pith. machine review for the scientific record. sign in

arxiv: 2511.23379 · v2 · submitted 2025-11-28 · 💻 cs.HC

TaskLens: Generating Task-Conditioned Scaffolded Interfaces for Learning Professional Creative Software

Yimeng Liu , Misha Sra This is my paper

Pith reviewed 2026-05-17 04:24 UTC · model grok-4.3

classification 💻 cs.HC
keywords scaffolded interfacesLLM-generated UIscreative software learningBlender 3D modelingworkflow guidancetask load reductiondomain concept learninguser studies in HCI
0
0 comments X

The pith

TaskLens uses LLMs to automatically generate scaffolded interfaces that reduce task load and improve learning for beginners using complex creative software like Blender.

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

The paper presents TaskLens as a method that takes a natural language description of a creative task and uses large language models to build a simplified, task-specific interface for software such as Blender. The generated interface breaks the task into workflow stages, shows only the relevant tools at each stage, connects those tools to key domain concepts, and gradually introduces more advanced features. A study with 32 beginners found these interfaces lowered perceived workload, helped users finish tasks more successfully, and increased their grasp of important concepts during the work itself. A follow-up study with eight experts suggested gains in efficiency and the ability to produce tailored interfaces for better productivity.

Core claim

TaskLens is an LLM-based method that automatically generates task-conditioned scaffolded UIs from natural language task descriptions. It works by having the LLM identify workflow stages and domain concepts, select task-relevant tools, generate the implementation code for the interface, and execute that code to display the result. The resulting interfaces surface relevant tools, organize them by workflow stage, link them to domain concepts, and progressively disclose advanced features. Evaluations in Blender with beginners showed significant reductions in perceived task load, improved performance via embedded guidance, and increased domain concept learning.

What carries the argument

TaskLens, the LLM pipeline that identifies workflow stages from a task description, selects appropriate tools, generates interface code, and runs the code to produce scaffolded UIs organized by stage and linked to concepts.

If this is right

  • Educators gain a way to produce guided interfaces for specific tasks in creative software without manual coding.
  • Beginners complete modeling tasks with lower workload and better results when tools are staged and linked to concepts.
  • Domain concept learning rises when interfaces connect tools directly to the ideas they represent.
  • Expert users can receive on-demand personalized interfaces that support higher efficiency in professional work.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same generation approach might apply to other steep-learning-curve tools such as video editors or CAD programs if the LLM handles their workflows accurately.
  • Real-time task detection could let interfaces adapt on the fly as a user shifts between subtasks inside the software.
  • Repeated exposure to scaffolded versions may help users build mental models that transfer when they later use the full, unscaffolded interface.

Load-bearing premise

Large language models can correctly identify workflow stages, pick the right tools, and produce error-free interface code for any creative task without adding misleading steps or leaving out essential ones.

What would settle it

If users of a newly generated TaskLens interface for a fresh Blender task show higher error rates, longer completion times, or no gains in concept understanding compared with the standard interface, the claim that these scaffolded UIs reliably help learning would be challenged.

Figures

Figures reproduced from arXiv: 2511.23379 by Misha Sra, Yimeng Liu.

Figure 1
Figure 1. Figure 1: With AugGen, users can generate scaffolded user interfaces (UIs) to augment software and domain concept learning while executing their tasks in professional creative software. Taking a task description as input, AugGen automatically analyzes user task to decompose the task into workflow stages and extract relevant tools and domain concepts. The functionality of selected tools are implemented through progra… view at source ↗
Figure 2
Figure 2. Figure 2: Challenges in professional creative software (C1– [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: AI copilots in Blender that offer step-by-step task instructions (left) [ [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Technical implementation pipeline built upon the [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: UV unwrapping using the default UI of Blender. Tools are located sparsely and sometimes hidden in keyboard [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: UV unwrapping Basic level of the scaffolded UI. While working on the task, the user can reveal relevant domain concepts by hovering over each tool to see the tooltips. For instance, “Mark Seam: Where to ‘cut’ the 3D model’s surface so it can be unfolded into a flat 2D layout”, “Unwrap: Flattens the 3D model’s surface into 2D space based on the marked seams”, and “UV Islands: Makes UV islands (separate unwr… view at source ↗
Figure 7
Figure 7. Figure 7: UV unwrapping Intermediate level of the scaf￾folded UI. with relevant tools impact user-perceived task load, task perfor￾mance, task completion time (efficiency), and user confidence. • RQ2: How does the scaffolded UI affect the learning of domain concepts and software operations? This question evaluates how progressive tool disclosure, concept-integrated UI organization, and bridging to native software he… view at source ↗
Figure 8
Figure 8. Figure 8: UV unwrapping using the scaffolded UI. Task-relevant tools are selected and grouped based on the workflow stages. Users can select complexity levels from basic, intermediate, to advanced for progressive tool disclosure. Hovering over each tool shows a tooltip explaining the related domain concepts. To connect the scaffolded interface with the native software environment, keyboard shortcuts, mouse clicks, a… view at source ↗
Figure 9
Figure 9. Figure 9: Study 1 participant responses using Ours (scaffolded) and Baseline (default Blender) interfaces for Tasks 1 and 2 (lower [PITH_FULL_IMAGE:figures/full_fig_p010_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Study 2 participant preferences for Ours (scaffolded) or Baseline (default Blender) interfaces regarding task perfor [PITH_FULL_IMAGE:figures/full_fig_p010_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Walk cycle animation using the scaffolded UI. [PITH_FULL_IMAGE:figures/full_fig_p016_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: a shows the scaffolded interface variations for UV unwrapping. For organic models like characters and creatures, the workflow focuses on creating smooth, low-distortion texture maps. The interface helps users define seams to flatten curved surfaces, apply algorithms that preserve surface flow, and refine UV layouts to keep textures accurate during animation. For hard-surface models, the workflow is design… view at source ↗
read the original abstract

Professional creative software has steep learning curves for novices due to complex interfaces, limited guidance, and unfamiliar terminology. To support educators and tool creators in addressing learner challenges, we introduce TaskLens, an LLM-based method that automatically generates task-conditioned scaffolded UIs from natural language task descriptions. Our method uses LLMs to identify workflow stages and domain concepts, select task-relevant tools, generate implementation code, and execute the code to produce scaffolded interfaces. The interfaces surface relevant tools, organize them by workflow stage, link them to domain concepts, and progressively disclose advanced features. We evaluate TaskLens by deploying two LLM-generated scaffolded interfaces in Blender, a professional 3D modeling software. A user study with beginners (n=32) showed that our scaffolded interfaces significantly reduced perceived task load, improved task performance through embedded workflow guidance, and increased domain concept learning in Blender during task execution. A second study with experts (n=8) showed improved task efficiency and potential to create personalized UIs for productivity and creativity.

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 / 2 minor

Summary. The paper introduces TaskLens, an LLM-based method for automatically generating task-conditioned scaffolded UIs from natural language task descriptions to support learning professional creative software such as Blender. LLMs are used to identify workflow stages and domain concepts, select relevant tools, generate implementation code, and produce interfaces that organize tools by workflow stage, link them to concepts, and progressively disclose advanced features. Evaluation consists of two user studies: a study with n=32 beginners claiming significant reductions in perceived task load, improved task performance via embedded guidance, and increased domain concept learning; and a study with n=8 experts claiming improved task efficiency and potential for personalized UIs.

Significance. If the empirical claims hold after proper validation and reporting, the work could meaningfully advance HCI by demonstrating a scalable LLM-driven approach to creating personalized learning scaffolds for complex creative tools, addressing steep learning curves in professional software. The core idea of task-conditioned interface generation via LLMs is a strength worth pursuing, provided the generated guidance is shown to be reliable.

major comments (2)
  1. [Evaluation / User Study 1] User study with beginners (n=32): The abstract claims that the scaffolded interfaces 'significantly reduced perceived task load, improved task performance... and increased domain concept learning,' yet no details are provided on control conditions, statistical tests, p-values, effect sizes, or validation of interface quality against LLM errors. This is load-bearing for the central performance and learning claims.
  2. [Method / Evaluation] Method and evaluation sections: The LLM-generated scaffolds for the two Blender tasks are deployed directly into the user studies without reported independent validation such as expert review, static analysis, or comparison against ground-truth workflows for workflow stage identification, tool selection, or concept linking. This assumption is load-bearing because any misidentification could mean the measured benefits reflect following flawed guidance rather than the intended scaffolding.
minor comments (2)
  1. [Abstract] The abstract would be strengthened by briefly specifying the exact Blender tasks used in the studies to contextualize the reported outcomes.
  2. [Discussion] Consider adding a limitations section discussing potential LLM hallucinations in workflow guidance and how they might affect learning outcomes.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed feedback. We appreciate the emphasis on strengthening the empirical claims and validation procedures. Below we respond point-by-point to the major comments and describe the revisions we will make.

read point-by-point responses

  1. Referee: [Evaluation / User Study 1] User study with beginners (n=32): The abstract claims that the scaffolded interfaces 'significantly reduced perceived task load, improved task performance... and increased domain concept learning,' yet no details are provided on control conditions, statistical tests, p-values, effect sizes, or validation of interface quality against LLM errors. This is load-bearing for the central performance and learning claims.

    Authors: We agree that the current reporting of the user study is insufficient to fully substantiate the claims. In the revised manuscript we will expand the Evaluation section (and update the abstract) to include: (1) a clear description of the control condition (standard Blender UI without TaskLens scaffolds), (2) the specific statistical tests performed (paired t-tests or repeated-measures ANOVA as appropriate), (3) exact p-values, (4) effect sizes (Cohen’s d or partial eta-squared), and (5) any post-hoc checks performed to assess interface quality or LLM-induced errors. These additions will make the quantitative results transparent and reproducible. revision: yes

  2. Referee: [Method / Evaluation] Method and evaluation sections: The LLM-generated scaffolds for the two Blender tasks are deployed directly into the user studies without reported independent validation such as expert review, static analysis, or comparison against ground-truth workflows for workflow stage identification, tool selection, or concept linking. This assumption is load-bearing because any misidentification could mean the measured benefits reflect following flawed guidance rather than the intended scaffolding.

    Authors: We acknowledge the importance of independent validation of the LLM outputs. We will add a dedicated subsection under Method that reports an expert validation study: two experienced Blender users independently reviewed the generated workflow stages, tool selections, and concept links for both tasks, comparing them against established Blender documentation and their own ground-truth workflows. Any discrepancies, resolution process, and inter-rater agreement will be reported. We will also add a limitations paragraph discussing residual risks of LLM errors. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical evaluation of LLM-generated scaffolds

full rationale

The paper describes an LLM-based pipeline for generating task-conditioned scaffolded UIs and evaluates the resulting interfaces via two user studies (n=32 beginners, n=8 experts) measuring task load, performance, and concept learning. No equations, fitted parameters, predictions, or derivation chains appear in the provided text. The central claims rest on direct empirical measurements rather than any self-referential reduction, self-citation load-bearing step, or ansatz smuggled through prior work. The work is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The approach assumes LLMs possess sufficient domain knowledge about Blender workflows and can produce correct, safe interface code without external verification. No free parameters or new entities are explicitly introduced in the abstract.

axioms (1)
  • domain assumption LLMs can accurately extract workflow stages and domain concepts from natural language task descriptions for professional creative software.
    Invoked in the method description as the basis for generating relevant scaffolds.

pith-pipeline@v0.9.0 · 5475 in / 1298 out tokens · 30032 ms · 2026-05-17T04:24:48.446693+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.

Reference graph

Works this paper leans on

95 extracted references · 95 canonical work pages · 2 internal anchors

  1. [1]

    2025.Adobe Express

    Adobe. 2025.Adobe Express. Adobe. https://www.adobe.com/express/

    work page 2025

  2. [2]

    2025.Adobe Firefly

    Adobe. 2025.Adobe Firefly. Adobe. https://www.adobe.com/products/firefly.h tml

    work page 2025

  3. [3]

    2025.Adobe Photoshop API

    Adobe. 2025.Adobe Photoshop API. Adobe. https://developer.adobe.com/photos hop/api/

    work page 2025

  4. [4]

    2025.Adobe Photoshop Generative Fill

    Adobe. 2025.Adobe Photoshop Generative Fill. Adobe. https://www.adobe.com/ products/photoshop/generative-fill.html

    work page 2025

  5. [5]

    2025.Illustrator on the iPad

    Adobe. 2025.Illustrator on the iPad. Draw on inspiration anywhere.Adobe. https://www.adobe.com/products/illustrator/ipad.html?msockid=36b836cad4c f6d4502aa20c7d5086c30

    work page 2025

  6. [6]

    1987.Knowledge Navigator

    Apple. 1987.Knowledge Navigator. Apple Inc. https://en.wikipedia.org/wiki/Kn owledge_Navigator

    work page 1987

  7. [7]

    2025.AutoCAD API

    Autodesk. 2025.AutoCAD API. Autodesk Inc. https://aps.autodesk.com/develop er/overview/autocad

    work page 2025

  8. [8]

    2025.Autodesk TinkerCAD

    Autodesk. 2025.Autodesk TinkerCAD. Autodesk. https://www.tinkercad.com/

    work page 2025

  9. [9]

    2025.Maya API

    Autodesk. 2025.Maya API. Autodesk Inc. https://aps.autodesk.com/developer /overview/maya

    work page 2025

  10. [10]

    2017.Instructional scaffolding in STEM education: Strategies and efficacy evidence

    Brian R Belland. 2017.Instructional scaffolding in STEM education: Strategies and efficacy evidence. Springer Nature

    work page 2017

  11. [11]

    2025.Blender

    Blender. 2025.Blender. Blender. https://www.blender.org/

    work page 2025

  12. [12]

    2025.Blender 3.6 Python API Documentation

    Blender. 2025.Blender 3.6 Python API Documentation. Blender. https://docs.ble nder.org/api/3.6/

    work page 2025

  13. [13]

    2025.Blender 3.6 Reference Manual

    Blender. 2025.Blender 3.6 Reference Manual. Blender. https://docs.blender.org /manual/en/3.6/index.html#

    work page 2025

  14. [14]

    2025.Blender Copilot

    Blender. 2025.Blender Copilot. Blender. https://blenderside.gumroad.com/l/blen der-copilot

    work page 2025

  15. [15]

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

    work page 2006

  16. [16]

    Saša Brdnik, Tjaša Heričko, and Boštjan Šumak. 2022. Intelligent user interfaces and their evaluation: a systematic mapping study.Sensors22, 15 (2022), 5830

    work page 2022

  17. [17]

    2025.Canva

    Canva. 2025.Canva. Canva. https://www.canva.com/

    work page 2025

  18. [18]

    John M Carroll and Caroline Carrithers. 1984. Training wheels in a user interface. Commun. ACM27, 8 (1984), 800–806

    work page 1984

  19. [19]

    Valerie Chen, Q Vera Liao, Jennifer Wortman Vaughan, and Gagan Bansal. 2023. Understanding the role of human intuition on reliance in human-AI decision- making with explanations.Proceedings of the ACM on Human-computer Interac- tion7, CSCW2 (2023), 1–32

    work page 2023

  20. [20]

    Parmit K Chilana, Nathaniel Hudson, Srinjita Bhaduri, Prashant Shashikumar, and Shaun Kane. 2018. Supporting remote real-time expert help: Opportuni- ties and challenges for novice 3d modelers. In2018 IEEE Symposium on Visual Languages and Human-Centric Computing (VL/HCC). IEEE, 157–166

    work page 2018

  21. [21]

    Parmit K Chilana, Amy J Ko, and Jacob O Wobbrock. 2012. LemonAid: selection- based crowdsourced contextual help for web applications. InProceedings of the SIGCHI Conference on Human Factors in Computing Systems. 1549–1558

    work page 2012

  22. [22]

    Andy Cockburn, Carl Gutwin, Joey Scarr, and Sylvain Malacria. 2014. Supporting Novice to Expert Transitions in User Interfaces.Comput. Surveys47 (01 2014), 1–36. doi:10.1145/2659796

    work page doi:10.1145/2659796 2014

  23. [23]

    2025.Windows 11’s Copilot Vision wants to help you learn to use complicated apps

    Andrew Cunningham. 2025.Windows 11’s Copilot Vision wants to help you learn to use complicated apps. ARS Technica. https://arstechnica.com/gadgets/2025/0 4/windows-11s-copilot-vision-wants-to-help-you-learn-to-use-complicated- apps/

    work page 2025

  24. [24]

    2025.Rhino Developer API Documentation

    Rhino Developer. 2025.Rhino Developer API Documentation. Rhino Developer. https://developer.rhino3d.com/

    work page 2025

  25. [25]

    Michael Ekstrand, Wei Li, Tovi Grossman, Justin Matejka, and George Fitzmau- rice. 2011. Searching for software learning resources using application context. InProceedings of the 24th annual ACM symposium on User interface software and technology. 195–204

    work page 2011

  26. [26]

    Mica R Endsley. 2017. From here to autonomy: lessons learned from human– automation research.Human factors59, 1 (2017), 5–27

    work page 2017

  27. [27]

    David K Farkas. 1993. The role of balloon help.ACM SIGDOC Asterisk Journal of Computer Documentation17, 2 (1993), 3–19

    work page 1993

  28. [28]

    Adam Fourney, Ben Lafreniere, Parmit Chilana, and Michael Terry. 2014. Inter- Twine: creating interapplication information scent to support coordinated use of software. InProceedings of the 27th annual ACM symposium on User interface software and technology. 429–438

    work page 2014

  29. [29]

    C Ailie Fraser, Tricia J Ngoon, Mira Dontcheva, and Scott Klemmer. 2019. Re- Play: contextually presenting learning videos across software applications. In Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems. 1–13

    work page 2019

  30. [30]

    2025.Gemini 3

    Google. 2025.Gemini 3. Google. https://deepmind.google/models/gemini/pro/

    work page 2025

  31. [31]

    2025.Google AI Studio

    Google. 2025.Google AI Studio. Google. https://aistudio.google.com/

    work page 2025

  32. [32]

    Tovi Grossman and George Fitzmaurice. 2010. ToolClips: an investigation of contextual video assistance for functionality understanding. InProceedings of the SIGCHI Conference on Human Factors in Computing Systems. 1515–1524

    work page 2010

  33. [33]

    Tovi Grossman, Justin Matejka, and George Fitzmaurice. 2010. Chronicle: capture, exploration, and playback of document workflow histories. InProceedings of the 23nd annual ACM symposium on User interface software and technology. 143–152

    work page 2010

  34. [34]

    Sandra G Hart. 2006. NASA-task load index (NASA-TLX); 20 years later. In Proceedings of the human factors and ergonomics society annual meeting, Vol. 50. Sage publications Sage CA: Los Angeles, CA, 904–908

    work page 2006

  35. [35]

    1999.Modern Information Retrieval—User Interfaces and Visualiza- tion

    Marti Hearst. 1999.Modern Information Retrieval—User Interfaces and Visualiza- tion. Addison-Wesley-Longman Publishing Co. https://people.ischool.berkeley. edu/~hearst/irbook/10/

    work page 1999

  36. [36]

    Jeffrey Heer. 2019. Agency plus automation: Designing artificial intelligence into interactive systems.Proceedings of the National Academy of Sciences116, 6 (2019), 1844–1850

    work page 2019

  37. [37]

    Eric J Horvitz, John S Breese, David Heckerman, David Hovel, and Koos Rom- melse. 2013. The Lumiere project: Bayesian user modeling for inferring the goals and needs of software users.arXiv preprint arXiv:1301.7385(2013)

    work page internal anchor Pith review Pith/arXiv arXiv 2013

  38. [38]

    2025.Adobe Illustrator

    Adobe Illustrator. 2025.Adobe Illustrator. Adobe. https://www.adobe.com/prod ucts/illustrator.html

    work page 2025

  39. [39]

    Anthony Jameson. 2007. Adaptive interfaces and agents. InThe human-computer interaction handbook. CRC press, 459–484

    work page 2007

  40. [40]

    Caitlin Kelleher and Randy Pausch. 2005. Stencils-based tutorials: design and evaluation. InProceedings of the SIGCHI conference on Human factors in computing systems. 541–550

    work page 2005

  41. [41]

    2009.Motivational design for learning and performance: The ARCS model approach

    John M Keller. 2009.Motivational design for learning and performance: The ARCS model approach. Springer Science & Business Media

    work page 2009

  42. [42]

    2025.Instructional Materials Motivation Survey

    John M Keller. 2025.Instructional Materials Motivation Survey. UNIVERSITÄT DUISBURG-ESSEN. https://learninglab.uni-due.de/research-instrument/13887

    work page 2025

  43. [43]

    2025.Do It For Me vs

    Anjali Khurana, Xiaotian Su, April Yi Wang, and Parmit K Chilana. 2025.Do It For Me vs. Do It With Me: Investigating User Perceptions of Different Paradigms of Automation in Copilots for Feature-Rich Software. Association for Computing Machinery, New York, NY, USA. https://doi.org/10.1145/3706598.3713431

    work page doi:10.1145/3706598.3713431 2025

  44. [44]

    Anjali Khurana, Hariharan Subramonyam, and Parmit K Chilana. 2024. Why and when llm-based assistants can go wrong: Investigating the effectiveness of prompt-based interactions for software help-seeking. InProceedings of the 29th International Conference on Intelligent User Interfaces. 288–303

    work page 2024

  45. [45]

    Kimia Kiani, George Cui, Andrea Bunt, Joanna McGrenere, and Parmit K Chilana

    work page

  46. [46]

    One-Size-Fits-All

    Beyond" One-Size-Fits-All" Understanding the Diversity in How Software Newcomers Discover and Make Use of Help Resources. InProceedings of the 2019 CHI Conference on Human Factors in Computing Systems. 1–14

    work page 2019

  47. [47]

    Jurek Kirakowski. 1996. The software usability measurement inventory: back- ground and usage.Usability evaluation in industry(1996), 169–178

    work page 1996

  48. [48]

    Thomas Kosch, Jakob Karolus, Johannes Zagermann, Harald Reiterer, Albrecht Schmidt, and Paweł W Woźniak. 2023. A survey on measuring cognitive workload in human-computer interaction.Comput. Surveys55, 13s (2023), 1–39. AugGen: Augmenting Task-Based Learning in Professional Creative Software with LLM-Generated Scaffolded UIs

    work page 2023

  49. [49]

    Nataliya Kosmyna, Eugene Hauptmann, Ye Tong Yuan, Jessica Situ, Xian-Hao Liao, Ashly Vivian Beresnitzky, Iris Braunstein, and Pattie Maes. 2025. Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task.arXiv preprint arXiv:2506.08872(2025)

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  50. [50]

    2025.Windows is about to get its biggest intelligent upgrade thanks to Copilot

    Jacob Krol. 2025.Windows is about to get its biggest intelligent upgrade thanks to Copilot. Tech Radar. https://www.techradar.com/computing/artificial- intelligence/windows-is-about-to-get-its-biggest-intelligent-upgrade-thanks- to-copilot

    work page 2025

  51. [51]

    Benjamin Lafreniere, Andrea Bunt, and Michael Terry. 2014. Task-centric inter- faces for feature-rich software. InProceedings of the 26th Australian Computer- Human Interaction Conference on Designing Futures: The Future of Design. 49–58

    work page 2014

  52. [52]

    Benjamin Lafreniere, Tovi Grossman, and George Fitzmaurice. 2013. Commu- nity enhanced tutorials: improving tutorials with multiple demonstrations. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems. 1779–1788

    work page 2013

  53. [53]

    Ghang Lee, Charles M Eastman, Tarang Taunk, and Chun-Heng Ho. 2010. Us- ability principles and best practices for the user interface design of complex 3D architectural design and engineering tools.International journal of human- computer studies68, 1-2 (2010), 90–104

    work page 2010

  54. [54]

    Jie Li, Hancheng Cao, Laura Lin, Youyang Hou, Ruihao Zhu, and Abdallah El Ali

    work page

  55. [55]

    InProceedings of the 2024 CHI Conference on Human Factors in Computing Systems

    User experience design professionals’ perceptions of generative artificial intelligence. InProceedings of the 2024 CHI Conference on Human Factors in Computing Systems. 1–18

    work page 2024

  56. [56]

    Wenjie Li, Yuxiao Zhou, Shijian Luo, and Yenan Dong. 2022. Design factors to improve the consistency and sustainable user experience of responsive Interface Design.Sustainability14, 15 (2022), 9131

    work page 2022

  57. [57]

    Xiangyang Li and Qiang Ji. 2005. Active affective state detection and user assistance with dynamic Bayesian networks.IEEE transactions on systems, man, and cybernetics-part a: systems and humans35, 1 (2005), 93–105

    work page 2005

  58. [58]

    2012.Improving understanding and trust with intelligibility in context-aware applications

    Brian Y Lim. 2012.Improving understanding and trust with intelligibility in context-aware applications. Carnegie Mellon University

    work page 2012

  59. [59]

    Yimeng Liu and Misha Sra. 2024. DanceGen: Supporting Choreography Ideation and Prototyping with Generative AI. InProceedings of the 2024 ACM Designing Interactive Systems Conference. 920–938

    work page 2024

  60. [60]

    Yimeng Liu, Misha Sra, and Chang Xiao. 2025. CrowdGenUI: Aligning LLM-Based UI Generation with Crowdsourced User Preferences. arXiv:2411.03477 [cs.HC] https://arxiv.org/abs/2411.03477

    work page arXiv 2025

  61. [61]

    Justin Matejka, Tovi Grossman, and George Fitzmaurice. 2011. Ambient help. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems. 2751–2760

    work page 2011

  62. [62]

    Justin Matejka, Tovi Grossman, and George Fitzmaurice. 2011. IP-QAT: in- product questions, answers, & tips. InProceedings of the 24th annual ACM sym- posium on User interface software and technology. 175–184

    work page 2011

  63. [63]

    Joanna McGrenere, Ronald M Baecker, and Kellogg S Booth. 2002. An evaluation of a multiple interface design solution for bloated software. InProceedings of the SIGCHI conference on Human factors in computing systems. 164–170

    work page 2002

  64. [64]

    2024.The Psychology Behind Effective UI/UX Design

    Shaan Nicol. 2024.The Psychology Behind Effective UI/UX Design. ChillyBin. https://www.chillybin.co/psychology-ui-ux-design/

    work page 2024

  65. [65]

    David G Novick, Oscar D Andrade, and Nathaniel Bean. 2009. The micro-structure of use of help. InProceedings of the 27th ACM international conference on Design of communication. 97–104

    work page 2009

  66. [66]

    David G Novick and Karen Ward. 2006. Why don’t people read the manual?. InProceedings of the 24th annual ACM international conference on Design of communication. 11–18

    work page 2006

  67. [67]

    2025.Blender Copilot 5.1

    OpenAI. 2025.Blender Copilot 5.1. OpenAI. https://chatgpt.com/g/g-w7Ewzzi94- blender-copilot

    work page 2025

  68. [68]

    2025.OpenShot

    OpenShot. 2025.OpenShot. OpenShot Studios, LLC. https://www.openshot.org/

    work page 2025

  69. [69]

    2025.Adobe Photoshop

    Adobe Photoshop. 2025.Adobe Photoshop. Adobe. https://www.adobe.com/pr oducts/photoshop.html

    work page 2025

  70. [70]

    Barron, and Ben Mildenhall

    Ben Poole, Ajay Jain, Jonathan T. Barron, and Ben Mildenhall. 2022. DreamFusion: Text-to-3D using 2D Diffusion.arXiv(2022)

    work page 2022

  71. [71]

    Vesna Popovic. 2000. Expert and novice user differences and implications for product design and useability. InProceedings of the Human Factors and Ergonomics Society Annual Meeting, Vol. 44. SAGE Publications Sage CA: Los Angeles, CA, 933–936

    work page 2000

  72. [72]

    2025.Prolific

    Prolific. 2025.Prolific. Prolific. https://www.prolific.com/

    work page 2025

  73. [73]

    Shuofei Qiao, Runnan Fang, Zhisong Qiu, Xiaobin Wang, Ningyu Zhang, Yong Jiang, Pengjun Xie, Fei Huang, and Huajun Chen. 2024. Benchmarking Agentic Workflow Generation.arXiv preprint arXiv:2410.07869(2024)

    work page arXiv 2024

  74. [74]

    Bosheng Qin, Juncheng Li, Siliang Tang, Tat-Seng Chua, and Yueting Zhuang

    work page

  75. [75]

    In2024 IEEE International Conference on Multimedia and Expo (ICME)

    Instructvid2vid: Controllable video editing with natural language instruc- tions. In2024 IEEE International Conference on Multimedia and Expo (ICME). IEEE, 1–6

    work page

  76. [76]

    John Rieman. 1996. A field study of exploratory learning strategies.ACM Transactions on Computer-Human Interaction (TOCHI)3, 3 (1996), 189–218

    work page 1996

  77. [77]

    Ian T Ruginski, Sarah H Creem-Regehr, Jeanine K Stefanucci, and Elizabeth Cashdan. 2019. GPS use negatively affects environmental learning through spatial transformation abilities.Journal of Environmental Psychology64 (2019), 12–20

    work page 2019

  78. [78]

    Skipper Seabold and Josef Perktold. 2010. statsmodels: Econometric and statistical modeling with python. InProceedings of the 9th Python in Science Conference. 57–61

    work page 2010

  79. [79]

    Abigail Sellen and Eric Horvitz. 2024. The rise of the ai co-pilot: Lessons for design from aviation and beyond.Commun. ACM67, 7 (2024), 18–23

    work page 2024

  80. [80]

    Ben Shneiderman. 2002. Promoting universal usability with multi-layer interface design.ACM SIGCAPH computers and the physically handicapped73-74 (2002), 1–8

    work page 2002

Showing first 80 references.