Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

Polymind: Parallel Visual Diagramming with Large Language Models to Support Prewriting Through Microtasks

T0 review · 3 major / 5 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 Polymind argues that prewriting with large language models should be a parallel, diagram-based collaboration of small configurable microtasks, not a back-and-forth chat.

desk verdict A genuine new workflow with an honest exploratory evaluation, but the causal claim about parallelism is not isolated and the quantitative support is thin; referee-worthy with revisions. read the letter →

arxiv 2502.09577 v2 pith:A6BTGR27 submitted 2025-02-13 cs.HC

Qian Wan , Jiannan Li , Huanchen Wang , Zhicong Lu This is my paper · ORCID
classification cs.HC
keywords prewritingdiagrammingcreativitysupportmicrotaskinghuman-AIcollaborationlargelanguagemodelsmixed-initiativeinteractionparallelthinking
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

Prewriting mixes divergent idea generation with convergent organising and is naturally visual and iterative, which makes turn-taking chatbot conversations a poor fit. Polymind's central proposal is to replace the single conversational LLM with a team of small, independent 'microtasks'—Brainstorm, Elaborate, Summarise, Draft, Freewrite, Associate, and user-defined ones—that run in parallel on a diagramming canvas as virtual collaborators. Users configure prompts, input/output diagram types, visibility, and initiative, so they can keep control during convergent work and let the AI take the lead during divergent brainstorming. In a study with ten non-expert writers, participants reported more customizability, agency, and control with Polymind than with a turn-taking conversational interface paired with a bare canvas, and described quickly expanding personalised idea trees. The two conditions did not differ significantly on expert-rated outline quality, so the paper's claim is about the collaboration experience rather than about producing better final outlines.

What carries the argument

The load-bearing object is the microtask: a small, independent LLM operation with an input diagram type (keyword, concept, sticky note, or section), an output diagram type, and a prompt template that fills in a placeholder with the node's text. The mechanism is parallelism: every few seconds several microtasks sample diagram nodes of each input type, process them concurrently, and post their results back onto the canvas, so divergent and convergent angles of thought are available at the same time. Task headers attached to nodes show notifications and previews, task cards on a task board hold each microtask's specifications, and initiative can be toggled between proactive and reactive modes, giving the writer mixed-initiative control over interruption and pace. Chaining happens when a generated node becomes the input for another microtask, which participants used to expand a small concept into a structured idea tree.

What would settle it

Run the same two-session prewriting study in three conditions: full Polymind, a version of Polymind with all microtasks available but only one allowed to execute at a time, and a turn-taking conversational interface with a bare canvas. If the single-microtask condition matches full Polymind on perceived customizability, agency, and speed of idea expansion, the paper's central claim about parallel execution is not supported.

Watch

Extended reading notes

Core claim

On its own terms, the discovery is that the bottleneck in human-LLM prewriting is interaction structure rather than model capability. Polymind operationalises parallel thinking by having several heterogeneous microtask agents work on the same diagram at once, each one a named LLM operation with a prompt template and an input/output diagram type. Results return to the canvas as hollow, colour-coded nodes that users can accept, discard, ask to explain, or regenerate through short feedback buttons; proactive microtasks sample nodes near the user's attention using a selection-cost formula based on mouse distance and node width, while reactive microtasks wait for a click. The study's finding is that this arrangement let writers act as managers of the collaboration: they re-prompted local pieces instead of rewriting full prompts, chained microtasks to grow idea trees quickly, and reported feeling that the results were their own ideas. The authors frame the contribution as a workflow and interface for human-AI co-creativity, not as a system that scores higher on final output quality.

Load-bearing premise

The study's baseline changed two things at once—it replaced the turn-taking chat with diagram-based microtasks and removed parallel execution—so the paper's conclusion that parallel microtasking is what improves the collaboration assumes that the benefits come from parallelism rather than from the diagram-based interface itself.

Editorial extensions

If this is right

  • If the workflow holds up, writers no longer need to compose long context-heavy prompts: they delegate small jobs to microtasks and accept, discard, or regenerate the resulting nodes.
  • Chaining microtasks becomes a first-class interaction, so a single concept can expand into a structured idea tree in minutes through operations such as Freewrite followed by Summarise.
  • Mixed-initiative control becomes practical: keep microtasks proactive during divergent brainstorming and switch them to reactive when focusing on convergent outlining.
  • LLM randomness becomes less disruptive, because outputs are short, localised, colour-coded, and attributable to a specific microtask instead of buried in a long conversational reply.
  • The design space extends beyond prewriting to any canvas-based activity—such as sketching, visual programming, or sensemaking—where multiple LLM roles can run in parallel.

Reading between the lines

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

  • Editorial inference: because the evaluation changed both modality and concurrency at once, a three-condition study that includes a serial diagram-microtask version of Polymind would isolate whether the benefit truly comes from running microtasks in parallel.
  • Editorial inference: the microtask abstraction should transfer to non-writing canvases; likely next tests are storyboarding, visual programming, and argument mapping, measuring whether the same task cards, headers, and initiative toggles remain useful.
  • Editorial inference: the paper's finding that Polymind outputs were shorter and more writer-owned suggests the value may be in the process—agency and exploration—rather than in the immediate outline; following users from prewriting to a finished draft would test whether that early agency compounds into better writing.
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 / 5 minor

Summary. Polymind is a visual diagramming tool that supports prewriting by running multiple configurable LLM-powered 'microtasks' in parallel on a canvas. The paper motivates the design through a formative study of LLM-assisted prewriting, derives three design goals (scaffolding diagramming through microtasks, task management, and mixed initiative), and reports a within-subjects evaluation with 10 non-expert writers comparing Polymind against a baseline consisting of ChatGPT-4 plus Polymind's canvas with all microtask features disabled. The main claims are that the parallel microtask workflow affords more customizability, agency, and control than turn-taking conversation, and that this helps users quickly expand personalized writing ideas. Quantitative results are largely non-significant: CSI expressiveness p=0.05, enjoyment p=0.67, exploration p=0.19, median idea counts were 3 in both conditions, and TTCW scores were 4 for Polymind versus 5.7 for the baseline (p=0.23). The paper's support for its central claims therefore rests mostly on qualitative interview data.

Significance. If the central claim were established, Polymind would be a useful contribution to human-AI co-creativity and prewriting support: it offers a concrete, well-specified workflow for orchestrating multiple LLM agents on a diagramming canvas, with a detailed interface design and a functional implementation. The formative study is sensible, the system description is thorough, and the qualitative findings about agency, controllability, and microtask chaining are plausible and substantiated by participant quotes. The paper is not a derivation or prediction paper, so circularity is not an issue. However, the evaluation as designed does not isolate the effect of parallelism, and the quantitative outcomes provide little support for the headline claim. The contribution is thus best viewed as an exploratory design study whose causal language exceeds what the evidence can support.

major comments (3)
  1. [§7.2.2] The baseline condition (GPT-4 & Canvas) differs from Polymind in two independent ways: turn-taking conversational interaction versus parallel, proactive multi-agent operation, and text-prompt-based LLM use versus diagram-mediated microtask delegation. Because there is no condition that holds the interface and modality constant while varying only parallelism (e.g., a serial diagram-microtask condition), the study cannot support the causal claim that parallelism, rather than the diagram-based microtask interface, drives the reported benefits. This attribution is central to the abstract, §1, and §8.3, so the experimental design leaves a load-bearing gap.
  2. [§7.3.2] The quantitative results do not support the headline claim that Polymind enables users to 'quickly expand personalised writing ideas.' Median idea counts were equal (3 vs. 3), the CSI expressiveness p-value is exactly 0.05 with n=10 (which is not a robust significance), and the TTCW expert scores favored the baseline (4 vs. 5.7, p=0.23). The paper's positive conclusions about customizability and agency are carried almost entirely by qualitative self-reports, which are particularly susceptible to novelty effects in a within-subjects comparison with a research prototype. The authors should either temper the causal language or provide additional quantitative or behavioral evidence.
  3. [§6.4] The implementation section contains unresolved placeholders: 'based on the saved dialogue (see ??)' and 'We refer our readers to ?? for more details.' The missing content includes the exact mechanism for explanation/regeneration prompts and the precise output constraints, which are needed to assess the reproducibility of the system and the basis for the customizability claim. These details should be supplied or the references should be corrected.
minor comments (5)
  1. [§7.3.1] There are typos in this section: 'mannually' should be 'manually' and 'intiative' should be 'initiative'.
  2. [§8.4] The word 'differnt' should be 'different' in the sentence about structural levels.
  3. [§8.1] The discussion refers to participants as 'S6', 'S8', and 'S10', but participants are consistently labeled V1-10 elsewhere in the paper; please unify the notation.
  4. [§8.5 vs §7.2.3] The limitations section states the evaluation was a 'controlled session of around 1 hour,' but the procedure section reports the whole study lasted 'around 2 hours.' Please clarify which duration is intended.
  5. [Table 1] The table header and formatting are a bit cramped; consider splitting the prompt column into separate example prompts for readability.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: Polymind's central claims rest on a user evaluation and external design rationale; the baseline confound is a validity limitation, not a circular reduction.

full rationale

Polymind's central claims are empirical and system-design claims rather than derived predictions. The system's components (microtasks, task board, initiative modes) are motivated by an external literature review, a formative study with 10 participants, and prior human-collaboration results such as GroupMind, not by the evaluation outcomes. No parameter is fitted to the study data and then reported as a prediction; the evaluation measures usability, CSI, TTCW, and qualitative perceptions of agency and control. Self-citation of the authors' earlier prewriting study [84] supplies background on human-AI co-creativity stages and is not the evidence for the paper's evaluation claims. The only substantial weakness is experimental: the baseline (ChatGPT-4 plus Polymind's canvas with all microtasking features turned off) varies parallelism and interaction modality simultaneously, so the causal attribution to parallelism is underdetermined. That is a validity limitation rather than a circular derivation, and it does not make the claims equivalent to their inputs by construction. No circular step can be exhibited by quoting equations, fitted parameters, or definitions that reduce to the conclusions.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The central claims rest more on design choices and a small empirical study than on derived mathematics. Hand-set system parameters and several domain assumptions about attention, task coverage, and baseline attribution are the main ledger entries; no invented entities are introduced.

free parameters (3)
  • Proactive sampling interval x = 5 seconds
    Hand-set after pilot study to balance notification frequency and latency; it directly controls how often proactive microtasks fire and therefore shapes the evaluated experience.
  • LLM temperature = 0.7
    Implementation choice in §6.4 to moderate generation randomness; not fitted to outcome data.
  • Generation output constraints = 3 keywords (<=3 words), 3 concepts (<=5 words), 1 sticky note (<=150 words)
    Hand-set to keep results digestible on the canvas; affects perceived relevance and brevity reported in the study.
assumptions (5)
  • domain assumption Fitts' law based node sampling approximates user attention.
    Used in §6.2.1 to select nodes for proactive microtasks; if it fails, proactive generation targets irrelevant nodes and the reported 'relevant, fast' expansions may not hold.
  • domain assumption Microtasks are sufficiently independent to run concurrently without harmful interference.
    The design rationale in §4 assumes parallel microtasks need little context from one another; the study does not measure interference directly.
  • domain assumption The six default microtasks cover the needed prewriting strategies.
    Derived from a literature survey in Goal 1 (§4, Table 1); the authors acknowledge in §8.5 that real-life prewriting may need more.
  • domain assumption The ChatGPT-plus-canvas baseline isolates the effect of the parallel workflow.
    §7.2.2 baseline differs in modality and serial/parallel structure, so differences cannot be cleanly attributed to parallelism.
  • domain assumption Results from 10 participants in 12-minute sessions generalize to real prewriting.
    Acknowledged in §8.5 as a limitation of the controlled ~1-hour session and convenience sample.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Polymind: Parallel Visual Diagramming with Large Language Models to Support Prewriting Through Microtasks." pith.science (2026). https://pith.science/paper/A6BTGR27

@misc{pith2026250209577,
  author       = {Pith},
  title        = {Pith review of: Polymind: Parallel Visual Diagramming with Large Language Models to Support Prewriting Through Microtasks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/A6BTGR27}},
  note         = {Machine review of arXiv:2502.09577}
}
read the original abstract

Prewriting is the process of generating and organising ideas before a first draft. It consists of a combination of informal, iterative, and semi-structured strategies such as visual diagramming, which poses a challenge for collaborating with large language models (LLMs) in a turn-taking conversational manner. We present Polymind, a visual diagramming tool that leverages multiple LLM-powered agents to support prewriting. The system features a parallel collaboration workflow in place of the turn-taking conversational interactions. It defines multiple ``microtasks'' to simulate group collaboration scenarios such as collaborative writing and group brainstorming. Instead of repetitively prompting a chatbot for various purposes, Polymind enables users to orchestrate multiple microtasks simultaneously. Users can configure and delegate customised microtasks, and manage their microtasks by specifying task requirements and toggling visibility and initiative. Our evaluation revealed that, compared to ChatGPT, users had more customizability over collaboration with Polymind, and were thus able to quickly expand personalised writing ideas during prewriting.

Figures

Figures reproduced from arXiv: 2502.09577 by the authors.

Figure 1
Figure 1. The microtasking workflow of Polymind: A) A user can delegate new microtasks, and configure microtasks by specifying input & output types, prompts, initiative modes, etc. B) An active microtask notifies users when results are ready and provides previews on the canvas. C) Once expanded, resulting diagrams are displayed using a hollow shape in contrast to user created diagrams, and distinctive border colours indicatin… view at source ↗
Figure 2
Figure 2. The interface of Polymind comprises: A. a diagramming canvas B. a toolbar C. a task board 6 Designing Polymind Polymind’s interface provides a range of diagramming features commonly used in prewriting strategies, and a “task board” overlaid on the canvas for microtask management, as shown in [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. Polymind supports three basic diagrams (or nodes), and allows users to draw a section over diagrams. The task board interface supports microtask management, and maps microtask input and output to different diagram types on the canvas. The results of microtasks are displayed as notifications and previews on the task header before being expanded and accepted. 6.1 Main Interface The interface of Polymind comprises a di… view at source ↗
Figures from the paper (8 more)
Figure 5
Figure 5. Figure 5: Once the mouse hovers over key points of a microtask on the preview panel for 1.5 seconds, the system will present a summary of the generated results using a news ticker effect. Results of Microtasks. To support Goal 2.2 we display all resulting diagram nodes using a h…
Figure 6
Figure 6. Figure 6: The task card of a microtask. The text label of the task name is indicated with a distinctive colour. (a) [PITH_FULL_IMAGE:figures/full_fig_p013_6.png]
Figure 7
Figure 7. Figure 7: The processing of a proactive microtask. Once results are obtained from the LLM, it will first “draw a [PITH_FULL_IMAGE:figures/full_fig_p015_7.png]
Figure 8
Figure 8. Figure 8: Different status of a microtask on a particular node. [PITH_FULL_IMAGE:figures/full_fig_p015_8.png]
Figure 9
Figure 9. Figure 9: The workflow of delegating a new microtask. (a) The user first clicks on the “add” icon and a new card [PITH_FULL_IMAGE:figures/full_fig_p016_9.png]
Figure 10
Figure 10. Figure 10: NASA Task Load Index of Polymind and Baseline conditions (the lower, the better). managing the canvas, adjusting its layout, and progressing through diagrams. Notably, V1 & V2 said that Polymind’s interface was easier to navigate, and easier to read, because its gener…
Figure 11
Figure 11. Figure 11: The perceived usefulness of Polymind features 7.3.2 Creativity Support. In terms of creativity, participants generally felt Polymind was more supportive (see [PITH_FULL_IMAGE:figures/full_fig_p019_11.png]
Figure 12
Figure 12. Figure 12: The results of Creativity Support Index (CSI) and Torrance Test of Creative Writing (TTCW) [PITH_FULL_IMAGE:figures/full_fig_p020_12.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Exploring Agentic Workflows for Generating High Quality Math Visual Aids

    cs.AI 2026-07 conditional novelty 4.0 of 10

    An exploratory agentic self-improvement loop for TikZ math diagrams yields modest human-rated gains but fails to fix many spatial and coverage errors.

Reference graph

Works this paper leans on

95 extracted references · 72 canonical work pages · cited by 1 Pith paper

  1. [1]

    M Al-Khataybeh and NS Al-Tarawneh. 2015. The effect of using the six thinking hats method on the development of EFL female eleventh grade students’ writing skill in Southern Al-Mazar directorate of education. International Journal of Arts and Humanities 1, 4 (2015), 24–37

    2015

  2. [2]

    Majid Mohammad Al-Khataybeh. 2018. The Effect of Using the ‘Six Thinking Hats’ and Fishbone Strategies for Developing Saudi EFL Learners’ Writing Competence. Asian EFL Journal Research Articles 1 (2018), 27

    2018

  3. [3]

    Ian Arawjo, Chelse Swoopes, Priyan Vaithilingam, Martin Wattenberg, and Elena L Glassman. 2024. ChainForge: A Visual Toolkit for Prompt Engineering and LLM Hypothesis Testing. In Proceedings of the CHI Conference on Human Factors in Computing Systems . 1–18

    2024

  4. [4]

    Ian Arawjo, Priyan Vaithilingam, Martin Wattenberg, and Elena Glassman. 2023. ChainForge: An open-source visual programming environment for prompt engineering. In Adjunct Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology . 1–3

    2023

  5. [5]

    Marva A Barnett. 1989. Writing as a process. The French Review 63, 1 (1989), 31–44

    1989

  6. [6]

    Ismail Baroudy. 2008. A Procedural Approach to Process Theory of Writing: Prewriting Techniques. The International Journal of Language Society and Culture 24, 4 (2008), 45–52

    2008

  7. [7]

    Karim Benharrak, Tim Zindulka, Florian Lehmann, Hendrik Heuer, and Daniel Buschek. 2024. Writer-Defined AI Personas for On-Demand Feedback Generation. In Proceedings of the CHI Conference on Human Factors in Computing Systems (Honolulu, HI, USA) (CHI ’24). Association for Computing Machinery, New York, NY, USA, Article 1049, 18 pages. doi:10.1145/3613904.3642406

    work page arXiv 2024

  8. [8]

    Michael S Bernstein, Greg Little, Robert C Miller, Björn Hartmann, Mark S Ackerman, David R Karger, David Crowell, and Katrina Panovich. 2010. Soylent: a word processor with a crowd inside. In Proceedings of the 23nd annual ACM symposium on User interface software and technology . 313–322

    2010

Show all 95 references
  1. [9]

    Advait Bhat, Saaket Agashe, Parth Oberoi, Niharika Mohile, Ravi Jangir, and Anirudha Joshi. 2023. Interacting with Next-Phrase Suggestions: How Suggestion Systems Aid and Influence the Cognitive Processes of Writing (IUI ’23). 436–452. doi:10.1145/3581641.3584060

    2023

  2. [10]

    Jeremy Birnholtz and Steven Ibara. 2012. Tracking changes in collaborative writing: edits, visibility and group maintenance. In Proceedings of the ACM 2012 conference on Computer Supported Cooperative Work . 809–818

    2012

  3. [11]

    Jeremy Birnholtz, Stephanie Steinhardt, and Antonella Pavese. 2013. Write here, write now! An experimental study of group maintenance in collaborative writing. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems. 961–970

    2013

  4. [12]

    E de Bono. 1985. Six thinking hats: An essential approach to business management. Proc. ACM Hum.-Comput. Interact., Vol. 9, No. 7, Article 316. Publication date: November 2025. 316:26 Qian Wan, Jiannan Li, Huanchen Wang, and Zhicong Lu

    1985

  5. [13]

    Daniel Buschek, Benjamin Bisinger, and Florian Alt. 2018. ResearchIME: A mobile keyboard application for studying free typing behaviour in the wild. In Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems . 1–14

    2018

  6. [14]

    Daniel Buschek, Martin Zürn, and Malin Eiband. 2021. The impact of multiple parallel phrase suggestions on email input and composition behaviour of native and non-native english writers. In Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems . 1–13

    2021

  7. [15]

    Tuhin Chakrabarty, Philippe Laban, Divyansh Agarwal, Smaranda Muresan, and Chien-Sheng Wu. 2023. Art or artifice? large language models and the false promise of creativity. arXiv preprint arXiv:2309.14556 (2023)

    2023 arXiv

  8. [16]

    Joel Chan, Steven Dang, and Steven P Dow. 2016. Improving crowd innovation with expert facilitation. In Proceedings of the 19th ACM Conference on Computer-Supported Cooperative Work & Social Computing . 1223–1235

    2016

  9. [17]

    Chen Chen, Xiaojun Meng, Shengdong Zhao, and Morten Fjeld. 2017. ReTool: Interactive microtask and workflow design through demonstration. In Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems . 3551–3556

    2017

  10. [18]

    Justin Cheng, Jaime Teevan, Shamsi T Iqbal, and Michael S Bernstein. 2015. Break it down: A comparison of macro-and microtasks. In Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems . 4061–4064

    2015

  11. [19]

    Erin Cherry and Celine Latulipe. 2014. Quantifying the creativity support of digital tools through the creativity support index. ACM Transactions on Computer-Human Interaction (TOCHI) 21, 4 (2014), 1–25

    2014

  12. [20]

    Lydia B Chilton, Savvas Petridis, and Maneesh Agrawala. 2019. VisiBlends: A flexible workflow for visual blends. In Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems . 1–14

    2019

  13. [21]

    John Joon Young Chung, Wooseok Kim, Kang Min Yoo, Hwaran Lee, Eytan Adar, and Minsuk Chang. 2022. TaleBrush: sketching stories with generative pretrained language models. In Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems . 1–19

    2022

  14. [22]

    John Joon Young Chung, Wooseok Kim, Kang Min Yoo, Hwaran Lee, Eytan Adar, and Minsuk Chang. 2022. TaleBrush: visual sketching of story generation with pretrained language models. InCHI Conference on Human Factors in Computing Systems Extended Abstracts. 1–4

    2022

  15. [23]

    Elizabeth Clark, Anne Spencer Ross, Chenhao Tan, Yangfeng Ji, and Noah A Smith. 2018. Creative writing with a machine in the loop: Case studies on slogans and stories. In 23rd International Conference on Intelligent User Interfaces . 329–340

    2018

  16. [24]

    Juliet Corbin and Anselm Strauss. 2014. Basics of qualitative research: Techniques and procedures for developing grounded theory. Sage publications

    2014

  17. [25]

    Wenzhe Cui, Suwen Zhu, Mingrui Ray Zhang, H Andrew Schwartz, Jacob O Wobbrock, and Xiaojun Bi. 2020. Justcorrect: Intelligent post hoc text correction techniques on smartphones. In Proceedings of the 33rd Annual ACM Symposium on User Interface Software and Technology . 487–499

    2020

  18. [26]

    Hai Dang, Karim Benharrak, Florian Lehmann, and Daniel Buschek. 2022. Beyond Text Generation: Supporting Writers with Continuous Automatic Text Summaries. In Proceedings of the 35th Annual ACM Symposium on User Interface Software and Technology. 1–13

    2022

  19. [27]

    Martin Davies. 2011. Concept mapping, mind mapping and argument mapping: what are the differences and do they matter? Higher education 62 (2011), 279–301

    2011

  20. [28]

    Edward De Bono. 2017. Six Thinking Hats: The multi-million bestselling guide to running better meetings and making faster decisions. Penguin uk

    2017

  21. [29]

    Claudio Dell’Era, Stefano Magistretti, Cabirio Cautela, Roberto Verganti, and Francesco Zurlo. 2020. Four kinds of design thinking: From ideating to making, engaging, and criticizing. Creativity and Innovation Management 29, 2 (2020), 324–344

    2020

  22. [30]

    Steven Dow, Anand Kulkarni, Brie Bunge, Truc Nguyen, Scott Klemmer, and Björn Hartmann. 2011. Shepherding the crowd: managing and providing feedback to crowd workers. In CHI’11 Extended Abstracts on Human Factors in Computing Systems. 1669–1674

    2011

  23. [31]

    Steven P Dow, Alana Glassco, Jonathan Kass, Melissa Schwarz, Daniel L Schwartz, and Scott R Klemmer. 2010. Parallel prototyping leads to better design results, more divergence, and increased self-efficacy. ACM Transactions on Computer-Human Interaction (TOCHI) 17, 4 (2010), 1–24

    2010

  24. [32]

    Haakon Faste and Honray Lin. 2012. The untapped promise of digital mind maps. In Proceedings of the SIGCHI conference on human factors in computing systems . 1017–1026

    2012

  25. [33]

    Paul M Fitts. 1954. The information capacity of the human motor system in controlling the amplitude of movement. Journal of experimental psychology 47, 6 (1954), 381

    1954

  26. [34]

    Linda Flower and John R Hayes. 1981. A cognitive process theory of writing. College composition and communication 32, 4 (1981), 365–387

    1981

  27. [35]

    Jos Fransen, Paul A Kirschner, and Gijsbert Erkens. 2011. Mediating team effectiveness in the context of collaborative learning: The importance of team and task awareness. Computers in human Behavior 27, 3 (2011), 1103–1113. Proc. ACM Hum.-Comput. Interact., Vol. 9, No. 7, Art...

    2011

  28. [36]

    Katy Ilonka Gero and Lydia B Chilton. 2019. How a Stylistic, Machine-Generated Thesaurus Impacts a Writer’s Process. In Proceedings of the 2019 on Creativity and Cognition . 597–603

    2019

  29. [37]

    Katy Ilonka Gero and Lydia B Chilton. 2019. Metaphoria: An algorithmic companion for metaphor creation. In Proceedings of the 2019 CHI conference on human factors in computing systems . 1–12

    2019

  30. [38]

    Jennifer Gluck, Andrea Bunt, and Joanna McGrenere. 2007. Matching attentional draw with utility in interruption. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems . 41–50

    2007

  31. [39]

    Sue Gregory and Yvonne Masters. 2012. Real thinking with virtual hats: A role-playing activity for pre-service teachers in Second Life. Australasian Journal of Educational Technology 28, 3 (2012)

    2012

  32. [40]

    Carl Gutwin and Saul Greenberg. 2002. A descriptive framework of workspace awareness for real-time groupware. Computer Supported Cooperative Work (CSCW) 11 (2002), 411–446

    2002

  33. [41]

    Sandra G Hart. 1986. NASA task load index (TLX). (1986)

    1986

  34. [42]

    Saskia Haug and Alexander Maedche. 2021. Feeasy: An Interactive Crowd-Feedback System. In Adjunct Proceedings of the 34th Annual ACM Symposium on User Interface Software and Technology . 41–43

    2021

  35. [43]

    Tom Hope, Ronen Tamari, Daniel Hershcovich, Hyeonsu B Kang, Joel Chan, Aniket Kittur, and Dafna Shahaf. 2022. Scaling Creative Inspiration with Fine-Grained Functional Aspects of Ideas. In Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems . 1–15

    2022

  36. [44]

    Eric Horvitz. 1999. Principles of mixed-initiative user interfaces. In Proceedings of the SIGCHI conference on Human Factors in Computing Systems . 159–166

    1999

  37. [45]

    Chieh-Yang Huang, Shih-Hong Huang, and Ting-Hao Kenneth Huang. 2020. Heteroglossia: In-situ story ideation with the crowd. In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems . 1–12

    2020

  38. [46]

    Yi-Ching Huang, Hao-Chuan Wang, and Jane Yung-jen Hsu. 2018. Feedback Orchestration: Structuring Feedback for Facilitating Reflection and Revision in Writing. In Companion of the 2018 ACM Conference on Computer Supported Cooperative Work and Social Computing . 257–260

    2018

  39. [47]

    Charles McLaughlin Hymes and Gary M Olson. 1992. Unblocking brainstorming through the use of a simple group editor. In Proceedings of the 1992 ACM conference on Computer-supported cooperative work . 99–106

    1992

  40. [48]

    Shamsi T Iqbal, Jaime Teevan, Dan Liebling, and Anne Loomis Thompson. 2018. Multitasking with play write, a mobile microproductivity writing tool. In Proceedings of the 31st Annual ACM Symposium on User Interface Software and Technology. 411–422

    2018

  41. [49]

    Maurice Jakesch, Advait Bhat, Daniel Buschek, Lior Zalmanson, and Mor Naaman. 2023. Co-Writing with Opinionated Language Models Affects Users’ Views. arXiv preprint arXiv:2302.00560 (2023)

    2023 arXiv

  42. [50]

    Youngseung Jeon, Seungwan Jin, Patrick C Shih, and Kyungsik Han. 2021. FashionQ: an ai-driven creativity support tool for facilitating ideation in fashion design. InProceedings of the 2021 CHI Conference on Human Factors in Computing Systems. 1–18

    2021

  43. [51]

    Peiling Jiang, Jude Rayan, Steven P Dow, and Haijun Xia. 2023. Graphologue: Exploring Large Language Model Responses with Interactive Diagrams. arXiv preprint arXiv:2305.11473 (2023)

    2023 arXiv

  44. [52]

    Levent Burak Kara and Thomas F Stahovich. 2004. Hierarchical parsing and recognition of hand-sketched diagrams. In Proceedings of the 17th annual ACM symposium on User interface software and technology . 13–22

    2004

  45. [53]

    Ronald T Kellogg. 1990. Effectiveness of prewriting strategies as a function of task demands. The American Journal of Psychology (1990), 327–342

    1990

  46. [54]

    Nicolas Kokkalis, Thomas Köhn, Johannes Huebner, Moontae Lee, Florian Schulze, and Scott R Klemmer. 2013. Taskgenies: Automatically providing action plans helps people complete tasks. ACM Transactions on Computer-Human Interaction (TOCHI) 20, 5 (2013), 1–25

    2013

  47. [55]

    David Kurlander and Steven Feiner. 1992. A history-based macro by example system. In Proceedings of the 5th annual ACM symposium on User interface software and technology . 99–106

    1992

  48. [56]

    Thomas D LaToza, W Ben Towne, Christian M Adriano, and André Van Der Hoek. 2014. Microtask programming: Building software with a crowd. In Proceedings of the 27th annual ACM symposium on User interface software and technology. 43–54

    2014

  49. [57]

    Claudia Leacock, Martin Chodorow, Michael Gamon, and Joel Tetreault. 2010. Automated grammatical error detection for language learners. Synthesis lectures on human language technologies 3, 1 (2010), 1–134

    2010

  50. [58]

    Mina Lee, Percy Liang, and Qian Yang. 2022. Coauthor: Designing a human-ai collaborative writing dataset for exploring language model capabilities. In Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems. 1–19

    2022

  51. [59]

    Jingyi Li, Joel Brandt, Radomír Mech, Maneesh Agrawala, and Jennifer Jacobs. 2020. Supporting visual artists in programming through direct inspection and control of program execution. In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems . 1–12

    2020

  52. [60]

    Chung Kwan Lo and Khe Foon Hew. 2017. A critical review of flipped classroom challenges in K-12 education: Possible solutions and recommendations for future research. Research and practice in technology enhanced learning 12, 1 (2017), Proc. ACM Hum.-Comput. Interact., Vol. 9, ...

    2017

  53. [61]

    Barbara Lorenz, Tim Green, and Abbie Brown. 2009. Using multimedia graphic organizer software in the prewriting activities of primary school students: What are the benefits? Computers in the Schools 26, 2 (2009), 115–129

    2009

  54. [62]

    Zhicong Lu, Mingming Fan, Yun Wang, Jian Zhao, Michelle Annett, and Daniel Wigdor. 2018. Inkplanner: Supporting prewriting via intelligent visual diagramming. IEEE transactions on visualization and computer graphics 25, 1 (2018), 277–287

    2018

  55. [63]

    Mogahed M Mogahed. 2013. Planning out pre-writing activities. International Journal of English and Literature 4, 3 (2013), 60–68

    2013

  56. [64]

    Changhoon Oh, Jungwoo Song, Jinhan Choi, Seonghyeon Kim, Sungwoo Lee, and Bongwon Suh. 2018. I lead, you help but only with enough details: Understanding user experience of co-creation with artificial intelligence. In Proceedings of the 2018 CHI Conference on Human Factors in ...

    2018

  57. [65]

    OpenAI. 2023. GPT-4 Technical Report. arXiv:2303.08774 [cs.CL]

    2023 arXiv

  58. [66]

    Stephen T O’Rourke and Rafael A Calvo. 2009. Visualizing paragraph closeness for academic writing support. In 2009 Ninth IEEE International Conference on Advanced Learning Technologies . IEEE, 688–692

    2009

  59. [67]

    Stephen T O’Rourke, Rafael A Calvo, and Danielle S McNamara. 2011. Visualizing Topic Flow in Students’ Essays. Journal of Educational Technology & Society 14, 3 (2011)

    2011

  60. [68]

    Tim O Peterson and Dale A Lunsford. 1998. Parallel thinking: A technique for group interaction and problem solving. Journal of Management Education 22, 4 (1998), 537–554

    1998

  61. [69]

    Alec Radford, Karthik Narasimhan, Tim Salimans, Ilya Sutskever, et al. 2018. Improving language understanding by generative pre-training. (2018)

    2018

  62. [70]

    D Gordon Rohman. 1965. Pre-writing the stage of discovery in the writing process. College composition and communi- cation 16, 2 (1965), 106–112

    1965

  63. [71]

    Mark A Runco. 2023. Creativity: Research, development, and practice . Academic Press

    2023

  64. [72]

    Thomas P Ryan and JP Morgan. 2007. Modern experimental design. Journal of Statistical Theory and Practice 1, 3-4 (2007), 501–506

    2007

  65. [73]

    John Sadauskas, Daragh Byrne, and Robert K Atkinson. 2015. Mining memories: Designing a platform to support social media based writing. In Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems . 3691–3700

    2015

  66. [74]

    Orit Shaer, Angelora Cooper, Osnat Mokryn, Andrew L Kun, and Hagit Ben Shoshan. 2024. AI-Augmented Brainwriting: Investigating the use of LLMs in group ideation. In Proceedings of the CHI Conference on Human Factors in Computing Systems. 1–17

    2024

  67. [75]

    Yang Shi, Nan Cao, Xiaojuan Ma, Siji Chen, and Pei Liu. 2020. Emog: Supporting the sketching of emotional expressions for storyboarding. In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems . 1–12

    2020

  68. [76]

    Yang Shi, Yang Wang, Ye Qi, John Chen, Xiaoyao Xu, and Kwan-Liu Ma. 2017. IdeaWall: Improving creative collaboration through combinatorial visual stimuli. In Proceedings of the 2017 ACM Conference on Computer Supported Cooperative Work and Social Computing. 594–603

    2017

  69. [77]

    Patrick C Shih, David H Nguyen, Sen H Hirano, David F Redmiles, and Gillian R Hayes. 2009. GroupMind: supporting idea generation through a collaborative mind-mapping tool. In Proceedings of the 2009 ACM International Conference on Supporting Group Work. 139–148

    2009

  70. [78]

    Nikhil Singh, Guillermo Bernal, Daria Savchenko, and Elena L Glassman. 2022. Where to hide a stolen elephant: Leaps in creative writing with multimodal machine intelligence. ACM Transactions on Computer-Human Interaction (2022)

    2022

  71. [79]

    Jessie J Smith, Saleema Amershi, Solon Barocas, Hanna Wallach, and Jennifer Wortman Vaughan. 2022. Real ml: Recognizing, exploring, and articulating limitations of machine learning research. In 2022 ACM Conference on Fairness, Accountability, and Transparency. 587–597

    2022

  72. [80]

    Sangho Suh, Bryan Min, Srishti Palani, and Haijun Xia. 2023. Sensecape: Enabling multilevel exploration and sensemaking with large language models. In Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology. 1–18

    2023

  73. [81]

    Jaime Teevan, Shamsi T Iqbal, and Curtis Von Veh. 2016. Supporting collaborative writing with microtasks. In Proceedings of the 2016 CHI conference on human factors in computing systems . 2657–2668

    2016

  74. [82]

    Masaki Uto and Maomi Ueno. 2015. Academic writing support system using bayesian networks. In 2015 IEEE 15th International Conference on Advanced Learning Technologies . IEEE, 385–387

    2015

  75. [83]

    Jorge Villalón, Paul Kearney, Rafael A Calvo, and Peter Reimann. 2008. Glosser: Enhanced feedback for student writing tasks. In 2008 eighth IEEE international conference on advanced learning technologies . IEEE, 454–458

    2008

  76. [84]

    It Felt Like Having a Second Mind

    Qian Wan, Siying Hu, Yu Zhang, Piaohong Wang, Bo Wen, and Zhicong Lu. 2024. " It Felt Like Having a Second Mind": Investigating Human-AI Co-creativity in Prewriting with Large Language Models. Proceedings of the ACM on Human-Computer Interaction 8, CSCW1 (2024), 1–26. Proc. AC...

    2024

  77. [85]

    Hao-Chuan Wang, Dan Cosley, and Susan R Fussell. 2010. Idea expander: supporting group brainstorming with conversationally triggered visual thinking stimuli. In Proceedings of the 2010 ACM conference on Computer supported cooperative work. 103–106

    2010

  78. [86]

    Hao-Chuan Wang, Susan R Fussell, and Dan Cosley. 2011. From diversity to creativity: Stimulating group brainstorming with cultural differences and conversationally-retrieved pictures. InProceedings of the ACM 2011 conference on Computer supported cooperative work. 265–274

    2011

  79. [87]

    Sitong Wang, Savvas Petridis, Taeahn Kwon, Xiaojuan Ma, and Lydia B Chilton. 2021. PopBlends: Strategies for Conceptual Blending with Large Language Models. arXiv preprint arXiv:2111.04920 (2021)

    2021 arXiv

  80. [88]

    Yunlong Wang, Priyadarshini Venkatesh, and Brian Y Lim. 2022. Interpretable Directed Diversity: Leveraging Model Explanations for Iterative Crowd Ideation. In Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems. 1–28

    2022

  81. [89]

    Di Wu, Zhiwang Yu, Nan Ma, Jianan Jiang, Yuetian Wang, Guixiang Zhou, Hanhui Deng, and Yi Li. 2023. StyleMe: Towards Intelligent Fashion Generation with Designer Style. In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems . 1–16

    2023

  82. [90]

    Tongshuang Wu, Ellen Jiang, Aaron Donsbach, Jeff Gray, Alejandra Molina, Michael Terry, and Carrie J Cai. 2022. Promptchainer: Chaining large language model prompts through visual programming. In CHI Conference on Human Factors in Computing Systems Extended Abstracts . 1–10

    2022

  83. [91]

    Tongshuang Wu, Michael Terry, and Carrie Jun Cai. 2022. Ai chains: Transparent and controllable human-ai interaction by chaining large language model prompts. In Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems. 1–22

    2022

  84. [92]

    Chuan Yan, John Joon Young Chung, Yoon Kiheon, Yotam Gingold, Eytan Adar, and Sungsoo Ray Hong. 2022. FlatMagic: Improving flat colorization through AI-driven design for digital comic professionals. In Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems . 1–17

    2022

  85. [93]

    Robert C Zeleznik, Andrew Bragdon, Chu-Chi Liu, and Andrew Forsberg. 2008. Lineogrammer: creating diagrams by drawing. In Proceedings of the 21st annual ACM symposium on User interface software and technology . 161–170

    2008

  86. [94]

    Mingrui Ray Zhang, He Wen, and Jacob O Wobbrock. 2019. Type, then correct: intelligent text correction techniques for mobile text entry using neural networks. In Proceedings of the 32nd Annual ACM Symposium on User Interface Software and Technology. 843–855

    2019

  87. [95]

    Zheng Zhang, Jie Gao, Ranjodh Singh Dhaliwal, and Toby Jia-Jun Li. 2023. VISAR: A Human-AI Argumentative Writing Assistant with Visual Programming and Rapid Draft Prototyping. arXiv preprint arXiv:2304.07810 (2023). Received July 2024; revised December 2024; accepted March 202...

    2023 arXiv

Pith tools

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