pith. sign in

arxiv: 2512.18388 · v2 · submitted 2025-12-20 · 💻 cs.HC · cs.AI

Exploration vs. Fixation: Scaffolding Divergent and Convergent Thinking for Human-AI Co-Creation with Generative Models

Chao Wen , Tung Phung , Pronita Mehrotra , Sumit Gulwani , Roger E. Beaty , Tomohiro Nagashima , Adish Singla This is my paper

Pith reviewed 2026-05-16 20:42 UTC · model grok-4.3

classification 💻 cs.HC cs.AI
keywords human-AI co-creationgenerative modelsdivergent thinkingconvergent thinkingcreativity supportimage generationuser studyGeneplore model
0
0 comments X

The pith

HAICo uses switchable divergent and convergent modes to scaffold creative image generation and outperforms ChatGPT in user tests.

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

Generative AI chatbots tend to produce finished images from the first prompt, which can anchor users to early ideas and limit how far they explore. The paper introduces HAICo, an interface that lets users toggle between a divergent mode for exploring distant concepts and a convergent mode for refining chosen ones, following the Geneplore model. In a within-subjects study, 24 participants created poster images and rated HAICo higher than ChatGPT on creativity dimensions and usability. The work shows that deliberately structuring the creative process into separate exploration and refinement phases improves outcomes over pure execution-focused tools. This points to the benefit of active guidance through creative stages rather than immediate artifact generation.

Core claim

The paper claims that an interface implementing switchable DIVERGENT and CONVERGENT modes, drawn directly from the Geneplore model, produces more creative and usable results in generative image tasks than standard chatbot interfaces, as shown by superior performance across multiple measures in a within-subjects experiment with 24 participants on a poster image creation task.

What carries the argument

HAICo system with its two switchable modes: DIVERGENT mode for broad exploration of remote conceptual ideas and CONVERGENT mode for targeted refinement of selected ideas.

Load-bearing premise

That directly scaffolding the Geneplore model's divergent and convergent phases in an AI interface will increase creativity without the mode-switching mechanism itself creating new fixation or bias.

What would settle it

A larger or cross-task replication study in which creativity metrics and usability ratings show no advantage or a disadvantage for HAICo compared with ChatGPT.

Figures

Figures reproduced from arXiv: 2512.18388 by Adish Singla, Chao Wen, Pronita Mehrotra, Roger E. Beaty, Sumit Gulwani, Tomohiro Nagashima, Tung Phung.

Figure 1
Figure 1. Figure 1: Workflow overview of HAIExplore. HAIExplore supports two stages: Brainstorming and Refinement, and allows users to switch between them at any time. In Brainstorming, the user provides an ideation prompt describing the desired image. Based on this prompt, HAIExplore generates diverse conceptual ideas, which the user can select, edit, or extend to create new ideas. When an idea seems promising, the user can … view at source ↗
Figure 2
Figure 2. Figure 2: The Brainstorming interface. (A) The user enters a prompt. (B) HAIExplore generates ideas and presents each as an Idea Card in the Idea Grid. Users can expand this set by (C) manually creating their own idea cards (e.g., the custom “Empty Bench Outside University” card shown) or by (D) clicking “More Ideas” to provide additional context and request further system-generated ideas. Once an idea card is selec… view at source ↗
Figure 3
Figure 3. Figure 3: Structure and interaction flow of an idea card. (Top Left) Each idea card contains five components: Title, Background, Description, Categories, and a Visual. The visual is to visualize the idea and help users better understand the idea. (Bottom Left) Users can edit the Title and Description to refine the idea, or (Right) click the spark icon to generate an image and an accompanying textual explanation that… view at source ↗
Figure 4
Figure 4. Figure 4: The Refinement Interface of HAIExplore. (A) The user opens a new tab to refine a base image selected from the Brainstorm tab. Multiple Refine tabs can be opened in parallel. (B) The user articulates their refinement intent in natural language and then clicks “Refine.” (C) HAIExplore analyzes the base image and the user’s refinement intent to generate dynamic Parameters and Options, exposing relevant parame… view at source ↗
Figure 5
Figure 5. Figure 5: Task instructions for “Spending Less Time on Phone” poster design used in the user study. [PITH_FULL_IMAGE:figures/full_fig_p012_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: User Study Procedure [40]. For details, see Section 5.2. We employed a within-subjects design to compare the creative workflows facilitated by HAIExplore and ChatGPT. To minimize learning effects, each participant completed two distinct image-creation sessions (one per system), using a different task topic for each. To control for task-specific variance and carryover effects, we employed a Balanced Incompl… view at source ↗
read the original abstract

Generative AI has democratized content creation, but popular chatbot-based interfaces often prioritize execution, generating fully rendered artifacts right away. This issue can lead to premature convergence and design fixation, where users are being anchored to initial outputs. Recent works have proposed new interfaces to address this issue by supporting exploration, though typically constrained to be semantically close to a user's initial task framing, potentially limiting the creativity of the outcomes. We examine an approach grounded in the Geneplore model of creative cognition and instantiate it in a human-AI co-creation system, HAICo, for creative image generation. HAICo explicitly structures the creative process into two switchable modes: DIVERGENT mode scaffolds the broad exploration of remote conceptual ideas; CONVERGENT mode supports a targeted refinement of selected ideas. Through a within-subjects study (N=24) on a poster image creation task, we demonstrate that HAICo outperforms ChatGPT across multiple dimensions of creativity and usability. Our results highlight the critical need to shift from pure execution-focused chatbots to scaffolded co-creation systems that actively guide exploration and foster the creative process.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper introduces HAICo, a human-AI co-creation interface for generative image tasks that explicitly scaffolds the Geneplore model's divergent (broad exploration of remote ideas) and convergent (targeted refinement) phases via switchable modes. In a within-subjects study (N=24) on poster image creation, it claims HAICo outperforms unmodified ChatGPT on multiple creativity and usability dimensions, arguing for a shift from execution-focused chatbots to structured co-creation systems.

Significance. If the causal link between the two-mode scaffolding and improved outcomes can be isolated, the work would provide actionable evidence for interface design in creative AI, highlighting how explicit phase support can mitigate fixation and promote remote associations beyond standard chatbot interactions.

major comments (2)
  1. [Abstract / Study description] Abstract and study description: the headline claim attributes superiority to the explicit DIVERGENT/CONVERGENT Geneplore scaffolding, yet the control is unmodified ChatGPT rather than an ablated HAICo variant (identical UI shell and generator but with modes collapsed). This leaves open confounds from interface structure, prompt differences, order effects, or novelty in the within-subjects design, so the load-bearing causal attribution is not isolated.
  2. [Abstract] Abstract: no information is supplied on the specific creativity metrics, statistical tests, operationalization of 'remote ideas,' or controls for order effects, preventing verification that the reported gains are robust and directly tied to the scaffolding mechanism.
minor comments (1)
  1. [Abstract] The abstract refers to 'multiple dimensions of creativity and usability' without naming them or citing the measurement instruments; adding these details would improve clarity even if the core design issue is addressed.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. The comments highlight important issues around causal attribution and abstract clarity. We address each point below, clarify our design rationale, and indicate revisions to strengthen the paper while maintaining the integrity of the reported comparison to the standard chatbot interface.

read point-by-point responses

  1. Referee: [Abstract / Study description] Abstract and study description: the headline claim attributes superiority to the explicit DIVERGENT/CONVERGENT Geneplore scaffolding, yet the control is unmodified ChatGPT rather than an ablated HAICo variant (identical UI shell and generator but with modes collapsed). This leaves open confounds from interface structure, prompt differences, order effects, or novelty in the within-subjects design, so the load-bearing causal attribution is not isolated.

    Authors: We acknowledge that the current control condition (unmodified ChatGPT) does not isolate the precise contribution of the switchable divergent/convergent modes from other interface differences such as structured prompting or mode switching UI. Our primary goal was to evaluate HAICo against the de facto standard chatbot interface that users currently employ for image generation tasks, thereby demonstrating practical advantages in a realistic setting. We agree an ablation (identical shell with modes collapsed) would provide stronger causal evidence and will add an explicit limitations section discussing potential confounds including UI novelty, prompt engineering differences, and order effects (which were mitigated via counterbalancing). We have revised the abstract and study description to clearly state that the comparison is to the baseline unmodified ChatGPT rather than an ablated HAICo variant, and we outline future ablation experiments. revision: yes

  2. Referee: [Abstract] Abstract: no information is supplied on the specific creativity metrics, statistical tests, operationalization of 'remote ideas,' or controls for order effects, preventing verification that the reported gains are robust and directly tied to the scaffolding mechanism.

    Authors: The abstract is length-constrained, but the full manuscript details the creativity metrics (fluency, flexibility, and originality drawn from established creative cognition scales), statistical tests (paired t-tests with reported effect sizes and p-values), operationalization of remote ideas (via semantic embedding distance from the initial prompt), and order-effect controls (counterbalanced task order in the within-subjects design). We will expand the abstract to include the primary metrics, key statistical outcomes, and a brief note on counterbalancing to improve verifiability without exceeding length limits. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical user study provides independent evidence for HAICo design

full rationale

The paper describes an interface (HAICo) that instantiates the established Geneplore model via explicit DIVERGENT and CONVERGENT modes and reports results from a within-subjects study (N=24) comparing it to ChatGPT on a poster task. No equations, fitted parameters, self-definitions, or load-bearing self-citations appear in the derivation chain; the central claims rest on measured outcomes rather than any reduction of predictions to inputs by construction. The Geneplore grounding is a standard external citation, not an author-derived uniqueness theorem or ansatz smuggled via prior work. This is a standard empirical HCI contribution with no detectable circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The claim depends on the applicability of the Geneplore model to AI-mediated image tasks and on the effectiveness of the newly implemented modes, neither of which receives independent validation outside the reported study.

axioms (1)
  • domain assumption The Geneplore model of creative cognition accurately describes processes that can be scaffolded in generative AI interfaces for image creation.
    The system is explicitly built as an instantiation of this model.
invented entities (1)
  • HAICo no independent evidence
    purpose: To provide switchable divergent and convergent scaffolding for human-AI image co-creation.
    Newly proposed system whose effectiveness is tested only within the paper's own study.

pith-pipeline@v0.9.0 · 5530 in / 1293 out tokens · 22468 ms · 2026-05-16T20:42:47.719274+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Beyond Compliance: How AI Could Help Creative Writers by Refusing Them

    cs.HC 2026-04 unverdicted novelty 6.0

    A qualitative study with 22 creative writers finds that the reflective value of AI refusals depends on alignment with users' situational thinking phases, cognitive beliefs, and views of AI roles.

Reference graph

Works this paper leans on

61 extracted references · 61 canonical work pages · cited by 1 Pith paper · 1 internal anchor

  1. [1]

    2025.Learning, Reinvented: Accelerating Human–AI Collaboration

    Accenture. 2025.Learning, Reinvented: Accelerating Human–AI Collaboration. Technical Report

    work page 2025

  2. [2]

    Robert E Adamson. 1952. Functional Fixedness as Related to Problem Solving: A Repetition of Three Experiments.Journal of Experimental Psychology 44 (1952)

    work page 1952

  3. [3]

    Ahmed, Maria Christakis, Aleksandr Efremov, Nigel Fernandez, Ahana Ghosh, Abhik Roychoudhury, and Adish Singla

    Umair Z. Ahmed, Maria Christakis, Aleksandr Efremov, Nigel Fernandez, Ahana Ghosh, Abhik Roychoudhury, and Adish Singla. 2020. Synthesizing Tasks for Block-based Programming. InAdvances in Neural Information Processing Systems: Annual Conference on Neural Information Processing Systems (NeurIPS)

    work page 2020

  4. [4]

    Ravinithesh Annapureddy, Alessandro Fornaroli, and Daniel Gatica-Perez. 2025. Generative AI Literacy: Twelve Defining Competencies.Digital Government: Research and Practice6 (2025)

    work page 2025

  5. [5]

    Roger E Beaty and Yoed N Kenett. 2023. Associative Thinking at the Core of Creativity.Trends in Cognitive Sciences27 (2023)

    work page 2023

  6. [6]

    Stephen Brade, Bryan Wang, Maurício Sousa, Sageev Oore, and Tovi Grossman. 2023. Promptify: Text-to-Image Generation through Interactive Prompt Exploration with Large Language Models. InSymposium on User Interface Software and Technology (UIST)

    work page 2023

  7. [7]

    Virginia Braun and Victoria Clarke. 2006. Using Thematic Analysis in Psychology.Qualitative Research in Psychology3 (2006)

    work page 2006

  8. [8]

    Erin Cherry and Celine Latulipe. [n. d.].Transactions on Computer-Human Interaction (TOCHI), title = Quantifying the Creativity Support of Digital Tools Through the Creativity Support Index, volume = 21, year = 2014([n. d.])

    work page 2014

  9. [9]

    Victoria Clarke and Virginia Braun. 2017. Thematic Analysis.The Journal of Positive Psychology12 (2017)

    work page 2017

  10. [10]

    Arthur Cropley. 2006. In Praise of Convergent Thinking.Creativity Research Journal18 (2006)

    work page 2006

  11. [11]

    Fiona Draxler, Anna Werner, Florian Lehmann, Matthias Hoppe, Albrecht Schmidt, Daniel Buschek, and Robin Welsch. 2024. The AI Ghostwriter Effect: When Users Do Not Perceive Ownership of AI-Generated Text but Self-Declare as Authors.Transactions on Computer-Human Interaction (TOCHI)31 (2024)

    work page 2024

  12. [12]

    Ziv Epstein, Aaron Hertzmann, Investigators of Human Creativity, Memo Akten, Hany Farid, Jessica Fjeld, Morgan R Frank, Matthew Groh, Laura Herman, Neil Leach, and Others. 2023. Art and the Science of Generative AI.Science380 (2023)

    work page 2023

  13. [13]

    Baptiste Rozière et al. 2023. Code Llama: Open Foundation Models for Code.CoRRabs/2308.12950 (2023)

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  14. [14]

    Yingchaojie Feng, Xingbo Wang, Kamkwai Wong, Sijia Wang, Yuhong Lu, Minfeng Zhu, Baicheng Wang, and Wei Chen. 2024. PromptMagician: Interactive Prompt Engineering for Text-to-Image Creation.Transactions on Visualization and Computer Graphics30 (2024)

    work page 2024

  15. [15]

    Akhilesh Gadde. 2025. Democratizing Software Engineering Through Generative Ai and Vibe Coding: The Evolution of No-code Development. Journal of Computer Science and Technology Studies7 (2025)

    work page 2025

  16. [16]

    Ahana Ghosh, Sebastian Tschiatschek, Sam Devlin, and Adish Singla. 2022. Adaptive Scaffolding in Block-Based Programming via Synthesizing New Tasks as Pop Quizzes. InArtificial Intelligence in Education (AIED)

    work page 2022

  17. [17]

    Joy Paul Guilford. 1956. The Structure of Intellect.Psychological Bulletin53 (1956)

    work page 1956

  18. [18]

    Joy Paul Guilford. 1967. The Nature of Human Intelligence. (1967)

    work page 1967

  19. [19]

    Jennifer Haase and Sebastian Pokutta. 2024. Human-AI Co-creativity: Exploring Synergies Across Levels of Creative Collaboration.CoRR abs/2411.12527 (2024)

    work page arXiv 2024

  20. [20]

    Evans Xu Han, Alice Qian Zhang, Haiyi Zhu, Hong Shen, Paul Pu Liang, and Jane Hsieh. 2025. POET: Supporting Prompting Creativity and Personalization with Automated Expansion of Text-to-Image Generation. InSymposium on User Interface Software and Technology

    work page 2025

  21. [21]

    Kent F Hubert, Kim N Awa, and Darya L Zabelina. 2024. The current state of artificial intelligence generative language models is more creative than humans on divergent thinking tasks.Scientific Reports14 (2024). 24 Wen et al

    work page 2024

  22. [22]

    Mina Huh, Ding Li, Kim Pimmel, Hijung Valentina Shin, Amy Pavel, and Mira Dontcheva. 2025. VideoDiff: Human-AI Video Co-Creation with Alternatives. InConference on Human Factors in Computing Systems (CHI)

    work page 2025

  23. [23]

    David G Jansson and Steven M Smith. 1991. Design Fixation.Design Studies12 (1991)

    work page 1991

  24. [24]

    Nicholas W Kohn and Steven M Smith. 2011. Collaborative Fixation: Effects of Others’ Ideas on Brainstorming.Applied Cognitive Psychology25 (2011)

    work page 2011

  25. [25]

    Harsh Kumar, Jonathan Vincentius, Ewan Jordan, and Ashton Anderson. 2025. Human Creativity in the Age of LLMs: Randomized Experiments on Divergent and Convergent Thinking. InConference on Human Factors in Computing Systems (CHI)

    work page 2025

  26. [26]

    2017.Research methods in human-computer interaction

    Jonathan Lazar, Jinjuan Heidi Feng, and Harry Hochheiser. 2017.Research methods in human-computer interaction. Morgan Kaufmann

    work page 2017

  27. [27]

    Lewis, Brian Utesch, and Deborah E

    James R. Lewis, Brian Utesch, and Deborah E. Maher. 2013. UMUX-LITE: When There’s No Time for the SUS. InConference on Human Factors in Computing Systems (CHI)

    work page 2013

  28. [28]

    Tong Li. 2025. AI-Driven Multi-Modal Interactive Learning Environment Using Deep Learning and Chain-of-Thought Reasoning.International Journal of Pattern Recognition and Artificial Intelligence39 (2025)

    work page 2025

  29. [29]

    Sarnoff Mednick. 1962. The Associative Basis of the Creative Process.Psychological Review69 (1962)

    work page 1962

  30. [30]

    Pronita Mehrotra, Aishni Parab, and Sumit Gulwani. 2024. Enhancing Creativity in Large Language Models Through Associative Thinking Strategies. CoRRabs/2405.06715 (2024)

    work page arXiv 2024

  31. [31]

    Aditi Mishra, Frederik Brudy, Qian Zhou, George Fitzmaurice, and Fraser Anderson. 2025. WhatIF: Branched Narrative Fiction Visualization for Authoring Emergent Narratives using Large Language Models. InConference on Creativity and Cognition

    work page 2025

  32. [32]

    Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, and Caiming Xiong. 2023. CodeGen: An Open Large Language Model for Code with Multi-Turn Program Synthesis. InInternational Conference on Learning Representations (ICLR)

    work page 2023

  33. [33]

    Michael Jonathan Reiss and Haiying Liang. [n. d.]. The Associations Between Students’ Attitudes Towards AI and Learning Engagement: Serial Mediating Roles of Perceived Autonomy and Learning Enjoyment.Frontiers in Psychology16 ([n. d.])

    work page

  34. [34]

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Björn Ommer. 2022. High-Resolution Image Synthesis with Latent Diffusion Models. InConference on Computer Vision and Pattern Recognition (CVPR)

    work page 2022

  35. [35]

    1991.Divergent Thinking

    Mark A Runco. 1991.Divergent Thinking

    work page 1991

  36. [36]

    Zainab Salma, Raquel Hijón-Neira, and Celeste Pizarro. 2025. Designing Co-Creative Systems: Five Paradoxes in Human–AI Collaboration. Information16 (2025)

    work page 2025

  37. [37]

    2017.The Reflective Practitioner: How Professionals Think in Action

    Donald A Schön. 2017.The Reflective Practitioner: How Professionals Think in Action

    work page 2017

  38. [38]

    Christoph Schuhmann, Romain Beaumont, Richard Vencu, Cade Gordon, Ross Wightman, Mehdi Cherti, Theo Coombes, Aarush Katta, Clayton Mullis, Mitchell Wortsman, Patrick Schramowski, Srivatsa Kundurthy, Katherine Crowson, Ludwig Schmidt, Robert Kaczmarczyk, and Jenia Jitsev

    work page

  39. [39]

    InAdvances in Neural Information Processing Systems 35: Annual Conference on Neural Information Processing Systems (NeurIPS)

    LAION-5B: An Open Large-scale Dataset for Training Next Generation Image-text Models. InAdvances in Neural Information Processing Systems 35: Annual Conference on Neural Information Processing Systems (NeurIPS)

    work page

  40. [40]

    Barry Schwartz. 2005. The Paradox of Choice: Why More Is Less (PS).New York(2005)

    work page 2005

  41. [41]

    Hanshu Shen, Lyukesheng Shen, Wenqi Wu, and Kejun Zhang. 2025. IdeationWeb: Tracking the Evolution of Design Ideas in Human-AI Co-Creation. InConference on Human Factors in Computing Systems (CHI)

    work page 2025

  42. [42]

    Hariharan Subramonyam, Roy Pea, Christopher Lawrence Pondoc, Maneesh Agrawala, and Colleen M. Seifert. 2024. Bridging the Gulf of Envisioning: Cognitive Challenges in Prompt Based Interactions with LLMs. InConference on Human Factors in Computing Systems (CHI)

    work page 2024

  43. [43]

    Sangho Suh, Meng Chen, Bryan Min, Toby Jia-Jun Li, and Haijun Xia. 2024. Luminate: Structured Generation and Exploration of Design Space with Large Language Models for Human-AI Co-Creation. InConference on Human Factors in Computing Systems (CHI)

    work page 2024

  44. [44]

    Sangho Suh, Jian Zhao, and Edith Law. 2022. CodeToon: Story Ideation, Auto Comic Generation, and Structure Mapping for Code-Driven Storytelling. InSymposium on User Interface Software and Technology (UIST)

    work page 2022

  45. [45]

    Sirui Tao, Ivan Liang, Cindy Peng, Zhiqing Wang, Srishti Palani, and Steven P. Dow. 2025. DesignWeaver: Dimensional Scaffolding for Text-to-Image Product Design. InConference on Human Factors in Computing Systems (CHI)

    work page 2025

  46. [46]

    Bekzat Tilekbay, Saelyne Yang, Michal Adam Lewkowicz, Alex Suryapranata, and Juho Kim. 2024. Expressedit: Video Editing with Natural Language and Sketching. InInternational Conference on Intelligent User Interfaces (IUI)

    work page 2024

  47. [47]

    René Vidal. 2010. Creative Problem Solving: An Applied University Course.Pesquisa Operacional30 (2010), 405–426

    work page 2010

  48. [48]

    Kelly, Saumya Pareek, Qiushi Zhou, and Eduardo Velloso

    Samangi Wadinambiarachchi, Ryan M. Kelly, Saumya Pareek, Qiushi Zhou, and Eduardo Velloso. 2024. The Effects of Generative AI on Design Fixation and Divergent Thinking. InConference on Human Factors in Computing Systems (CHI)

    work page 2024

  49. [49]

    1926.The Art of Thought

    Graham Wallas. 1926.The Art of Thought

    work page 1926

  50. [50]

    Chen, and Bing-Yu Chen

    Wen-Fan Wang, Ting-Ying Lee, Chien-Ting Lu, Che-Wei Hsu, Nil Ponsa Campanyà, Yu Chen, Mike Y. Chen, and Bing-Yu Chen. 2025. GenTune: Toward Traceable Prompts to Improve Controllability of Image Refinement in Environment Design. InSymposium on User Interface Software and Technology (UIST)

    work page 2025

  51. [51]

    Wen-Fan Wang, Chien-Ting Lu, Nil Ponsa Campanyà, Bing-Yu Chen, and Mike Y. Chen. 2025. AIdeation: Designing a Human-AI Collaborative Ideation System for Concept Designers. InConference on Human Factors in Computing Systems (CHI)

    work page 2025

  52. [52]

    Zhijie Wang, Yuheng Huang, Da Song, Lei Ma, and Tianyi Zhang. 2024. PromptCharm: Text-to-Image Generation through Multi-modal Prompting and Refinement. Exploration vs. Fixation: Scaffolding Divergent and Convergent Thinking for Human-AI Co-Creation 25

    work page 2024

  53. [53]

    Chao Wen, Ahana Ghosh, Jacqueline Staub, and Adish Singla. 2024. Task Synthesis for Elementary Visual Programming in XLogoOnline Environment. InAIED Track on Late Breaking Results

    work page 2024

  54. [54]

    Chao Wen, Jacqueline Staub, and Adish Singla. 2025. Program Synthesis Benchmark for Visual Programming in XLogoOnline Environment. In Annual Meeting of the Association for Computational Linguistics (ACL)

    work page 2025

  55. [55]

    Frank Wilcoxon. 1992. Individual Comparisons by Ranking Methods. InBreakthroughs in Statistics: Methodology and Distribution. Springer

    work page 1992

  56. [56]

    Yuxin Xu, Mengqiu Cheng, and Anastasia Kuzminykh. 2024. What Makes It Mine? Exploring Psychological Ownership Over Human-AI Co-Creations. InGraphics Interface Conference (GI)

    work page 2024

  57. [57]

    Ryan Yen, Jian Zhao, and Daniel Vogel. 2025. Code Shaping: Iterative Code Editing with Free-form AI-Interpreted Sketching. InConference on Human Factors in Computing Systems (CHI)

    work page 2025

  58. [58]

    Yongsheng Yu, Ziyun Zeng, Hang Hua, Jianlong Fu, and Jiebo Luo. 2024. PromptFix: You Prompt and We Fix the Photo. InAdvances in Neural Information Processing Systems 38: Annual Conference on Neural Information Processing Systems (NeurIPS)

    work page 2024

  59. [59]

    J. D. Zamfirescu-Pereira, Richmond Y. Wong, Bjoern Hartmann, and Qian Yang. 2023. Why Johnny Can’t Prompt: How Non-AI Experts Try (and Fail) to Design LLM Prompts. InConference on Human Factors in Computing Systems (CHI)

    work page 2023

  60. [60]

    Hanqing Zhang, Haolin Song, Shaoyu Li, Ming Zhou, and Dawei Song. 2024. A Survey of Controllable Text Generation Using Transformer-based Pre-trained Language Models.Comput. Surveys56 (2024)

    work page 2024

  61. [61]

    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. InConference on Human Factors in Computing Systems (CHI)

    work page 2024