Back

Screen Reader Programmers in the Vibe Coding Era: Adaptation, Empowerment, and New Accessibility Landscape

Nan ChenLuna K. QiuArran Zeyu WangZilong WangYuqing Yang

article2025arXiv6 citations

Reveals how blind and low-vision developers interact with AI code assistants through a longitudinal study, identifying key accessibility barriers in interpreting machine-generated output and situational awareness while establishing actionable design principles for inclusive programming tools.

Listen

Artificial intelligence coding assistants and autonomous agents are rapidly altering software engineering through natural language-driven programming, often termed vibe coding. While these advancements shift general developer workflows from manual coding to supervising machine-driven actions, software developers with visual impairments face unique accessibility barriers. These developers rely on screen readers to interpret graphical interfaces and code linearly through synthesized speech. As integrated development environments increasingly introduce automated, multi-turn AI features, there is an urgent need to understand whether these tools empower screen reader users or introduce new navigational and cognitive obstacles.

The article aims to evaluate how screen reader programmers interact with advanced AI code assistants in real-world programming contexts. Specifically, it assesses the extent of empowerment provided by these tools, identifies novel accessibility and usability friction points, and analyzes how user preferences evolve between fully autonomous workflows and user-controlled interactions.

To evaluate these interactions, researchers conducted a two-week, three-phase longitudinal study involving 16 blind and low-vision programmers using GitHub Copilot within Visual Studio Code. The methodology comprised an initial hands-on programming session with structured tasks and semi-structured interviews, followed by a two-week exploration period where participants used the assistant during their routine coding activities and logged diary entries, concluding with follow-up reflective interviews with 15 participants. The research team analyzed interaction timelines, screen recordings, transcripts, and diary logs using thematic analysis grounded in Activity Theory.

The investigation revealed several key findings regarding efficiency, accessibility, and interaction trade-offs. Advanced AI code assistants significantly enhanced programming efficiency, with participants spending only about 25% of their active task time on manual code writing while the assistant handled boilerplate tasks and cognitive burdens. Multimodal and generative capabilities bridged long-standing visual accessibility gaps, enabling blind developers to undertake tasks such as user interface creation and image-to-code extraction. However, reviewing AI outputs proved to be the most time-consuming activity, averaging 11.73 minutes per session, due to linear screen reader navigation, ambiguous diff representations, and cognitive strain across multiple interface views. Furthermore, a pronounced behavioral shift occurred over the two-week study: while initial interest favored highly autonomous agent modes, longitudinal usage saw users retreat toward conversational query modes to retain verification control and avoid unintended codebase modifications. Significant onboarding hurdles also persisted due to unlabeled interface elements and a lack of non-visual learning resources.

These findings demonstrate that while generative AI holds immense potential to democratize complex programming tasks for visually impaired engineers, current human-AI interaction models assume visual oversight. Autonomous modifications by AI agents without transparent, non-visual status updates heighten cognitive load and undermine user trust. Consequently, accessibility in AI-assisted programming requires more than basic screen reader compatibility; it demands architectural considerations around process transparency, predictable navigation, and verifiability to prevent compounding the digital divide for developers with disabilities.

The article recommends designing AI assistants around interaction simplicity, predictable keyboard shortcuts, and transparent communication. Tool developers should implement accessible change-tracking mechanisms with structured audio or textual cues, provide explicit status notifications detailing agent actions, and enable models to proactively ask clarifying questions when prompts are ambiguous. Organizations should also support customizable interfaces—such as allowing users to toggle between detailed message views and simplified text displays—and provide screen reader-tailored onboarding documentation to facilitate adoption.

Confidence in these findings is supported by the longitudinal, multi-method qualitative design and real-world task validation. However, limitations include the relatively small sample size of 16 participants, a participant pool primarily based in China, a male-skewed gender balance, and reliance on self-reported diary data during the exploration phase. Future initiatives should evaluate more demographically diverse developer populations, examine different commercial models, and implement telemetry-based tracking to validate long-term adoption dynamics.

MicrosoftUniversity of North Carolina at Chapel Hillautomationgenerative aimultiple viewsscreen reader programmersblind software developerscode assistants
arXiv: 2506.13270
  • Paper: Human-Centered Artificial Intelligence: Reliable, Safe & Trustworthy, Ben Shneiderman (2020). It establishes the foundational human-centered AI framework for balancing high automation with high human control, which directly underpins the source paper's analysis of developer agency and supervision.
  • Paper: Evaluating Large Language Models Trained on Code, Mark Chen et al. (2021). It introduces OpenAI Codex and foundational code generation benchmarks, representing the underlying technology and interaction model that GitHub Copilot tools studied in the source build upon.
  • Paper: SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering, John Yang et al. (2024). It formalizes Agent-Computer Interfaces (ACIs) for interacting with AI coding assistants and managing context feedback, directly informing the UI and accessibility challenges evaluated in the source.
  • Paper: Program Synthesis with Large Language Models, Jacob Austin et al. (2021). It details how large language models handle program synthesis and interactive refinement with human feedback, establishing baseline principles for developer-AI collaborative coding.
  • Paper: Generative AI, Stefan Feuerriegel et al. (2023). It conceptualizes generative AI's paradigm shift from one-way task delegation to iterative co-creation and supervisory interaction in human-computer workflows.
Cover for Screen Reader Programmers in the Vibe Coding Era: Adaptation, Empowerment, and New Accessibility Landscape

Abstract

Generative AI agents are reshaping human-computer interaction, shifting users from direct task execution to supervising machine-driven actions, especially the rise of "vibe coding" in programming. Yet little is known about how screen reader programmers interact with AI code assistants in practice. We conducted a longitudinal study with 16 blind and low-vision programmers. Participants completed a GitHub Copilot tutorial, engaged with a programming task, and provided initial feedback. After two weeks of AI-assisted programming, follow-ups examined how their practices and perceptions evolved. Our findings show that code assistants enhanced programming efficiency and bridged accessibility gaps. However, participants struggled to convey intent, interpret AI outputs, and manage multiple views while maintaining situational awareness. They showed diverse preferences for accessibility features, expressed a need to balance automation with control, and encountered barriers when learning to use these tools. Furthermore, we propose design principles and recommendations for more accessible and inclusive human-AI collaborations.

Topics

automationgenerative aimultiple viewsrecommendationscreen readerssituation awarenesshuman-ai collaborationprogramming languagescoding scheme

Keywords

screen reader programmersblind software developerscode assistantsaccessible computingaccessible developmentaccessibility barriersprompt engineeringUI development

Table of Contents

  • 1 Introduction
  • 2 Related Work
  • 2.1 AI Code Assistants and Vibe Coding
  • 2.2 Challenges for Programmers with Visual Impairments
  • 2.3 Human-AI Interaction for People with Visual Impairments
  • 3 Methodology
  • 3.1 Copilot-assisted Programming
  • 3.2 Participants
  • 3.3 Initial Study
  • 3.3.1 Task
  • 3.4 Two-Week Exploration Phase
  • 3.5 Follow-up Study
  • 3.6 Data Analysis
  • 3.7 Positionality
  • 4 Findings
  • 4.1 AI’s Value and Empowerment
  • 4.1.1 Initial Study: How Can AI Assistants Support and Reshape the Coding Process
  • 4.1.2 Follow-up Study: How Can AI Assistants Support Real-World Programming Tasks
  • 4.2 Challenges in Complex Human-AI Interaction Workflows
  • 4.2.1 Communicating with AI Code Assistant
  • 4.2.2 Reviewing Generated Responses
  • 4.2.3 Switching Between Views
  • 4.2.4 System Status and Next Steps
  • 4.3 Accessibility Features: Trade-offs and User Perceptions
  • 4.3.1 Accessible View: Ease of Use vs. Information Completeness
  • 4.3.2 Automatic Playback and Status Notifications: Convenience vs. Disruption
  • 4.3.3 Sound Cues for Code Changes: Helpfulness vs. Cognitive Load
  • 4.4 Feature Preferences: Automation vs. Control
  • 4.5 Barriers to Learning for Screen Reader Users
  • 5 Discussion
  • 5.1 Reflections on Vibe Coding
  • 5.2 Toward a Sense of Control
  • 5.3 For the Future: Towards Accessible Human-AI Collaboration Design
  • 5.3.1 High-Level Principles
  • 5.3.2 Concrete Recommendations
  • 5.4 Limitation and Future Work
  • 6 Conclusion
  • References

Knowls

  1. Knowl 1 — Three-Phase Longitudinal Study Setup for AI Code Assistant Accessibility

    experimental setup

    To investigate how blind and low-vision (BLV) programmers interact with modern AI coding assistants (such as GitHub Copilot in Visual Studio Code), a three-phase longitudinal study was designed:

    1. Phase 1: Initial Lab Study (N=16N = 16). Participants engaged in a 110–140 minute session comprising: (a) a semi-structured background interview assessing prior programming and AI tool experience; (b) a customized tutorial introducing GitHub Copilot features (code completion, inline chat, and chat panel with Ask, Edit, and Agent modes) configured for screen readers; (c) a 60-minute programming task randomly assigned from four collaborative software engineering scenarios (T1T_1: Chat Server, T2T_2: Chat Client, T3T_3: Data Analysis, T4T_4: Calculator Debugging) designed so that no task could be completed with a single prompt; and (d) a 10-item, 7-point Likert-scale survey accompanied by a semi-structured post-task interview.

    2. Phase 2: Two-Week In-Situ Exploration. Participants used GitHub Copilot during their daily programming work and logged noteworthy positive and negative interaction experiences using a structured diary template (yielding 31 collected diary cases).

    3. Phase 3: Follow-Up Reflection Study (N=15N = 15). In a 30-minute session, participants reflected on their diary logs, explained changes in their interaction strategies, repeated the 7-point Likert-scale questionnaire, and participated in an exit interview.

  2. Knowl 2 — Time Distribution and Cognitive Role Shift in Non-Visual AI-Assisted Coding

    empirical result

    Integrating AI assistants shifted screen reader users' primary activity from direct code authoring to prompt steering, output inspection, and validation:

    • Overall Activity Breakdown: During the initial programming tasks, participants spent an average of 43.1243.12 minutes on programming-related activities. However, only 10.6910.69 minutes (25.41%25.41\%) was spent on direct coding activities (with purely manual code typing accounting for just 2.892.89 minutes).
    • Distribution Across AI-Assisted Sub-activities:
      • Reviewing AI Outputs: 11.7311.73 minutes (the single most time-consuming activity across all tasks).
      • Prompt Engineering: 7.517.51 minutes (average of 8.388.38 prompts submitted per user).
      • Waiting for AI Generation: 5.595.59 minutes (averaging 1.461.46 minutes of waiting per request across an average of 3.683.68 explicit AI requests).
      • Validation (running tests/programs): 5.185.18 minutes.
      • Fixing Generated Errors: 2.422.42 minutes.
    • Delegation of Review Tasks: 37.5%37.5\% of participants (6 out of 16) actively delegated verification tasks to the AI itself (e.g., prompting the assistant to check unit test coverage, evaluate function edge cases, or verify non-standard library dependencies) to reduce auditory review fatigue.

  3. Knowl 3 — Longitudinal Evolution of Interaction Modes: From Automated Agents to Supervised Conversations

    empirical result

    User preferences across GitHub Copilot's interaction modes exhibited a marked longitudinal divergence between initial exploration and extended regular use:

    • Initial Study Preferences: In the initial lab tasks, participants heavily favored the fully automated Agent mode, utilizing it 7272 times, compared to Edit mode (2828 uses), Ask mode (2222 uses), Inline Chat (1313 uses), and Code Completion (11 use).
    • Longitudinal Real-World Shift: After two weeks of in-situ development, participant preference shifted toward safer, more conversational modes. Twelve participants frequently used Ask mode, 1010 used Agent mode, 55 used Edit mode, 55 used Code Completion, and 22 used Inline Chat.
    • Underlying Drivers of Preference Shift:
      • Loss of Agency and Quota Consumption: Unsupervised Agent mode occasionally executed misaligned file searches, altered unrequested code blocks, or consumed API quotas unnecessarily.
      • Verification Overhead: Multi-file, autonomous changes produced by Agent mode imposed substantial cognitive load on screen reader users, who could not quickly glance across workspace diffs to verify changes.
      • Predictability of Ask Mode: Ask mode served as a non-destructive sandboxed interface where users could clarify requirements, generate isolated snippets, and maintain supervisory control before manually applying changes.

  4. Knowl 4 — Accessibility Enablement and Visual Gap Bridging via AI Code Assistants

    empirical result

    Advanced AI code assistants enabled screen reader programmers to overcome long-standing accessibility barriers in inherently visual domains:

    1. User Interface (UI) Development: Visually impaired developers historically struggled with UI layout and spatial component placement. With multimodal and agentic code assistants, participants drafted front-end interfaces by translating textual user feedback into layout prompts, asking the assistant to evaluate UI code, and submitting screenshots back to multimodal models to verify layout hierarchy and prevent overlapping elements.
    2. Multimodal Code and Diagram Interpretation: Participants leveraged multimodal models to extract editable source code from non-accessible image screenshots in technical blogs and interpret visual hardware pin diagrams for embedded systems.
    3. Rapid Tool Onboarding and Prototyping: AI assistants lowered the barrier to learning unfamiliar programming languages, identifying third-party frameworks, and refactoring legacy codebases for custom DIY accessibility tools.

  5. Knowl 5 — Interaction Frictions in Non-Visual Multi-View Navigation and Status Tracking

    empirical result

    Screen reader programmers encountered several interaction bottlenecks when operating within multi-view, AI-driven IDE environments:

    • Keyboard Shortcut Collisions: Standard keybindings conflicted with screen reader navigation habits. For instance, pressing the Up/Down arrow keys in the chat input box cycled through prior prompt command history rather than navigating line-by-line within a multiline prompt.
    • Focus Loss and Multi-View Disorientation: Navigating between the code editor, inline chat, chat message list, accessible view, diff viewer, and integrated terminal led to frequent focus traps. Users inadvertently typed prompt queries into active source code files or missed background build/execution failures in the terminal.
    • Ambiguous Status and Notification Masking: Live auditory status indicators lacked semantic clarity. During multi-step Agent execution, the IDE system announced "Action required: Run command in terminal" but immediately overwrote it with periodic "Progress" announcements, causing users to miss required confirmations.
    • Diff View Auditory Mixing: Screen reader line-by-line reading in unified diff views intermingled added, deleted, and unchanged lines without auditory separation, making spatial comparison of code modifications cognitively exhausting.

  6. Knowl 6 — Usability Trade-Offs in Non-Visual Modalities for AI Code Assistants

    empirical result

    Evaluation of specialized accessibility features revealed crucial design trade-offs between simplicity and information richness:

    • Accessible View vs. Message List:
      • Accessible View: Presents output as a plain-text read-only buffer navigated with standard arrow keys, eliminating focus traps. However, it strips rich semantic markup, degrades table readability, omits modified file lists, and requires manual refresh to display new output streaming.
      • Structured Message List: Retains rich semantic structure (enabling heading and button jumps via screen reader browse mode), but introduces steep navigation hierarchies that overwhelm novice users.
    • Auditory Cues vs. Speech Interruptions:
      • Automatic Speech Playback: Automatically reading generated text upon arrival interrupted ongoing thought processes and clashed with active screen reader speech.
      • Non-Speech Earcons (Sound Cues): Distinct sound cues denoting insertions, deletions, and modifications provided non-disruptive feedback once memorized, but required explicit training to distinguish reliably.
      • Continuous Status Sounds: Background sound loops indicating active generation allowed users to multitask, provided they did not mask screen reader speech.

  7. Knowl 7 — Onboarding, Discoverability, and Interface Inaccessibility Barriers for Screen Reader Programmers

    empirical result

    The adoption of AI coding assistants by screen reader users is hindered by accessibility defects and documentation gaps:

    • Unlabeled UI Components: Competing AI development environments (such as Cursor and Cline) contained accessibility bugs where input boxes and code snippet containers lacked proper Accessible Rich Internet Applications (ARIA) roles and labels (e.g., rendering both simply as generic 'Input' elements), forcing users to write custom NVDA screen reader add-ons to parse interfaces.
    • Feature Invisibility: Because screen reader users navigate sequentially, nested or dynamically inserted controls were frequently missed. For example, GitHub Copilot's built-in 'Undo' button remained undiscovered by multiple participants after two weeks because it was located within historical prompt messages rather than the assistant's response block.
    • Absence of Non-Visual Learning Material: Mainstream AI tutorials rely heavily on visual directions (e.g., 'click the icon here'). The absence of screen-reader-specific keyboard walkthroughs and documentation caused significant onboarding delays.

  8. Knowl 8 — Human-AI Collaboration Design Principles and Recommendations for Screen Reader Developers

    model/method

    To create inclusive, non-visual AI programming assistants, software systems should follow three foundational principles and nine concrete recommendations:

    1. High-Level Principles

    • Embrace Interaction Simplicity: Minimize cognitive load through predictable, uniform keyboard navigation patterns and simplified structural hierarchies.
    • Maintain Transparent Communication: Deliver comprehensive, real-time, non-visual feedback detailing system progress, executed commands, and verifiable rationale.
    • Design for Inclusive Learning Journeys: Provide multimodal, non-visual documentation and conversational onboarding to build conceptual and procedural mental models.

    2. Concrete Recommendations

    • Interaction and Navigation:
      1. Consistent Shortcuts: Standardize hotkeys to prevent conflicts with native screen reader and editor navigation keys.
      2. Structured Content Navigation: Group multi-turn dialogue into distinct sections with headings, tables of contents, and collocated results.
      3. Multi-Attribute Change Tracking: Offer dedicated non-visual summaries describing the scope (number of edits), context (file/line positions), and content of code modifications.
      4. Interruption Control: Implement customizable notification modes (such as a 'Do Not Disturb' setting) alongside on-demand status querying.
    • Transparency and Verifiability: 5. Proactive Clarification: Enable AI agents to prompt users for clarification when instructions are ambiguous before making workspace edits. 6. Verifiable Outputs: Provide direct citations, documentation links, and execution traces to facilitate confidence calibration. 7. Actionable Status Updates: Provide distinct, unambiguous auditory notifications distinguishing background processing from states awaiting user input.
    • Learning and Customization: 8. Intent-Based Model Selection: Automatically suggest or select optimal LLM backends based on task type (e.g., debugging vs. documentation). 9. Conversational Onboarding Guidance: Leverage interactive conversational agents to guide screen reader users through feature discovery and accessibility configuration.

  9. Knowl 9 — Participant Demographics, Tool Familiarity, and Task Completion Rates

    data/table

    The study evaluated 16 screen reader programmers across four real-world software engineering tasks (T1T_1: Chat Server, T2T_2: Chat Client, T3T_3: Data Analysis, T4T_4: Calculator Debugging). The table summarizes participant visual status, programming experience, AI assistant usage frequency, task assigned, and requirement completion success:

    ID Gender/Age Status Exp. Occupation AI Frequency Task Result
    P1 Male / 34 Blind 7 yrs Freelancer Daily T2 1/2
    P2 Male / 39 Blind 19 yrs Game Dev. Daily T4 3/9
    P3 Male / 29 Blind 2 yrs Masseur Daily T3 1/2
    P4 Male / 42 Blind 10 yrs Software Eng. Monthly T4 9/10
    P5 Male / 24 Blind 3 yrs Accessibility Dev. Daily T2 1/2
    P6 Male / 36 Blind 18 yrs Accessibility Dev. Daily T3 1/2
    P7 Female / 27 Blind 7 yrs Freelancer Monthly T4 6/10
    P8 Male / 41 Low Vision 21 yrs Developer Monthly T3 1/2
    P9 Female / 22 Low Vision 1 yr Student Daily T2 0/2
    P10 Male / 28 Blind 9 yrs Accessibility Dev. Daily T1 2/2
    P11 Male / 29 Low Vision 5 yrs Accessibility Dev. Weekly T1 0/2
    P12 Male / 24 Blind 3 yrs Student Weekly T1 0/2
    P13 Male / 32 Blind 10 yrs Accessibility QA Daily T1 1/2
    P14 Male / 29 Blind 6 yrs Software Eng. Weekly T4 8/10
    P15 Male / 35 Blind 10 yrs Accessibility Dev. Daily T3 1/2
    P16 Male / 45 Blind 10 yrs Masseur Weekly T2 0/2

    Visual status comprised 13 totally blind and 3 low-vision developers. Programming experience ranged from 1 to 21 years (median ≈8\approx 8 years).

  10. Knowl 10 — Limitations in Sample Representation, Privacy Constraints, and Model Variance

    limitation

    The empirical findings of the study are bounded by several methodological and ecological limitations:

    • Sample Size and Demographic Skew: The cohort included 16 participants (14 male, 2 female), all residing in China. Cultural norms, localized programming conventions, and regional screen reader ecosystems (primarily NVDA in Mandarin) may influence findings.
    • Telemetry vs. Privacy Constraints: Due to privacy safeguards, IDE telemetry and raw conversational chat logs were not passively tracked during the two-week exploration phase; longitudinal data relied on self-reported diary entries and retrospective interviews.
    • Confounding Model Variance: The specific impact of underlying foundation models (e.g., GPT-4o vs. Claude 3.7) versus IDE interface design was not isolated as an independent experimental variable.
    • End-User Programming Scope: The participant pool was restricted to individuals with existing programming background (1–21 years of experience), leaving the impact of 'vibe coding' on non-programmer BLV end users unaddressed.

Coverage note — None. All core empirical findings, design principles, interaction analyses, data tables, and study methodologies were incorporated.

References

  1. 1.Jacob Abbott, Haley MacLeod, Novia Nurain, Gustave Ekobe, and Sameer Patil. 2019. Local standards for anonymization practices in health, wellness, accessibility, and aging research at CHI. In Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems. 1–14.
  2. 2.Rudaiba Adnin and Maitraye Das. 2024. " I look at it as the king of knowledge": How Blind People Use and Understand Generative AI Tools. In Proceedings of the 26th International ACM SIGACCESS Conference on Computers and Accessibility. 1–14.
  3. 3.Mehmet Akhoroz and Caglar Yildirim. 2025. Conversational AI as a Coding Assistant: Understanding Programmers’ Interactions with and Expectations from Large Language Models for Coding. arXiv preprint arXiv:2503.16508 (2025).
  4. 4.Khaled Albusays and Stephanie Ludi. 2016. Eliciting programming challenges faced by developers with visual impairments: exploratory study. In Proceedings of the 9th international workshop on cooperative and human aspects of software engineering. 82–85.
  5. 5.Khaled Albusays, Stephanie Ludi, and Matt Huenerfauth. 2017. Interviews and observation of blind software developers at work to understand code navigation challenges. In Proceedings of the 19th International ACM SIGACCESS Conference on Computers and Accessibility. 91–100.
  6. 6.Rahaf Alharbi, Pa Lor, Jaylin Herskovitz, Sarita Schoenebeck, and Robin N Brewer. 2024. Misfitting With AI: How Blind People Verify and Contest AI Errors. In Proceedings of the 26th International ACM SIGACCESS Conference on Computers and Accessibility. 1–17.
  7. 7.Owura Asare, Meiyappan Nagappan, and N Asokan. 2024. A user-centered security evaluation of copilot. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–11.
  8. 8.Alex Atcheson, Omar Khan, Brian Siemann, Anika Jain, and Karrie Karahalios. 2025. " I’d Never Actually Realized How Big An Impact It Had Until Now": Perspectives of University Students with Disabilities on Generative Artificial Intelligence. (2025).
  9. 9.Gagan Bansal, Jennifer Wortman Vaughan, Saleema Amershi, Eric Horvitz, Adam Fourney, Hussein Mozannar, Victor Dibia, and Daniel S Weld. 2024. Challenges in human-agent communication. arXiv preprint arXiv:2412.10380 (2024).
  10. 10.Shraddha Barke, Michael B James, and Nadia Polikarpova. 2023. Grounded copilot: How programmers interact with code-generating models. Proceedings of the ACM on Programming Languages 7, OOPSLA1 (2023), 85–111.
  11. 11.David R. Barstow. 1984. A Perspective on Automatic Programming. AI Magazine 5, 1 (Mar. 1984), 5.
  12. 12.Be My Eyes. 2025. Introducing Be My AI. https://www.bemyeyes.com/blog/introducing-be-my-ai/. Accessed: 2025-05-03.
  13. 13.Andrew Begel, Yit Phang Khoo, and Thomas Zimmermann. 2010. Codebook: discovering and exploiting relationships in software repositories. In Proceedings of the 32nd ACM/IEEE International Conference on Software Engineering-Volume 1. 125–134.
  14. 14.Christian Bird, Denae Ford, Thomas Zimmermann, Nicole Forsgren, Eirini Kalliamvakou, Travis Lowdermilk, and Idan Gazit. 2022. Taking Flight with Copilot: Early insights and opportunities of AI-powered pair-programming tools. Queue 20, 6 (2022), 35–57.
  15. 15.Yevgen Borodin, Jeffrey P Bigham, Glenn Dausch, and IV Ramakrishnan. 2010. More than meets the eye: a survey of screen-reader browsing strategies. In Proceedings of the 2010 international cross disciplinary conference on web accessibility (W4A). 1–10.
  16. 16.Virginia Braun and Victoria Clarke. 2006. Using thematic analysis in psychology. Qualitative research in psychology 3, 2 (2006), 77–101.
  17. 17.Avyay Casheekar, Archit Lahiri, Kanishk Rath, Kaushik Sanjay Prabhakar, and Kathiravan Srinivasan. 2024. A contemporary review on chatbots, AI-powered virtual conversational agents, ChatGPT: Applications, open challenges and future research directions. Computer Science Review 52 (2024), 100632.
  18. 18.Yoonha Cha, Isabela Figueira, Jessy Ayala, Emory James Edwards, Joshua Garcia, André van der Hoek, and Stacy Marie Branham. 2024. " Do You Want Me to Participate or Not?": Investigating the Accessibility of Software Development Meetings for Blind and Low Vision Professionals. In Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems. 1–17.
  19. 19.Yoonha Cha, Victoria Jackson, Karina Kohl, Rafael Prikladnicki, André van der Hoek, and Stacy Branham. 2025. The dilemma of building do-it-yourself (diy) solutions for workplace accessibility. In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems. 1–17.
  20. 20.Ana Paula Chaves and Marco Aurelio Gerosa. 2021. How should my chatbot interact? A survey on social characteristics in human–chatbot interaction design. International Journal of Human–Computer Interaction 37, 8 (2021), 729–758.
  21. 21.Janet X Chen, Allison McDonald, Yixin Zou, Emily Tseng, Kevin A Roundy, Acar Tamersoy, Florian Schaub, Thomas Ristenpart, and Nicola Dell. 2022. Trauma-informed computing: Towards safer technology experiences for all. In Proceedings of the 2022 CHI conference on human factors in computing systems. 1–20.
  22. 22.Mark Chen, Jerry Tworek, and Heewoo Jun et al. 2021. Evaluating Large Language Models Trained on Code. arXiv:2107.03374 [cs.LG] https: //arxiv.org/abs/2107.03374
  23. 23.Nan Chen, Yuge Zhang, Jiahang Xu, Kan Ren, and Yuqing Yang. 2024. Viseval: A benchmark for data visualization in the era of large language models. IEEE Transactions on Visualization and Computer Graphics (2024).
  24. 24.Cline. 2025. Cline. https://cline.bot/. Accessed: 2025-05-03.
  25. 25.Juliet Corbin and Anselm Strauss. 2014. Basics of qualitative research: Techniques and procedures for developing grounded theory. Sage publications.
  26. 26.David Coyle, James Moore, Per Ola Kristensson, Paul Fletcher, and Alan Blackwell. 2012. I did that! Measuring users’ experience of agency in their own actions. In Proceedings of the SIGCHI conference on human factors in computing systems. 2025–2034.
  27. 27.Cursor. 2025. Cursor. https://www.cursor.com/. Accessed: 2025-05-03.
  28. 28.Cursor. 2025. Introducing Agent Mode in Cursor. https://docs.cursor.com/chat/agent. Accessed: 2025-05-05.
  29. 29.Arghavan Moradi Dakhel, Vahid Majdinasab, Amin Nikanjam, Foutse Khomh, Michel C Desmarais, and Zhen Ming Jack Jiang. 2023. Github copilot ai pair programmer: Asset or liability? Journal of Systems and Software 203 (2023), 111734.
  30. 30.Paul Denny, Viraj Kumar, and Nasser Giacaman. 2023. Conversing with copilot: Exploring prompt engineering for solving cs1 problems using natural language. In Proceedings of the 54th ACM technical symposium on computer science education V. 1. 1136–1142.
  31. 31.Yrjo Engestrom. 2000. Activity theory as a framework for analyzing and redesigning work. Ergonomics 43, 7 (2000), 960–974.
  32. 32.Deborah Etsenake and Meiyappan Nagappan. 2024. Understanding the Human-LLM Dynamic: A Literature Survey of LLM Use in Programming Tasks. arXiv preprint arXiv:2410.01026 (2024).
  33. 33.Claire Ferrari and Amy Hurst. 2021. Accessible web development: Opportunities to improve the education and practice of web development with a screen reader. ACM Transactions on Accessible Computing (TACCESS) 14, 2 (2021), 1–32.
  34. 34.Claudia Flores-Saviaga, Shangbin Feng, and Saiph Savage. 2022. Datavoidant: An AI System for Addressing Political Data Voids on Social Media. Proc. ACM Hum.-Comput. Interact. 6, CSCW2, Article 503 (Nov. 2022), 29 pages. doi:10.1145/3555616
  35. 35.Claudia Flores-Saviaga, Benjamin V Hanrahan, Kashif Imteyaz, Steven Clarke, and Saiph Savage*. 2025. The Impact of Generative AI Coding Assistants on Developers Who Are Visually Impaired. In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems. 1–17.
  36. 36.GitHub. 2025. GitHub Copilot. https://github.com/features/copilot. Accessed: 2025-05-03.
  37. 37.GitHub. 2025. GitHub Copilot Accessibility Conformance Report for Visual Studio Code. https://accessibility.github.com/conformance/copilot-vs-code/. Accessed: 2024-05-02.
  38. 38.Ricardo E Gonzalez Penuela, Jazmin Collins, Cynthia Bennett, and Shiri Azenkot. 2024. Investigating use cases of ai-powered scene description applications for blind and low vision people. In Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems. 1–21.
  39. 39.Ricardo E Gonzalez Penuela, Ruiying Hu, Sharon Lin, Tanisha Shende, and Shiri Azenkot. 2025. Towards Understanding the Use of MLLM-Enabled Applications for Visual Interpretation by Blind and Low Vision People. In Proceedings of the Extended Abstracts of the CHI Conference on Human Factors in Computing Systems. 1–8.
  40. 40.Google. 2025. Google Gemini in Workspace. https://workspace.google.com/solutions/ai/. Accessed: 2025-05-03.
  41. 41.Rajesh Hegde and Prasun Dewan. 2008. Connecting programming environments to support ad-hoc collaboration. In 2008 23rd IEEE/ACM international conference on automated software engineering. IEEE, 178–187.
  42. 42.Jaylin Herskovitz, Andi Xu, Rahaf Alharbi, and Anhong Guo. 2023. Hacking, switching, combining: understanding and supporting DIY assistive technology design by blind people. In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems. 1–17.
  43. 43.Ellen Jiang, Edwin Toh, Alejandra Molina, Kristen Olson, Claire Kayacik, Aaron Donsbach, Carrie J Cai, and Michael Terry. 2022. Discovering the syntax and strategies of natural language programming with generative language models. In Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems. 1–19.
  44. 44.Jazette Johnson, Andrew Begel, Richard Ladner, and Denae Ford. 2022. Program-l: Online help seeking behaviors by blind and low vision programmers. In 2022 IEEE Symposium on Visual Languages and Human-Centric Computing (VL/HCC). IEEE, 1–6.
  45. 45.Shuli Jones, Isabella Pedraza Pineros, Daniel Hajas, Jonathan Zong, and Arvind Satyanarayan. 2024. “Customization is Key”: Reconfigurable Textual Tokens for Accessible Data Visualizations. In Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems (Honolulu, HI, USA) (CHI ’24). Association for Computing Machinery, New York, NY, USA, Article 47, 14 pages. doi:10.1145/3613904.3641970
  46. 46.J Bern Jordan, Victoria Van Hyning, Mason A Jones, Rachael Bradley Montgomery, Elizabeth Bottner, and Evan Tansil. 2024. Information Wayfinding of Screen Reader Users: Five Personas to Expand Conceptualizations of User Experiences. In Proceedings of the 26th International ACM SIGACCESS Conference on Computers and Accessibility. 1–7.
  47. 47.Andrej Karpathy. 2025. There’s a new kind of coding I call “vibe coding”, where you fully give in to the vibes, embrace exponentials, and forget that the code even exists. Tweet from account @karpathy. https://x.com/karpathy/status/1886192184808149383 Online; accessed 31-May-2025.
  48. 48.Majeed Kazemitabaar, Justin Chow, Carl Ka To Ma, Barbara J Ericson, David Weintrop, and Tovi Grossman. 2023. Studying the effect of AI code generators on supporting novice learners in introductory programming. In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems. 1–23.
  49. 49.Claire Kearney-Volpe and Amy Hurst. 2021. Accessible web development: Opportunities to improve the education and practice of web development with a screen reader. ACM Transactions on Accessible Computing (TACCESS) 14, 2 (2021), 1–32.
  50. 50.Rafal Kocielnik, Saleema Amershi, and Paul N Bennett. 2019. Will you accept an imperfect ai? exploring designs for adjusting end-user expectations of ai systems. In Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems. 1–14.
  51. 51.Harsh Kumar, Ilya Musabirov, Mohi Reza, Jiakai Shi, Xinyuan Wang, Joseph Jay Williams, Anastasia Kuzminykh, and Michael Liut. 2024. Guiding Students in Using LLMs in Supported Learning Environments: Effects on Interaction Dynamics, Learner Performance, Confidence, and Trust. Proceedings of the ACM on Human-Computer Interaction 8, CSCW2 (2024), 1–30.
  52. 52.Hao-Ping Lee, Advait Sarkar, Lev Tankelevitch, Ian Drosos, Sean Rintel, Richard Banks, and Nicholas Wilson. 2025. The impact of generative AI on critical thinking: Self-reported reductions in cognitive effort and confidence effects from a survey of knowledge workers. In Proceedings of the 2025 CHI conference on human factors in computing systems. 1–22.
  53. 53.Roberto Legaspi, Wenzhen Xu, Tatsuya Konishi, Shinya Wada, Nao Kobayashi, Yasushi Naruse, and Yuichi Ishikawa. 2024. The sense of agency in human–AI interactions. Knowledge-Based Systems 286 (2024), 111298.
  54. 54.Jenny T. Liang, Chenyang Yang, and Brad A. Myers. 2024. A Large-Scale Survey on the Usability of AI Programming Assistants: Successes and Challenges. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering (Lisbon, Portugal) (ICSE ’24). Article 52, 13 pages.
  55. 55.Hannah Limerick, David Coyle, and James W Moore. 2014. The experience of agency in human-computer interactions: a review. Frontiers in human neuroscience 8 (2014), 643.
  56. 56.Michael Xieyang Liu, Advait Sarkar, Carina Negreanu, Benjamin Zorn, Jack Williams, Neil Toronto, and Andrew D. Gordon. 2023. “What It Wants Me To Say”: Bridging the Abstraction Gap Between End-User Programmers and Code-Generating Large Language Models. In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems (Hamburg, Germany) (CHI ’23). Association for Computing Machinery, New York, NY, USA, Article 598, 31 pages. doi:10.1145/3544548.3580817
  57. 57.Mi Luo, Fei Chen, Pengxiang Cheng, Zhenhua Dong, Xiuqiang He, Jiashi Feng, and Zhenguo Li. 2020. Metaselector: Meta-learning for recommendation with user-level adaptive model selection. In Proceedings of The Web Conference 2020. 2507–2513.
  58. 58.Kelly Mack, Emma McDonnell, Dhruv Jain, Lucy Lu Wang, Jon E. Froehlich, and Leah Findlater. 2021. What do we mean by “accessibility research”? A literature survey of accessibility papers in CHI and ASSETS from 1994 to 2019. In Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems. 1–18.
  59. 59.S. Mealin and E. Murphy-Hill. 2012. An exploratory study of blind software developers . In 2012 IEEE Symposium on Visual Languages and Human-Centric Computing (VL/HCC 2012). IEEE Computer Society, Los Alamitos, CA, USA, 71–74. doi:10.1109/VLHCC.2012.6344485
  60. 60.Microsoft. 2025. Microsoft Copilot. https://www.microsoft.com/en-us/microsoft-365/copilot. Accessed: 2025-05-03.
  61. 61.Meredith Ringel Morris, Jazette Johnson, Cynthia L Bennett, and Edward Cutrell. 2018. Rich representations of visual content for screen reader users. In Proceedings of the 2018 CHI conference on human factors in computing systems. 1–11.
  62. 62.Aboubakar Mountapmbeme, Obianuju Okafor, and Stephanie Ludi. 2022. Addressing accessibility barriers in programming for people with visual impairments: A literature review. ACM Transactions on Accessible Computing (TACCESS) 15, 1 (2022), 1–26.
  63. 63.Hussein Mozannar, Gagan Bansal, Adam Fourney, and Eric Horvitz. 2024. Reading between the lines: Modeling user behavior and costs in AI-assisted programming. In Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems. 1–16.
  64. 64.Hussein Mozannar, Valerie Chen, Mohammed Alsobay, Subhro Das, Sebastian Zhao, Dennis Wei, Manish Nagireddy, Prasanna Sattigeri, Ameet Talwalkar, and David Sontag. 2024. The RealHumanEval: Evaluating Large Language Models’ Abilities to Support Programmers. arXiv:2404.02806 [cs.SE]
  65. 65.Notion. 2025. Notion AI. https://www.notion.com/help/guides/category/ai. Accessed: 2025-05-03.
  66. 66.NV Access. 2025. NVDA Screen Reader. https://www.nvaccess.org/. Accessed: 2025-05-03.
  67. 67.Maulishree Pandey, Sharvari Bondre, Sile O’Modhrain, and Steve Oney. 2022. Accessibility of ui frameworks and libraries for programmers with visual impairments. In 2022 IEEE Symposium on Visual Languages and Human-Centric Computing (VL/HCC). IEEE, 1–10.
  68. 68.Maulishree Pandey, Vaishnav Kameswaran, Hrishikesh V Rao, Sile O’Modhrain, and Steve Oney. 2021. Understanding accessibility and collaboration in programming for people with visual impairments. Proceedings of the ACM on Human-Computer Interaction 5, CSCW1 (2021), 1–30.
  69. 69.Maulishree Pandey, Steve Oney, and Andrew Begel. 2024. Towards Inclusive Source Code Readability Based on the Preferences of Programmers with Visual Impairments. In Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems. 1–18.
  70. 70.Minoli Perera, Swamy Ananthanarayan, Cagatay Goncu, and Kim Marriott. 2025. The Sky is the Limit: Understanding How Generative AI can Enhance Screen Reader Users’ Experience with Productivity Applications. In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems. 1–17.
  71. 71.Neil Perry, Megha Srivastava, Deepak Kumar, and Dan Boneh. 2023. Do users write more insecure code with AI assistants?. In Proceedings of the 2023 ACM SIGSAC conference on computer and communications security. 2785–2799.
  72. 72.Veronica Pimenova, Sarah Fakhoury, Christian Bird, Margaret-Anne Storey, and Madeline Endres. 2025. Good Vibrations? A Qualitative Study of Co-Creation, Communication, Flow, and Trust in Vibe Coding. arXiv preprint arXiv:2509.12491 (2025).
  73. 73.Venkatesh Potluri, Maulishree Pandey, Andrew Begel, Michael Barnett, and Scott Reitherman. 2022. Codewalk: Facilitating shared awareness in mixed-ability collaborative software development. In Proceedings of the 24th International ACM SIGACCESS Conference on Computers and Accessibility. 1–16.
  74. 74.Venkatesh Potluri, Priyan Vaithilingam, Suresh Iyengar, Y Vidya, Manohar Swaminathan, and Gopal Srinivasa. 2018. Codetalk: Improving programming environment accessibility for visually impaired developers. In Proceedings of the 2018 chi conference on human factors in computing systems. 1–11.
  75. 75.Ben Puryear and Gina Sprint. 2022. Github copilot in the classroom: learning to code with AI assistance. Journal of Computing Sciences in Colleges 38, 1 (2022), 37–47.
  76. 76.Vinitha Ranganeni, Mike Sinclair, Eyal Ofek, Amos Miller, Jonathan Campbell, Andrey Kolobov, and Edward Cutrell. 2023. Exploring levels of control for a navigation assistant for blind travelers. In Proceedings of the 2023 ACM/IEEE International Conference on Human-Robot Interaction. 4–12.
  77. 77.MIT Technology Review. 2025. What is vibe coding, exactly? — technologyreview.com. https://www.technologyreview.com/2025/04/16/1115135/what-is-vibe-coding-exactly/. [Accessed 05-05-2025].
  78. 78.Harini Sampath, Alice Merrick, and Andrew Macvean. 2021. Accessibility of command line interfaces. In Proceedings of the 2021 CHI conference on human factors in computing systems. 1–10.
  79. 79.Aylin Sarioğlu, Haydar Metin, and Dominik Bork. 2025. Accessibility in conceptual modeling—A systematic literature review, a keyboard-only UML modeling tool, and a research roadmap. Data & Knowledge Engineering (2025), 102423.
  80. 80.Advait Sarkar and Ian Drosos. 2025. Vibe coding: programming through conversation with artificial intelligence. In Proceedings of the 36th Annual Conference of the Psychology of Programming Interest Group (PPIG 2025).
  81. 81.JooYoung Seo and Megan Rogge. 2023. Coding non-visually in visual studio code: collaboration towards accessible development environment for blind programmers. In Proceedings of the 25th International ACM SIGACCESS Conference on Computers and Accessibility. 1–9.
  82. 82.Agnia Sergeyuk, Yaroslav Golubev, Timofey Bryksin, and Iftekhar Ahmed. 2025. Using AI-based coding assistants in practice: State of affairs, perceptions, and ways forward. Information and Software Technology 178 (2025), 107610.
  83. 83.Agnia Sergeyuk, Sergey Titov, and Maliheh Izadi. 2024. In-ide human-ai experience in the era of large language models; a literature review. In Proceedings of the 1st ACM/IEEE Workshop on Integrated Development Environments. 95–100.
  84. 84.Anshul Shah, Anya Chernova, Elena Tomson, Leo Porter, William G Griswold, and Adalbert Gerald Soosai Raj. 2025. Students’ Use of GitHub Copilot for Working with Large Code Bases. In Proceedings of the 56th ACM Technical Symposium on Computer Science Education V. 1. 1050–1056.
  85. 85.Ather Sharif, Sanjana Shivani Chintalapati, Jacob O Wobbrock, and Katharina Reinecke. 2021. Understanding screen-reader users’ experiences with online data visualizations. In Proceedings of the 23rd International ACM SIGACCESS Conference on Computers and Accessibility. 1–16.
  86. 86.Kristen Shinohara, Murtaza Tamjeed, Michael McQuaid, and Dymen A Barkins. 2022. Usability, accessibility and social entanglements in advanced tool use by vision impaired graduate students. Proceedings of the ACM on Human-Computer Interaction 6, CSCW2 (2022), 1–21.
  87. 87.Robert M Siegfried. 2006. Visual programming and the blind: the challenge and the opportunity. ACM SIGCSE Bulletin 38, 1 (2006), 275–278.
  88. 88.Michael Silverstein. 2003. Indexical order and the dialectics of sociolinguistic life. Language & communication 23, 3-4 (2003), 193–229.
  89. 89.Fangchen Song, Ashish Agarwal, and Wen Wen. 2024. The impact of generative AI on collaborative open-source software development: Evidence from GitHub Copilot. arXiv preprint arXiv:2410.02091 (2024).
  90. 90.Stack Overflow. 2022. Stack Overflow Developer Survey 2022. https://survey.stackoverflow.co/2022. Accessed: 2025-05-03.
  91. 91.Kevin M Storer, Harini Sampath, and M Alice Alice Merrick. 2021. ” It’s Just Everything Outside of the IDE that’s the Problem”: Information Seeking by Software Developers with Visual Impairments. In Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems. 1–12.
  92. 92.Mohammad Tahaei, Marios Constantinides, and Daniele Quercia et al. 2023. Human-Centered Responsible Artificial Intelligence: Current & Future Trends. In Extended Abstracts of the 2023 CHI Conference on Human Factors in Computing Systems (Hamburg, Germany) (CHI EA ’23). Article 515, 4 pages.
  93. 93.Anfu Tang, Laure Soulier, and Vincent Guigue. 2025. Clarifying ambiguities: on the role of ambiguity types in prompting methods for clarification generation. In Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval. 20–30.
  94. 94.Ningzhi Tang, Meng Chen, Zheng Ning, Aakash Bansal, Yu Huang, Collin McMillan, and T Li. 2023. An empirical study of developer behaviors for validating and repairing ai-generated code. In 13th Workshop on the Intersection of HCI and PL.
  95. 95.Xinru Tang, Ali Abdolrahmani, Darren Gergle, and Anne Marie Piper. 2025. Everyday Uncertainty: How Blind People Use GenAI Tools for Information Access. In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems. 1–17.
  96. 96.Visual Studio Code Team. 2025. Introducing Copilot Agent Mode. https://code.visualstudio.com/blogs/2025/02/24/introducing-copilot-agent-mode. Accessed: 2025-05-05.
  97. 97.Ruotong Wang, Ruijia Cheng, Denae Ford, and Thomas Zimmermann. 2024. Investigating and designing for trust in AI-powered code generation tools. In Proceedings of the 2024 ACM Conference on Fairness, Accountability, and Transparency. 1475–1493.
  98. 98.Justin D Weisz, Jessica He, Michael Muller, Gabriela Hoefer, Rachel Miles, and Werner Geyer. 2024. Design principles for generative AI applications. In Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems. 1–22.
  99. 99.Justin D. Weisz, Shraddha Vijay Kumar, Michael Muller, Karen-Ellen Browne, Arielle Goldberg, Katrin Ellice Heintze, and Shagun Bajpai. 2025. Examining the Use and Impact of an AI Code Assistant on Developer Productivity and Experience in the Enterprise. In Proceedings of the Extended Abstracts of the CHI Conference on Human Factors in Computing Systems (CHI EA ’25). Association for Computing Machinery, New York, NY, USA, Article 673, 13 pages. doi:10.1145/3706599.3706670
  100. 100.Wen Wen and Hiroshi Imamizu. 2022. The sense of agency in perception, behaviour and human–machine interactions. Nature Reviews Psychology 1, 4 (2022), 211–222.
  101. 101.Michel Wermelinger. 2023. Using GitHub Copilot to solve simple programming problems. In Proceedings of the 54th ACM Technical Symposium on Computer Science Education V. 1. 172–178.
  102. 102.Joel Wester, Bhakti Moghe, Katie Winkle, and Niels van Berkel. 2024. Facing LLMs: Robot Communication Styles in Mediating Health Information between Parents and Young Adults. Proceedings of the ACM on Human-Computer Interaction 8, CSCW2 (2024), 1–37.
  103. 103.Jim Whitehead. 2007. Collaboration in software engineering: A roadmap. In Future of Software Engineering (FOSE’07). IEEE, 214–225.
  104. 104.Wikipedia contributors. 2024. Vibe coding. https://en.wikipedia.org/wiki/Vibe_coding. Accessed: 2025-05-12.
  105. 105.Zhiheng Xi, Wenxiang Chen, Xin Guo, Wei He, Yiwen Ding, Boyang Hong, Ming Zhang, Junzhe Wang, Senjie Jin, Enyu Zhou, et al. 2025. The rise and potential of large language model based agents: A survey. Science China Information Sciences 68, 2 (2025), 121101.
  106. 106.Jingyi Xie, Rui Yu, He Zhang, Syed Masum Billah, Sooyeon Lee, and John M Carroll. 2025. Beyond Visual Perception: Insights from Smartphone Interaction of Visually Impaired Users with Large Multimodal Models. In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems. 1–17.
  107. 107.Eliana Zen, Vinicius Kruger Da Costa, and Tatiana Aires Tavares. 2023. Understanding the accessibility barriers faced by learners with visual impairments in computer programming. In Proceedings of the XXII Brazilian Symposium on Human Factors in Computing Systems. 1–11.
  108. 108.Michael JQ Zhang and Eunsol Choi. 2025. Clarify when necessary: Resolving ambiguity through interaction with lms. In Findings of the Association for Computational Linguistics: NAACL 2025. 5526–5543.
  109. 109.Yuge Zhang, Qiyang Jiang, XingyuHan XingyuHan, Nan Chen, Yuqing Yang, and Kan Ren. 2024. Benchmarking Data Science Agents. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 5677–5700.
  110. 110.Albert Ziegler, Eirini Kalliamvakou, X Alice Li, Andrew Rice, Devon Rifkin, Shawn Simister, Ganesh Sittampalam, and Edward Aftandilian. 2024. Measuring GitHub Copilot’s impact on productivity. Commun. ACM 67, 3 (2024), 54–63.

Citation

MLA
Chen, N., et al. “Screen Reader Programmers in the Vibe Coding Era: Adaptation, Empowerment, and New Accessibility Landscape”. arXiv, 2025, http://arxiv.org/abs/2506.13270v2.
APA
Chen, N., Qiu, L. K., Wang, A. Z., Wang, Z., & Yang, Y. (2025). Screen Reader Programmers in the Vibe Coding Era: Adaptation, Empowerment, and New Accessibility Landscape. arXiv. http://arxiv.org/abs/2506.13270v2
Chicago
Chen, N., L. K. Qiu, A. Z. Wang, Z. Wang, and Y. Yang. 2025. “Screen Reader Programmers in the Vibe Coding Era: Adaptation, Empowerment, and New Accessibility Landscape”. arXiv. http://arxiv.org/abs/2506.13270v2.
Harvard
Chen, N. et al. (2025) “Screen Reader Programmers in the Vibe Coding Era: Adaptation, Empowerment, and New Accessibility Landscape”, arXiv [Preprint]. Available at: http://arxiv.org/abs/2506.13270v2.
Vancouver
1. Chen N, Qiu LK, Wang AZ, Wang Z, Yang Y (2025) Screen Reader Programmers in the Vibe Coding Era: Adaptation, Empowerment, and New Accessibility Landscape. arXiv

BibTeX

@article{chen2025screen,
  title = {Screen Reader Programmers in the Vibe Coding Era: Adaptation, Empowerment, and New Accessibility Landscape},
  author = {Chen, Nan and Qiu, Luna K. and Wang, Arran Zeyu and Wang, Zilong and Yang, Yuqing},
  year = {2025},
  journal = {arXiv},
  url = {http://arxiv.org/abs/2506.13270v2},
  eprint = {2506.13270}
}
Metadata:arXiv

Access the Paper

This paper is available from its original source. Click below to access the PDF.

Open PDF
License: Published with permission