Pith. sign in

REVIEW 3 major objections 4 minor 51 references

"If we misunderstand the client, we misspend 100 hours": Exploring conversational AI and response types for information elicitation

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read This paper claims that adding conversational AI and choice-based responses to a pre-meeting elicitation tool makes clients' written input clearer while lowering their ratings of dependability and, with AI, efficiency.

desk verdict A well-structured study with a credible UX trade-off, but the 'clearer client input' result is likely a format artifact of a rubric that scores grammatical completeness. read the letter →

arxiv 2506.11610 v1 pith:DUO526IA submitted 2025-06-13 cs.HC

Daniel Hove Paludan , Julie Fredsg{aa}rd , Kasper Patrick Bährentz , Ilhan Aslan This is my paper
classification cs.HC
keywords conversationalAIrequirementselicitationclient-designeralignmentuserexperiencechoice-basedresponsesGriceanmaximsdesignbriefshuman-AIcollaboration
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Design projects go wrong when clients and designers start from mismatched mental models, and design briefs often fail to surface clear, usable requirements. This paper reports a three-phase study of a digital pre-meeting elicitation tool, testing four versions in a 2x2 factorial design: conversational AI versus no AI, and free-text versus choice-based responses. The central finding is a trade-off: both AI and choice-based formats made clients' written input measurably clearer, while making the experience feel less dependable, and AI also made it feel less efficient. Self-reported preparedness and the remaining UX dimensions were unaffected. The authors argue that embedding the AI assistant in a form-like interface, rather than a pure chatbot, can keep the clarity benefit while giving clients back a sense of control.

What carries the argument

The central object is a four-version elicitation tool defined by a 2x2 factorial design: conversational AI presence (AI vs no AI) crossed with response type (free-text vs choice-based). Response quality is measured through Gricean maxim proxies—manual 0-2 ratings of specificity, relevance, and clarity, plus response length—and user experience through the UEQ. The argument is carried by two-way ANOVA comparisons across these conditions: the only significant effects are lower dependability for both factors, lower efficiency for AI, and higher clarity and longer responses for the AI/choice combinations, which is what grounds the trade-off claim.

What would settle it

Have two independent raters score the same 50 participants' responses on the paper's 0-2 clarity scale and check inter-rater agreement; if agreement is low, or if the AI/choice clarity advantage disappears when real designers judge which responses would produce a usable brief, the central claim fails.

Watch

Extended reading notes

Core claim

The paper's central claim is that integrating conversational AI and choice-based response formats into a pre-meeting information elicitation tool improves the clarity of client input at the cost of perceived dependability. In a 2x2 evaluation with 50 mock clients, both factors produced significantly lower UEQ dependability scores, and AI presence also lowered efficiency; the clarity gains were significant for AI over no AI and for choice-based over free-text, with the largest gain for AI-generated smart options in the choice-based condition. Preparedness ratings did not differ across conditions. The authors conclude that elicitation tools should give clients control, always allow free text, and present layered, editable AI outputs, rather than replacing the form with a chat.

Load-bearing premise

The positive clarity result rests on one researcher's manual ratings of response clarity, relevance, and specificity (Section 4.4.2), with no second rater checking agreement; if those ratings are idiosyncratic, or grammatical clarity is not what designers actually need, the paper's central upside is unsupported.

Editorial extensions

If this is right

  • Designers can expect clearer client input when using either conversational AI or choice-based formats, and the clearest input comes from combining them with AI-generated smart options.
  • The dependability and efficiency penalties mean these formats should not be added blindly; the tool must preserve free-text options, allow revision, and keep navigation flexible.
  • Because preparedness ratings were equal across conditions, clients will not perceive the clarity benefit themselves, so the value must be communicated or absorbed by the designer.
  • Layered outputs—raw, summarised, interpreted, and actionable—should be part of such tools, with AI interpretations made editable and verifiable by clients.
  • Follow-up question strategies need to be based on semantic content rather than a fixed character threshold to avoid repetitive, mechanical conversations.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Beyond the paper: if the trade-off generalises to real projects, the client's UX score may be the wrong success metric; the designer, not the client, is the one who reaps the clarity benefit, so pilot deployments should measure whether design briefs improve.
  • Beyond the paper: the clarity scale's definition (grammatically complete sentences) may reward AI-polished phrasing rather than information useful for a design brief; a testable follow-up is to have designers rate anonymised responses for brief-usability.
  • Beyond the paper: the reported trade-off is likely tied to this prompt configuration, including the 60-character follow-up threshold and rigid question order; varying those parameters could move the efficiency cost or the clarity gain, so the specific magnitudes should not be treated as universal.
  • Beyond the paper: re-running the 2x2 with real clients on live projects, and with two independent raters for clarity, would be the direct way to test whether the mock-client result survives contact with actual design practice.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. This paper reports a three-phase study on a digital elicitation tool for early-stage client-designer collaboration. Phase 1 consisted of semi-structured interviews with 11 designers from 10 companies, informing the design of a prototype. Phase 2 evaluated four variants of the tool in a 2x2 factorial design (AI vs. no AI, free-text vs. choice-based responses) with 50 mock clients, measuring user experience via the UEQ, perceived preparedness, and response quality via Gricean-maxim-based manual ratings. Phase 3 gathered feedback from seven of the original design companies. The paper reports that conversational AI and choice-based responses lower dependability scores on the UEQ (and AI lowers efficiency), while claiming that these formats produce client input with greater clarity. It concludes with three design implications for future elicitation tools.

Significance. If the clarity finding were fully supported, the paper would make a useful, non-obvious contribution: a measurable trade-off between perceived dependability and the interpretability of client input in pre-meeting elicitation tools. The study has notable strengths: a grounded design process tied to practitioner interviews, an appropriate 2x2 between-subjects design with inferential statistics and post-hoc adjustments, candid acknowledgment of core limitations (hypothetical clients, no designer-client dyads), and design implications that are clearly traceable to the qualitative data. The 60-character follow-up threshold is explicitly acknowledged as arbitrary in Section 6.3 and is not a fitted constant, so I see no circularity problem. However, the paper's headline positive claim depends on a clarity measure whose construct validity is questionable, and the reported statistical pattern for clarity does not match the abstract's main-effect wording. These issues are load-bearing and require revision.

major comments (3)
  1. [§4.4.2 and §4.1.2] The clarity rubric is operationalized as grammatical well-formedness: a response scores 2 if it consists of 'complete sentences with no significant grammatical issues.' In the choice-based conditions, however, the logged response is by construction the text of a system-supplied option: predefined in the no-AI condition and generated dynamically by the LLM in the AI condition. Choice-based responses will therefore mechanically outperform typed free text on a grammatical-completeness scale, making the reported clarity advantage at least partly a format artifact. This is load-bearing because the paper's central trade-off—accepting lower dependability in exchange for 'client input with greater clarity'—rests on this measure. The authors should either re-score responses with a content-focused rubric that does not conflate grammatical form with the quality of elicited requirements, or substantially soften the claim.
  2. [§4.6] The analysis of clarity is described as revealing a significant interaction effect between AI presence and response type, followed by pairwise comparisons. The text does not report the main-effect F statistics for clarity, and the reported pairwise results show a choice-based advantage only 'when AI was present.' The abstract and introduction state that 'both conversational AI and choice-based responses' result in greater clarity, which is not supported by an interaction-only result. The full ANOVA results for the Gricean metrics should be reported, and the claims in the abstract, introduction, and conclusion should be qualified to match the actual pattern of simple effects.
  3. [§4.4.2] All specificity, relevance, and clarity ratings were made by a single researcher, blind to experimental condition, but no inter-rater reliability is reported. Because the scales require subjective judgment—especially 'relevance' and the grammatical-completeness interpretation of 'clarity'—a second rater and an agreement statistic (e.g., Cohen's kappa or ICC) are necessary to establish that the effects are not idiosyncratic. If additional rating is not feasible, the absence of inter-rater reliability should be explicitly discussed as a limitation affecting the confidence in the clarity result.
minor comments (4)
  1. [§4.4.2] The footnote 'Scoring examples for each metric are provided in Appendix??' contains an unresolved cross-reference; the appendix should be included or the reference removed.
  2. [§4.1.3] The text 'powered by gpt-4o-minithrough the OpenAI API' appears to contain a typo; it should read 'gpt-4o-mini through the OpenAI API.'
  3. [§4.4.2] The phrase 'creditable corpora of Danish words' should be 'credible corpora of Danish words.'
  4. [Abstract and §1] The wording 'both conversational AI and choice-based responses lead to lower dependability scores ... yet result in client input with greater clarity' should be revised to reflect that the clarity result is an interaction effect and that the choice-based advantage was observed only in the AI-present condition.

Circularity Check

1 steps flagged · score 6.0 of 10

The clarity benefit in the headline trade-off is partially circular: the rubric defines clarity as grammatical completeness, and choice-based/AI formats supply prewritten complete sentences by construction.

  1. other [Section 4.4.2 (clarity rubric) with Section 4.1.2 (response-format manipulation) and Section 4.6 (clarity ANOVA results)]
    "Findings show that both conversational AI and choice-based responses lead to lower dependability scores on the User Experience Questionnaire, yet result in client input with greater clarity. ... To quantify their clarity (i.e., how easy they are to understand) we employed a scale from 0 (illegible text) to 2 (complete sentences with no significant grammatical issues). ... For the choice-based (no AI) condition, these options are all predefined. For the choice-based (AI) condition, they are smart options, generated dynamically based on users' previous responses."

    Choice-based answers are not typed by the client but are selected from a list of prewritten response options (static in the no-AI condition, LLM-generated in the AI condition). The rubric used to establish 'greater clarity' awards the top score to 'complete sentences with no significant grammatical issues.' A selected prewritten option is, by construction, a complete grammatical sentence, so choice-based responses will mechanically outscore spontaneous free-text on this rubric; AI-generated options also tend to be polished sentences, explaining the AI effect. Thus the paper's headline upside—that AI and choice formats yield 'client input with greater clarity'—is an artifact of aligning the outcome's operational definition with the input format's intrinsic properties.

full rationale

Most of this paper is an empirical 2x2 evaluation with no fitted constants, no predictive model, and no derivation that could reduce to its inputs. The self-citations (e.g., [35]) are related-work asides and carry none of the load for the reported effects. The UX findings—lower dependability/efficiency with AI and choice-based formats—are independent of the measurement issue and are not circular. However, the positive half of the headline claim, 'client input with greater clarity,' is partially circular: clarity is operationalized as grammatical completeness, and choice-based/AI response formats supply exactly that property by construction (predefined or LLM-generated option sentences). This makes the main upside a measurement consequence rather than independent evidence. Since the downside findings remain intact and only half of the central trade-off is affected, the appropriate score is 6 (partial circularity). The phase-3 limitation that designers did not test the tool with a client is a validity caveat, not a circularity issue.

Assumptions & free parameters 1 free parameters · 4 assumptions · 0 invented entities

The paper's empirical conclusions rely on standard statistical assumptions and on domain assumptions about measurement validity. There are no fitted mathematical constants in the central claim; the only hand-chosen parameter is the 60-character follow-up threshold, which conditions the AI's behaviour.

free parameters (1)
  • Follow-up question length threshold = 60 characters
    The system only asked a follow-up question when a free-text response was shorter than 60 characters. This threshold was chosen to balance depth and completion time, and it shapes how much AI probing occurred, so the clarity and efficiency results are conditional on it.
assumptions (4)
  • domain assumption The adapted Gricean proxy scales (specificity, relevance, clarity, response length) are valid and meaningful measures of the quality of client input for design briefs.
    The paper scores responses on 0-2 scales adapted from Xiao et al. and Jacobsen et al., but does not validate these proxies against designer interpretations or actual brief quality.
  • domain assumption Single-rater, condition-blind scoring produces reliable response-quality measurements.
    No inter-rater reliability statistic is reported, so the clarity, relevance, and specificity results assume the single rater's judgments are reproducible.
  • domain assumption Behaviour of mock clients answering about hypothetical projects is representative of real client behaviour.
    The paper acknowledges participants were not real clients with real projects, so this is an explicit assumption about external validity.
  • standard math ANOVA remains valid despite Shapiro-Wilk violations for attractiveness and perspicuity.
    The authors cite Blanca et al. for ANOVA robustness; this is a standard statistical assumption and is reasonable given approximately equal group sizes and homogeneous variances.

how reviews work

0 comments
Cite this review

Pith. "Pith review of "If we misunderstand the client, we misspend 100 hours": Exploring conversational AI and response types for information elicitation." pith.science (2026). https://pith.science/paper/DUO526IA

@misc{pith2026250611610,
  author       = {Pith},
  title        = {Pith review of: "If we misunderstand the client, we misspend 100 hours": Exploring conversational AI and response types for information elicitation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DUO526IA}},
  note         = {Machine review of arXiv:2506.11610}
}
read the original abstract

Client-designer alignment is crucial to the success of design projects, yet little research has explored how digital technologies might influence this alignment. To address this gap, this paper presents a three-phase study investigating how digital systems can support requirements elicitation in professional design practice. Specifically, it examines how integrating a conversational agent and choice-based response formats into a digital elicitation tool affects early-stage client-designer collaboration. The first phase of the study inquired into the current practices of 10 design companies through semi-structured interviews, informing the system's design. The second phase evaluated the system using a 2x2 factorial design with 50 mock clients, quantifying the effects of conversational AI and response type on user experience and perceived preparedness. In phase three, the system was presented to seven of the original 10 companies to gather reflections on its value, limitations, and potential integration into practice. Findings show that both conversational AI and choice-based responses lead to lower dependability scores on the User Experience Questionnaire, yet result in client input with greater clarity. We contribute design implications for integrating conversational AI and choice-based responses into elicitation tools to support mutual understanding in early-stage client-designer collaboration.

Figures

Figures reproduced from arXiv: 2506.11610 by the authors.

Figure 1
Figure 1. An illustration of the four experimental conditions of the developed elicitation tool to support mutual understanding in [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The three phases of the conducted study. [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. The study setup of the phase one designer interviews. One researcher facilitated the interview while another took notes. 18 [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Three of the 18 inspiration cards created for exploring how AI might be integrated into designers’ current practices of client [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Illustrations of the four experimental conditions in the 2x2 factorial design. [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]
Figure 6
Figure 6. Figure 6: Screenshots of the system asking the same question in each of the four experimental conditions. [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 7
Figure 7. Figure 7: Plot of the UEQ scales for all conditions. Y-axis shows the mean with standard error bars. [PITH_FULL_IMAGE:figures/full_fig_p015_7.png]
Figure 8
Figure 8. Figure 8: Illustration of how insights—from the phase one designer interviews, the phase two client evaluations, and the phase three [PITH_FULL_IMAGE:figures/full_fig_p023_8.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

51 extracted references · 36 canonical work pages

  1. [1]

    Cansu Akverdi and Gökçe Elif Baykal. 2024. Generative AI Tools in Design Fields: Opportunities and Challenges in the Ideation Process. InAdjunct Proceedings of the 2024 Nordic Conference on Human-Computer Interaction. 1–5

    work page 2024

  2. [2]

    Gabriela Goldschmidt and. 2007. To see eye to eye: the role of visual representations in building shared mental models in design teams.CoDesign3, 1 (2007), 43–50. doi:10.1080/15710880601170826 arXiv:https://doi.org/10.1080/15710880601170826

    work page doi:10.1080/15710880601170826 2007

  3. [3]

    Muhammad Ashfaq, Jiang Yun, Shubin Yu, and Sandra Maria Correia Loureiro. 2020. I, Chatbot: Modeling the determinants of users’ satisfaction and continuance intention of AI-powered service agents.Telematics and Informatics54 (2020), 101473. doi:10.1016/j.tele.2020.101473

    arXiv 2020

  4. [4]

    Ilhan Aslan, Tabea Schmidt, Jens Woehrle, Lukas Vogel, and Elisabeth André. 2018. Pen + Mid-Air Gestures: Eliciting Contextual Gestures. In Proceedings of the 20th ACM International Conference on Multimodal Interaction(Boulder, CO, USA)(ICMI ’18). Association for Computing Machinery, New York, NY, USA, 135–144. doi:10.1145/3242969.3242979

    arXiv 2018

  5. [5]

    Mohammadmehdi Ataei, Hyunmin Cheong, Daniele Grandi, Ye Wang, Nigel Morris, and Alexander Tessier. 2024. Elicitron: An LLM Agent-Based Simulation Framework for Design Requirements Elicitation.ArXivabs/2404.16045 (2024). doi:10.48550/arXiv.2404.16045

    arXiv doi:10.48550/arxiv.2404.16045 2024

  6. [6]

    Björn Bittner, Ilhan Aslan, Chi Tai Dang, and Elisabeth André. 2019. Of Smarthomes, IoT Plants, and Implicit Interaction Design. InProceedings of the Thirteenth International Conference on Tangible, Embedded, and Embodied Interaction(Tempe, Arizona, USA)(TEI ’19). Association for Computing Machinery, New York, NY, USA, 145–154. doi:10.1145/3294109.3295618

    arXiv 2019

  7. [7]

    M José Blanca Mena, Rafael Alarcón Postigo, Jaume Arnau Gras, Roser Bono Cabré, and Rebecca Bendayan. 2017. Non-normal data: Is ANOVA still a valid option?Psicothema, 2017, vol. 29, num. 4, p. 552-557(2017)

    work page 2017

  8. [8]

    Virginia Braun and Victoria Clarke. 2006. Using thematic analysis in psychology.Qualitative research in psychology3, 2 (2006), 77–101

    2006

Show all 51 references
  1. [9]

    Hernan Casakin and Petra Badke-Schaub. 2017. Sharedness of team mental models in the course of design-related interaction between architects and clients.Design Science3 (2017), e14

    2017

  2. [10]

    Li-Yuan Chiou, Peng-Kai Hung, Rung-Huei Liang, and Chun-Teng Wang. 2023. Designing with AI: An Exploration of Co-Ideation with Image Generators. InProceedings of the 2023 ACM Designing Interactive Systems Conference(Pittsburgh, PA, USA)(DIS ’23). Association for Computing Mach...

    2023

  3. [11]

    Hatim Dafaalla, Mohammed Abaker, Abdelzahir Abdelmaboud, Mohammed Alghobiri, Ahmed Abdelmotlab, Nazir Ahmad, Hala Eldaw, and Aiman Hasabelrsoul. 2022. Deep Learning Model for Selecting Suitable Requirements Elicitation Techniques.Applied Sciences(2022). doi:10.3390/ app12189060

    2022

  4. [12]

    Alan Dix and Layda Gongora. 2011. Externalisation and design. InProcedings of the second conference on creativity and innovation in design. 31–42

    2011

  5. [13]

    Laila Dybkjær, Niels Ole Bernsen, and Hans Dybkjær. 1996. Grice incorporated: cooperativity in spoken dialogue. InCOLING 1996 Volume 1: The 16th International Conference on Computational Linguistics

    1996

  6. [14]

    R. Frary. 1985. Multiple-Choice versus Free-Response: A Simulation Study.Journal of Educational Measurement22 (1985), 21–31. doi:10.1111/J.1745- 3984.1985.TB01046.X

    1985

  7. [15]

    Mirta Galesic and Michael Bosnjak. 2009. Effects of questionnaire length on participation and indicators of response quality in a web survey.Public opinion quarterly73, 2 (2009), 349–360

    2009

  8. [16]

    Elisa Giaccardi, Johan Redström, and Iohanna Nicenboim and. 2025. The making(s) of more-than-human design: introduction to the spe- cial issue on more-than-human design and HCI.Human–Computer Interaction40, 1-4 (2025), 1–16. doi:10.1080/07370024.2024.2353357 arXiv:https://doi....

    2025

  9. [17]

    Paul Grice. 1975. Logic and Conversation. InSyntax and Semantics, Volume 3: Speech Acts, Peter Cole and Jerry L. Morgan (Eds.). Academic Press, 41–58. 26 Daniel Hove Paludan, Julie Fredsgård, Kasper Patrick Bährentz, and Ilhan Aslan

    1975

  10. [18]

    Dirk Heerwegh and Geert Loosveldt. 2008. Face-to-face versus web surveying in a high-internet-coverage population: Differences in response quality.Public opinion quarterly72, 5 (2008), 836–846

    2008

  11. [19]

    Jiaxiong Hu, Jingya Guo, Ningjing Tang, Xiaojuan Ma, Yuan Yao, Changyuan Yang, and Yingqing Xu. 2024. Designing the conversational agent: Asking follow-up questions for information elicitation.Proceedings of the ACM on Human-Computer Interaction8, CSCW1 (2024), 1–30

    2024

  12. [20]

    Rune M Jacobsen, Samuel Rhys Cox, Carla F Griggio, and Niels van Berkel. 2025. Chatbots for Data Collection in Surveys: A Comparison of Four Theory-Based Interview Probes.arXiv preprint arXiv:2503.08582(2025)

    2025 arXiv

  13. [21]

    Soomin Kim, Joonhwan Lee, and Gahgene Gweon. 2019. Comparing Data from Chatbot and Web Surveys: Effects of Platform and Conversational Style on Survey Response Quality. InProceedings of the 2019 CHI Conference on Human Factors in Computing Systems(Glasgow, Scotland Uk)(CHI ’19...

    2019

  14. [22]

    Georgios Koronis, Hernan Casakin, and Arlindo Silva. 2021. Crafting briefs to stimulate creativity in the design studio.Thinking Skills and Creativity 40 (2021), 100810

    2021

  15. [23]

    Georgios Koronis, Pei Zhi Chia, Jacob Kang Kai Siang, Arlindo Silva, Christine Yogiaman, and Nilanjan Raghunath. 2019. An Empirical Study on the Impact of Design Brief Information on the Creativity of Design Outcomes With Consideration of Gender and Gender Diversity.Jour- nal ...

    2019 doi

  16. [24]

    Jon A Krosnick. 1991. Response strategies for coping with the cognitive demands of attitude measures in surveys.Applied cognitive psychology5, 3 (1991), 213–236

    1991

  17. [25]

    Claire Lauer, Danielle Storey, and Romit Soley. 2024. Vector Personas: How UX Researchers Can Use AI to Bring New Dimension to Traditional Persona Development. InProceedings of the 42nd ACM International Conference on Design of Communication. 151–157

    2024

  18. [26]

    Bettina Laugwitz, Theo Held, and Martin Schrepp. 2008. Construction and evaluation of a user experience questionnaire. InSymposium of the Austrian HCI and usability engineering group. Springer, 63–76

    2008

  19. [27]

    Seung Won Lee, Tae Hee Jo, Semin Jin, Jiin Choi, Kyungwon Yun, Sergio Bromberg, Seonghoon Ban, and Kyung Hoon Hyun. 2024. The Impact of Sketch-guided vs. Prompt-guided 3D Generative AIs on the Design Exploration Process. InProceedings of the CHI Conference on Human Factors in ...

    2024

  20. [28]

    Seo-young Lee, Matthew Law, and Guy Hoffman. 2024. When and How to Use AI in the Design Process? Implications for Human-AI Design Collaboration.International Journal of Human–Computer Interaction(2024), 1–16

    2024

  21. [29]

    Elsie Lee-Robbins, Arran Ridley, and Eytan Adar. 2024. Client-Designer Negotiation in Data Visualization Projects.IEEE Transactions on Visualization and Computer Graphics(2024)

    2024

  22. [30]

    Asghar Mohammad Moradi Leyla Alipour, Mohsen Faizi and Gholamreza Akrami. 2018. A review of design fixation: research direc- tions and key factors.International Journal of Design Creativity and Innovation6, 1-2 (2018), 22–35. doi:10.1080/21650349.2017.1320232 arXiv:https://doi...

    2018

  23. [31]

    Jie Li, Hancheng Cao, Laura Lin, Youyang Hou, Ruihao Zhu, and Abdallah El Ali. 2024. User Experience Design Professionals’ Perceptions of Generative Artificial Intelligence. arXiv:2309.15237 [cs.CY] https://arxiv.org/abs/2309.15237

    2024 arXiv

  24. [32]

    Zhe Liu. 2025. Interview AI-ssistant: Designing for Real-Time Human-AI Collaboration in Interview Preparation and Execution. InCompanion Proceedings of the 30th International Conference on Intelligent User Interfaces. 225–228

    2025

  25. [33]

    Lucas Memmert and Eva Bittner. 2022. Complex problem solving through human-AI collaboration: Literature review on research contexts. (2022)

    2022

  26. [34]

    Murugesan, Rashina Hoda, and Z

    L. Murugesan, Rashina Hoda, and Z. Salcic. 2017. Identifying Design Features Using Combination of Requirements Elicitation Techniques.2017 IEEE/ACM 1st International Workshop on Design and Innovation in Software Engineering (DISE)(2017), 6–12. doi:10.1109/DISE.2017.9

    2017 doi

  27. [35]

    Daniel Hove Paludan, Julie Fredsgård, Kasper Patrick Bährentz, Ilhan Aslan, and Niels van Berkel. 2025. Towards Sustainable Creativity Support: An Exploratory Study on Prompt Based Image Generation. arXiv:2504.07879 [cs.HC] https://arxiv.org/abs/2504.07879

    2025 arXiv

  28. [36]

    Nadia Piet. 2020. AI Ideation Card Deck. https://aixdesign.co/posts/ai-ideation-cards. Accessed: 2025-03-14

    2020

  29. [37]

    Bob Price. 2002. Laddered questions and qualitative data research interviews.Journal of advanced nursing37, 3 (2002), 273–281

    2002

  30. [38]

    2022.Collaboration in the Machine Age: Trustworthy Human-AI Collaboration

    Liana Razmerita, Armelle Brun, and Thierry Nabeth. 2022.Collaboration in the Machine Age: Trustworthy Human-AI Collaboration. Springer International Publishing, Cham, 333–356. doi:10.1007/978-3-030-93052-3_14

    2022 doi

  31. [39]

    Melanie Revilla and Carlos Ochoa. 2017. Ideal and maximum length for a web survey.International Journal of Market Research59, 5 (2017), 557–565

    2017

  32. [40]

    Jungwook Rhim, Minji Kwak, Yeaeun Gong, and Gahgene Gweon. 2022. Application of humanization to survey chatbots: Change in chatbot perception, interaction experience, and survey data quality.Computers in Human Behavior126 (2022), 107034

    2022

  33. [41]

    Schnotz and Christian Kürschner

    W. Schnotz and Christian Kürschner. 2008. External and internal representations in the acquisition and use of knowledge: visualization effects on mental model construction.Instructional Science36 (2008), 175–190. doi:10.1007/S11251-007-9029-2

    2008 doi

  34. [42]

    Linus Tan and Max Luhrs. 2024. Using Generative AI Midjourney to enhance divergent and convergent thinking in an architect’s creative design process.The Design Journal27, 4 (2024), 677–699

    2024

  35. [43]

    Xilin Tang, Jerrod Windham, and Benjamin Bush. 2025. Pre-AI and post-AI design: balancing human Creativity and AI Tools in the Industrial Design Process. InProceeding of the 2024 International Conference on Artificial Intelligence and Future Education (AIFE ’24). Association f...

    2025

  36. [44]

    Severi Uusitalo, Antti Salovaara, Tero Jokela, and Marja Salmimaa. 2024. ” Clay to Play With”: Generative AI Tools in UX and Industrial Design Practice. InProceedings of the 2024 ACM Designing Interactive Systems Conference. 1566–1578. Exploring conversational AI and response ...

    2024

  37. [45]

    Wobbrock, Meredith Ringel Morris, and Andrew D

    Jacob O. Wobbrock, Meredith Ringel Morris, and Andrew D. Wilson. 2009. User-defined gestures for surface computing. InProceedings of the SIGCHI Conference on Human Factors in Computing Systems(Boston, MA, USA)(CHI ’09). Association for Computing Machinery, New York, NY, USA, 1...

    2009

  38. [46]

    Alexander Wuttke, Matthias Aßenmacher, Christopher Klamm, Max M Lang, Quirin Würschinger, and Frauke Kreuter. 2024. AI Conversational Interviewing: Transforming Surveys with LLMs as Adaptive Interviewers.arXiv preprint arXiv:2410.01824(2024)

    2024 arXiv

  39. [47]

    Ziang Xiao, Michelle X. Zhou, Q. Vera Liao, Gloria Mark, Changyan Chi, Wenxi Chen, and Huahai Yang. 2020. Tell Me About Yourself: Using an AI-Powered Chatbot to Conduct Conversational Surveys with Open-ended Questions.ACM Transactions on Computer-Human Interaction27, 3 (June 2...

    2020 doi

  40. [48]

    Harin Yoon, Changhoon Oh, and Soojin Jun. 2024. How Can I Trust AI?: Extending a UXer-AI Collaboration Process in the Early Stages. InExtended Abstracts of the CHI Conference on Human Factors in Computing Systems. 1–7

    2024

  41. [49]

    Jiayi Zhou, Renzhong Li, Junxiu Tang, Tan Tang, Haotian Li, Weiwei Cui, and Yingcai Wu. 2024. Understanding nonlinear collaboration between human and AI agents: A co-design framework for creative design. InProceedings of the 2024 CHI Conference on Human Factors in Computing Sy...

    2024

  42. [50]

    Qi Zhou, Bin Li, Lei Han, and Min Jou. 2023. Talking to a bot or a wall? How chatbots vs. human agents affect anticipated communication quality. Computers in Human Behavior143 (2023), 107674

    2023

  43. [51]

    Zijian Zhu, Hyemin Lee, Younghwan Pan, and Pengyu Cai. 2024. AI assistance in enterprise UX design workflows: enhancing design brief creation for designers.Frontiers in Artificial Intelligence7 (2024), 1404647

    2024

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.