Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 70 of the 70 references it cites were found in a registry or an index; 5 of the 15 sources its claims cite were checked for support.

The Effects of Generative AI on Design Fixation and Divergent Thinking

Open access
Wadinambiarachchi, Samangi, Kelly, Ryan M., Pareek, Saumya, Zhou, Qiushi, Velloso, Eduardo · 2024 · Proceedings of the CHI Conference on Human Factors in Computing Systems

Abstract

Generative AI systems have been heralded as tools for augmenting human creativity and inspiring divergent thinking, though with little empirical evidence for these claims. This paper explores the effects of exposure to AI-generated images on measures of design fixation and divergent thinking in a visual ideation task. Through a between-participants experiment (N=60), we found that support from an AI image generator during ideation leads to higher fixation on an initial example. Participants who used AI produced fewer ideas, with less variety and lower originality compared to a baseline. Our qualitative analysis suggests that the effectiveness of co-ideation with AI rests on participants’ chosen approach to prompt creation and on the strategies used by participants to generate ideas in response to the AI’s suggestions. We discuss opportunities for designing generative AI systems for ideation support and incorporating these AI tools into ideation workflows.

Overview

CoChat Summary

TL;DR

In a between-participants experiment (N=60), novice designers who used an AI image generator (Midjourney V4) for inspiration during a chatbot-avatar sketching task showed higher design fixation on the provided example and produced fewer, less varied, and less original ideas than a no-support baseline. The authors argue fixation can arise at prompt creation and at ideation from AI images, and describe a "fixation displacement" phenomenon where users shift fixation from the example onto the AI's outputs.

Key findings

  • GenAI support increased design fixation relative to baseline (mean = .32, 89% CI [.11, .55]; Bayes Factor = 124, "extreme" support), and Image Search also increased it but less so (mean = .27, 89% CI [.05, .49]; BF = 42.48).
  • Fluency was lower in both tool conditions: GenAI total effect mean = -.21 (89% CI [-.7, .28], BF = 3.20, 76% posterior probability of a negative effect); Image Search mean = -.49 (89% CI [-.9, -.10], BF = 46.62, 98% posterior probability).
  • Variety was lower for both stimuli (GenAI total effect mean = -.29, 89% CI [-.75, .16], BF = 5.93; Image Search mean = -.40, [-.87, -.07], BF = 11.38); Image Search's effect was fully mediated by Fluency, while GenAI retained an additional negative direct effect.
  • Originality was slightly lower for GenAI (mean = -.03, 89% CI [-.06, .00], BF = 35) and negligibly lower for Image Search (mean = -0.01, [-.04, .01], BF = 3.9).
  • Qualitative/descriptive results: 44% (206/468) of AI-generated images portrayed humanoid robots conceptually similar to the example; 44% of prompts (52/117) contained at least one word from the design brief, and 66.6% (78/117) included robot-related terms alongside brief words; a moderate positive correlation (Spearman ρ = 0.56) was found between the fixation score of sketches and that of the immediately preceding AI images (82 sketches analysed).

Methods

  • Between-participants lab experiment with three conditions of the independent variable "Inspiration Stimulus": Baseline (no support), Image Search (Google Images), and GenAI (paid Midjourney V4 via Discord); 60 participants (mean age 25.8, range 18–49, SD = 5.4; 10 women and 10 men per condition), screened for visual design experience.
  • Task: sketch as many chatbot-avatar ideas as possible in 20 minutes after being shown a written brief containing an example avatar with 14 salient features; sessions lasted 45–60 min and included pre/post questionnaires (NASA-TLX) and a 15–20 min semi-structured interview.
  • Analysis: Bayesian multilevel models in brms/Stan (binomial for design fixation, negative binomial for fluency and variety, linear regression for originality; 89% compatibility intervals and Bayes Factors), plus reflexive thematic analysis of interviews; two blind raters scored sketches (277 sketches grouped into 83 clusters).

Limitations

  • Stated: generative AI tools were nascent and use was "somewhat naïve"; the fixed 20-minute task limits insights to short-term rapid ideation; participants were screened for visual design skill but few had professional experience, so no claims about expertise; design fixation was operationalised only via denotative features of the example, excluding connotative aspects (art style, emotional expression, cultural references); only image generators in a visual ideation task were tested, so transfer to other modalities is unknown.
  • Stated: the AI system did not return results instantaneously, potentially delaying the GenAI condition, though Image Search participants (who got instant results) also showed lower fluency, possibly due to scrolling and added decision-making.
  • Evident from the text: the study reports Bayesian posterior estimates and Bayes Factors rather than frequentist significance, and the authors explicitly caution that no "statistical significance" claims should be derived; several effects (e.g., GenAI on fluency and variety) have compatibility intervals spanning zero.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1145/3613904.3642919

Cites

70 of 70 references verified0 citation errors · 0 not checkable

Statistics

Verification

Claims the paper makes

4 Valid · 0 Not valid · 8 Not checked · 58 Own results and interpretations

30 more claims

Create a free account to see all 70 claims and to verify the 1 reference still open.

See all claims — free account
  • Probably not a substantial claim. 33
  • It may state two separate claims as one. 9
  • The quote only partly states this claim. 1

Create a free account to see these claims.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

3 more claims about this paper

Create a free account to see every claim other papers make about this one.

See all claims — free account

Statistics recomputed

Every reported test result, recomputed from its test statistic and degrees of freedom. A result shows as not adding up only once an independent check has confirmed it.

  • ConsistentF(2, 57) = 1reported p = .37 · recomputed p = .374

    “F(2, 57) = 1, p = .37”

1 of 1

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

2 of 2

Citations

Cites

Show all 66 references

Research works printed but not linked: 6

  1. Carina Andersson, Yvonne Eriksson, Lasse Frank, and Bill Nicholl. 2012. Design fixations among information design students: What has been seen cannot be unseen. In DS 74: Proceedings of the 14th International Conference on Engineering & Product Design Education (E&PDE12) Design Education for Future Wellbeing,. Design Society, Antwerp, Belguim. https://www.designsociety.org/download-publication/33183/design_fixations_among_information_design_students_what_has_been_seen_cannot_be_unseen
  2. Audi MediaCenter. 2022. Reinventing the wheel? “FelGAN” inspires new rim designs with AI | Audi MediaCenter. https://www.audi-mediacenter.com/en/press-releases/reinventing-the-wheel-felgan-inspires-new-rim-designs-with-ai-15097
  3. N. Davis, S. Siddiqui, P. Karimi, M.L. Maher, and K. Grace. 2019. Creative sketching partner: A co-creative sketching tool to inspire design creativity. In Proceedings of the 10th International Conference on Computational Creativity, ICCC 2019. Association for Computational Creativity, North Carolina, 358–359.
  4. Edward de Bono. 2008. Six Thinking Hats (revised edition ed.). Penguin, United Kingdom.
  5. Thomas Howard, Anja Maier, Balder Onarheim, and Morten. Friis-Olivarius. 2013. Overcoming design fixation through education and creativity methods. In Proceedings of the International Conference on Engineering Design, ICED, Vol. 7 DS75-07. The Design Society, Seoul Korea, 139–148. https://www.designsociety.org/download-publication/34578/overcoming_design_fixation_through_education_and_creativity_methods
  6. L.A. Vasconcelos, M.A. Neroni, and N. Crilly. 2016. Fluency results in design fixation experiments: An additional explanation. In 4th International Conference on Design Creativity, ICDC 2016. The Design Society, Atlanta, GA, USA, 1–8.

Cited by

See all 320

Metadata

Authors
Wadinambiarachchi, S., Kelly, R. M., Pareek, S., Zhou, Q., & Velloso, E.
Affiliations
The University of Melbourne (AU)
Venue
Proceedings of the CHI Conference on Human Factors in Computing Systems
Publisher
ACM
Published
2024-05-11
Pages
1-18
OpenAlex
W4392972228
Cited by
320
References
66
Citations by year
2026: 197 · 2025: 108 · 2024: 15
FWCI
358.48
Type
conference paper
Language
en
Open access
Locations
  • CHI Conference on Human Factors in Computing Systems (publishedVersion, cc-by)
  • arXiv (Cornell University) (submittedVersion, cc-by)
  • Minerva Access (University of Melbourne) (submittedVersion, cc-by)
  • arXiv (Cornell University) (cc-by)
Funders
  • Rowden White Scholarship, The University of Melbourne
  • Melbourne Research Scholarship, The University of Melbourne
Topics
  • Physical Sciences › Engineering › Mechanical Engineering › Design Education and Practice
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Creativity in Education and Neuroscience
  • Physical Sciences › Computer Science › Computer Graphics and Computer-Aided Design › Digital Media and Visual Art
Keywords
generative AI, design fixation, divergent thinking, AI image generation, idea originality

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-09, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-09 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: The Effects of Generative AI on Design Fixation and Divergent Thinking. https://app.cochat.ai/papers/10.1145/3613904.3642919
The paper
Wadinambiarachchi, S., Kelly, R. M., Pareek, S., Zhou, Q., & Velloso, E. (2024). The Effects of Generative AI on Design Fixation and Divergent Thinking. Proceedings of the CHI Conference on Human Factors in Computing Systems. https://doi.org/10.1145/3613904.3642919

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.