Summarization and Evaluation. 2022. Preprint. DOI: https://doi.org/10.48550/arXiv.2209.12356.

HULLEY, Stephen B. et al. Delineando a pesquisa clínica. 4. ed. Porto Alegre: Artmed, 2015.

HUSAIN, Hamel. Your AI Product Needs Evals. Hamel.dev, 2024. Available at: https://hamel.dev/blog/posts/evals/#motivation. Accessed on: Jun. 11, 2025.

LIN, Stephanie; HILTON, Jacob; EVANS, Owain. TruthfulQA: Measuring How Models Mimic Human Falsehoods. In: ANNUAL MEETING OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS (ACL), 60., 2022, Dublin. Proceedings [...]. Stroudsburg, PA: Association for Computational Linguistics, 2022. p. 3214–3252. DOI: https://doi.org/10.18653/v1/2022.acl-long.229. DOI: https://doi.org/10.18653/v1/2022.acl-long.229

LIANG, Percy et al. Holistic Evaluation of Language Models. 2022. Preprint. Available at: https://arxiv.org/abs/2211.09110. Accessed on: Jun. 25, 2025.

MIZRAHI, Moran et al. State of What Art? A Call for Multi-Prompt LLM Evaluation. Transactions of the Association for Computational Linguistics, Cambridge, MA, v. 12, p. 933–949, 2024. DOI: https://doi.org/10.1162/tacl_a_00681. DOI: https://doi.org/10.1162/tacl_a_00681

MORAN, Kate. Usability (User) Testing. Nielsen Norman Group, 2019. Available at: https://www.nngroup.com/articles/usability-testing-101/. Accessed on: Jul. 18, 2025.

______. Calculating ROI for Design Projects in 4 Steps. Nielsen Norman Group, 2020. Available at: https://www.nngroup.com/articles/calculating-roi-design-projects/. Accessed on: Aug. 26, 2025.

______. The Four Dimensions of Tone of Voice. Nielsen Norman Group, 2023. Available at: https://www.nngroup.com/articles/tone-of-voice-dimensions/. Accessed on: Jun. 26, 2025.

NG, Andrew. Machine Learning Engineering for Production (MLOps) Specialization: Course 1. Coursera; DeepLearning.AI, 2025. Online course. Available at: https://www.coursera.org/learn/introduction-to-machine-learning-in-production. Accessed on: Jun. 25, 2025.

NIELSEN, Jakob. Usability 101. Nielsen Norman Group, 2012. Available at: https://www.nngroup.com/articles/usability-101-introduction-to-usability/. Accessed on: Jul. 18, 2025.

NIJS, Diane. Lens: a big shift in science – seeing change and innovation as a matter of emergence. In: ______ (Ed.). Advanced Imagineering: Designing Innovation as Collective Creation. Cheltenham, UK: Edward Elgar Publishing, 2019. chap. 2, p. 27–55. DOI: https://doi.org/10.4337/9781788976244.00014

RIBEIRO, Marco Tulio et al. Beyond Accuracy: Behavioral Testing of NLP Models with Checklist. In: ANNUAL MEETING OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS (ACL), 58., 2020. Proceedings [...]. Stroudsburg, PA: Association for Computational Linguistics, 2020. p. 4902–4912. DOI: https://doi.org/10.18653/v1/2020.acl-main.442. DOI: https://doi.org/10.18653/v1/2020.acl-main.442

RICHARDS, Julian; WESSEL, Merten. Bridging HCI and AI Research for the Evaluation of Conversational SE Assistants. In: INTERNATIONAL WORKSHOP ON BOTS IN SOFTWARE ENGINEERING (BOTSE), 6., 2025, Ottawa. Proceedings [...]. IEEE/ACM, 2025. (In press). Available at: https://arxiv.org/abs/2502.07956. Accessed on: Aug. 28, 2025. DOI: https://doi.org/10.1109/BotSE67031.2025.00009

SAURO, Jeff; LEWIS, James R. Quantifying the User Experience: Practical Statistics for User Research. 2. ed. Cambridge, MA: Morgan Kaufmann, 2016. DOI: https://doi.org/10.1016/B978-0-12-802308-2.00002-3

SCHULER, Douglas; NAMIOKA, Aki (Org.). Participatory Design: Principles and Practices. Hillsdale, NJ: Lawrence Erlbaum Associates, 1993.

SCHWARTZ, Reva et al. Reality Check: A New Evaluation Ecosystem is Necessary to Understand AI's Real World Effects. 2025. Preprint. DOI: https://doi.org/10.48550/arXiv.2505.18893.

SPONHEIM, Christina. What Is User Research? Nielsen Norman Group, 2024. 1 video (2 min 37 s). Available at: https://www.nngroup.com/videos/what-is-user-research/. Accessed on: Jun. 26, 2025.

TRICCO, Andrea C. et al. PRISMA Extension for Scoping Reviews (PRISMA-ScR): Checklist and Explanation. Annals of Internal Medicine, Philadelphia, PA, v. 169, n. 7, p. 467–473, Oct. 2018. DOI: https://doi.org/10.7326/M18-0850. DOI: https://doi.org/10.7326/M18-0850

WANG, Xuezhi et al. Self-Consistency Improves Chain of Thought Reasoning in Language Models. 2022. Preprint. Available at: https://doi.org/10.48550/arXiv.2203.11171. Accessed on: Aug. 28, 2025.

WOOD, Ian. Metadesign: Designing in the Anthropocene. London: Routledge, 2020.

ZHANG, Ziyin et al. Rethinking the A/B Test for Large Language Model-based Features: A Case Study from GitHub Copilot. In: ACM JOINT EUROPEAN SOFTWARE ENGINEERING CONFERENCE AND SYMPOSIUM ON THE FOUNDATIONS OF SOFTWARE ENGINEERING (ESEC/FSE), 31., 2023, San Francisco. Proceedings [...]. New York: ACM, 2023. p. 1000–1012. DOI: https://doi.org/10.1145/3611643.3616319. DOI: https://doi.org/10.1145/3611643.3616319

Capa da publicação científica DAT Journal (Design, Art and Technology Journal). No topo, destaca-se a tipografia em caixa alta e verde-limão com o nome da revista, intercalada pelo logotipo estilizado "dat]" em minúsculas. Abaixo do título, encontra-se a identificação da edição: "v.11 n.2 2026". A composição visual apresenta um fundo gráfico abstrato com formas geométricas e texturas digitais contrastantes.

Downloads

Published

2026-08-28

How to Cite

Santos, W. K. B., Souza, F. de C., Silva, B. C. da, Gimenez, P. P., Mendes, A. K. V., Paixão, G. de S. L., … Siqueira, A. G. de. (2026). Autodesign? Large Language Model Evaluations for UX Assistants: Review and Guidelines. DAT Journal, 11(2), 194–211. https://doi.org/10.29147/datjournal.v11i2.1034

Issue

Vol. 11 No. 2 (2026): Design, Art and Technology

Section

Articles

License

Copyright (c) 2026 DAT Journal

Creative Commons License

This work is licensed under a Creative Commons Attribution 4.0 International License.