Paper deep dive
Channelling, Coordinating, Collaborating: A Three-Layer Framework for Disability-Centered Human-Agent Collaboration
Lan Xiao, Catherine Holloway
Intelligence
Status: succeeded | Model: google/gemini-3.1-flash-lite-preview | Prompt: intel-v1 | Confidence: 96%
Last extracted: 3/31/2026, 1:40:43 AM
Summary
The paper proposes a three-layer framework—Channelling, Coordinating, and Co-Creating—to rethink AI's role in ability-diverse human-agent collaboration. Moving away from individual-focused assistive technology, the framework centers on interdependence, grounding theory, and knowledge boundary management to support triadic collaboration where AI acts as a partner in workflows involving people with disabilities.
Entities (5)
Relation Signals (4)
Channelling → ispartof → Three-Layer Framework
confidence 100% · We propose a three-layer framework, Channelling, Coordinating, and Co-Creating
Coordinating → ispartof → Three-Layer Framework
confidence 100% · We propose a three-layer framework, Channelling, Coordinating, and Co-Creating
Co-Creating → ispartof → Three-Layer Framework
confidence 100% · We propose a three-layer framework, Channelling, Coordinating, and Co-Creating
Three-Layer Framework → groundedin → Carlile’s 3T framework
confidence 95% · Grounded in the Ability-Diverse Collaboration framework, grounding theory, and Carlile’s 3T framework
Cypher Suggestions (0)
No Cypher suggestions yet.
Abstract
Abstract:AI accessibility tools have mostly been designed for individual use, helping one person overcome a specific functional barrier. But for many people with disabilities, complex tasks are accomplished through collaboration with others who bring complementary abilities, not solitary effort. We propose a three-layer framework, Channelling, Coordinating, and Co-Creating, that rethinks AI's role in ability-diverse collaboration: establishing shared informational ground across abilities, mediating workflows between collaborators with different abilities, and contributing as a bounded partner toward shared goals. Grounded in the Ability-Diverse Collaboration framework, grounding theory, and Carlile's 3T framework, it extends the ``agents as remote collaborators'' vision by centring the collaborative, interdependent ways people with disabilities already work.
Tags
Links
- Source: https://arxiv.org/abs/2603.26252v1
- Canonical: https://arxiv.org/abs/2603.26252v1
Trouble viewing inline? Open PDF directly →
Full Text
32,277 characters extracted from source content.
Expand or collapse full text
Channelling, Coordinating, Collaborating: A Three-Layer Framework for Disability-Centered Human-Agent Collaboration Lan Xiao Global Disability Innovation Hub UCL Interaction Centre University College London London, United Kingdom l.xiao.22@ucl.ac.uk Catherine Holloway Global Disability Innovation Hub UCL Interaction Centre University College London London, United Kingdom c.holloway@ucl.ac.uk Figure 1: Disability-Centered Human-Agent Collaboration Framework Abstract AI accessibility tools have mostly been designed for individual use, helping one person overcome a specific functional barrier. But for many people with disabilities, complex tasks are accomplished through collaboration with others who bring complementary abili- ties, not solitary effort. We propose a three-layer framework, Chan- nelling, Coordinating, and Co-Creating, that rethinks AI’s role in ability-diverse collaboration: establishing shared informational Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than the author(s) must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from permissions@acm.org. CHI’26, April 13–17, 2026, Barcelona, Spain © 2026 Copyright held by the owner/author(s). Publication rights licensed to ACM. ACM ISBN 978-1-4503-X-X/18/06 https://doi.org/X.X ground across abilities, mediating workflows between collabora- tors with different abilities, and contributing as a bounded partner toward shared goals. Grounded in the Ability-Diverse Collabora- tion framework, grounding theory, and Carlile’s 3T framework, it extends the “agents as remote collaborators” vision by centring the collaborative, interdependent ways people with disabilities already work. CCS Concepts • Human-centered computing→Accessibility theory, con- cepts and paradigms. Keywords human-agent collaboration, disability, CSCW, accessibility, co-agency, AI coordination arXiv:2603.26252v1 [cs.HC] 27 Mar 2026 CHI’26, April 13–17, 2026, Barcelona, SpainXiao et al. ACM Reference Format: Lan Xiao and Catherine Holloway. 2026. Channelling, Coordinating, Col- laborating: A Three-Layer Framework for Disability-Centered Human- Agent Collaboration. In Proceedings of 2026 CHI Conference on Human Fac- tors in Computing Systems (CHI’26). ACM, New York, NY, USA, 5 pages. https://doi.org/X.X 1 Introduction People with disabilities are increasingly using LLM-based tools across their everyday workflows, from information access [1] and day-to-day decision support [23] to end-to-end content creation [26]. However, AI accessibility tools and much of the surrounding HCI research remain focused on individual assistance: compen- sating for a specific functional limitation through a one-person, one-tool interaction model. Disability scholarship has long argued that this framing is in- complete. Bennett et al. [6] argue that interdependence, not inde- pendence alone, better captures how people with disabilities navi- gate daily life, and should sit at the centre of assistive technology research. Xiao et al. map recurring patterns in ability-diverse collab- oration, showing that collaboration is itself the mechanism through which access gets produced [25]. This collaborative reality is evi- dent across settings: blind runners coordinate with sighted guides through continuous, fine-grained communication that weaves to- gether voice cues and bodily movement [5], and creators with sensory impairments split the work of video production across multiple stages with trusted partners, from filming through editing and publishing [26]. For many people with disabilities, collabora- tion is the ordinary infrastructure behind complex professional and creative work. Human-agent collaboration research offers relevant foundations for understanding this collaborative reality, with frameworks for trust, coordination, and shared agency [2,15,18,28] building on grounding theory [10], workspace awareness [14,16], and knowl- edge boundary management [9]. Yet these frameworks have not engaged with ability-diverse settings, and carry assumptions that need rethinking when ability differences are present. In this position paper, we propose a three-layer framework that bridges accessibility and human-agent collaboration. Layer 1, Channelling, concerns modality-adapted, equivalent access to task-relevant information. Layer 2, Coordinating, concerns how AI can mediate workflows, communication, and handoffs among collaborators with different abilities. Layer 3, Co-Creating, moves beyond support to consider AI as a bounded contributor that ad- vances shared goals alongside human partners. Grounded in the Ability-Diverse Collaboration framework [25], grounding theory [10], and Carlile’s 3T framework [9], together these layers extend both fields by centring interdependence in people with disabili- ties’ everyday practices and by treating triadic collaboration as an explicit design target. 2 Background and Positioning Ability-diverse collaboration, where collaborators bring fundamen- tally different abilities to shared work, differs from typical collabora- tion in two key ways [25]. First, collaborators may not enjoy equal access to information, nor possess congruent knowledge concern- ing the content and process of collaboration. Second, this inherent asymmetry in abilities and information access can engender distinct roles and potentially divergent goals. These two features challenge foundational assumptions in the theories most commonly used to explain collaboration. Clark and Brennan’s [10] theory of grounding describes how collaborators establish mutual understanding through a collective process shaped by the communication medium’s constraints: cop- resence, visibility, audibility, cotemporality, and others. A central principle is least collaborative effort: participants minimise the to- tal work needed to reach mutual understanding. In ability-diverse collaboration, however, these constraints are not properties of the medium alone but of the person-medium coupling. A video call affords visibility, but not for a blind collaborator. A voice channel affords audibility, but not for a Deaf collaborator. The first feature of ability-diverse collaboration, unequal information access, means that grounding costs become fundamentally asymmetric. The least- effort path for the group may concentrate information gatekeeping in one party, efficient in aggregate but constraining for the person with a disability. Carlile’s [9] 3T framework describes three progressively com- plex knowledge boundaries: syntactic (a shared lexicon suffices), semantic (interpretive differences require translation), and prag- matic (conflicting interests mean knowledge is “at stake” and must be transformed). The second feature of ability-diverse collabora- tion, that asymmetry creates distinct roles and divergent goals, maps directly onto these pragmatic boundaries. The knowledge at stake is not only domain-specific expertise but ability-specific: the embodied, situated understanding of how to perceive, move, and work. When a collaboration must be renegotiated because abilities or roles change, this knowledge carries real costs, and mismatches between the boundary type and the process used produce costly failures. Human-agent collaboration research has been exploring agents as collaborative partners, building on grounding theory and workspace awareness [14,16], and developing frameworks for trust, coordina- tion, and shared agency [2,18,28]. This creates a natural opportu- nity: AI could address both features of ability-diverse collaboration, reducing information asymmetry and mediating the distinct roles that ability differences create. However, current frameworks pre- sume shared perceptual access to the same evidence, a common modality for presenting awareness information, and a dyadic struc- ture. In ability-diverse settings, AI typically enters an existing part- nership, creating a triad with fundamentally different coordination demands. Addressing these challenges requires a framework that accounts for both: how AI can reconfigure the grounding condi- tions so that information access is no longer dependent on a single collaborator, and how AI can enter the negotiation of roles and goals without overriding the agency of the people involved. 3 The Three-Layer Framework We propose three layers of agent involvement in ability-diverse collaboration, Channelling, Coordinating, and Co-Creating, each progressively deeper (Figure 1). The layers are nested: Coordinating depends on Channelling, and Co-Creating depends on both. This nesting parallels Carlile’s [9] progression from syntactic to semantic to pragmatic boundaries: each layer requires not just more support, Channelling, Coordinating, Collaborating: A Three-Layer Framework for Disability-Centered Human-Agent CollaborationCHI’26, April 13–17, 2026, Barcelona, Spain but a qualitatively different kind of shared knowledge. We ground each layer in the ADC technology categories and illustrate with examples across disability types and domains. 3.1 Layer 1: Channelling In ability-diverse collaboration, information asymmetry is often the root barrier. A Deaf collaborator in a design team cannot hear verbal feedback during a critique session. A collaborator using switch- based input cannot scan a dense project dashboard at the pace their team assumes. A neurodivergent team member may receive the same meeting notes as everyone else but find the unstructured format unusable for extracting priorities. This asymmetry creates unequal costs: some collaborators must invest significantly more effort to access the same information, rely more heavily on partners to interpret or relay it, or miss information altogether. Over time, these imbalances can constrain agency and reinforce dependency on others. Research on accessible collaborative writing [12] and collaboration awareness in document editing [19] has begun to address fragments of this problem, but typically for a single user rather than a collaborative pair. At this layer, AI functions as an ability supporter, providing equivalent, modality-adapted information to all collaborators. This does not mean identical information, but functionally equivalent access sufficient to support informed decision-making. Real-time captioning can give a Deaf team member access to rapid-fire design discussion. Structured meeting summaries with prioritised actions can give a collaborator with executive function difficulties a usable overview of what was decided. AI-generated descriptions can give a blind collaborator independent access to visual workspace state. In Clark and Brennan’s [10] terms, this layer reconfigures the grounding conditions so that the costs of establishing mutual un- derstanding no longer fall disproportionately on one party. The key shift is from delegation to informed direction. When a collaborator has independent access to task-relevant information, the collab- oration dynamic changes: rather than asking “what did I miss?”, they can direct “let us revisit that decision.” However, establishing shared information does not resolve how to coordinate around it. For that, the collaboration must move to the next layer. 3.2 Layer 2: Coordinating Even when all collaborators can access the same content, ability- diverse collaboration requires substantial coordination work: nego- tiating roles, managing handoffs, tracking each other’s activities, and resolving differences in judgment. This coordination work is often invisible and burdensome, what Branham and Kane [8] call the “invisible work of accessibility” in mixed-ability workplaces. At this layer, AI functions as a communication supporter, medi- ating workflow between collaborators with different abilities. This means maintaining task state across ability boundaries: providing awareness cues adapted to each collaborator’s sensory and cogni- tive modality, tracking energy-aware task distribution in a team where a member with chronic fatigue works alongside non-disabled colleagues, or supporting structured handoffs when collaborators share a single device, a common setup in low-resource contexts [4,26]. Prior work on paired remote sighted assistance [27] shows that even human-human coordination in these settings is effortful; AI-mediated coordination could reduce this burden across disability types. This layer addresses workspace awareness [16] in ability-diverse settings, where the standard assumption that all collaborators can monitor the same shared workspace does not hold. In Carlile’s [9] terms, it develops semantic capacity: not just transmitting informa- tion, but creating shared meanings around what differences in pace, process, and ability mean for how the team works together. The shift is from ad hoc verbal negotiation to structured, AI-mediated workflow continuity, reducing the invisible labour that falls dispro- portionately on people with disabilities and their partners. However, when differences generate not just interpretive challenges but con- flicting interests, when changing one’s practice carries real costs, a different process is needed. 3.3 Layer 3: Co-Creating In Channelling and Coordinating, the collaboration model remains ability sharing: the primary goal-holder directs, and both human collaborator and AI serve supporting roles. Co-Creating marks a shift toward ability combining, in which AI contributes as a third participant with distinct capabilities directed toward a shared goal. At this layer, AI generates, suggests, and contributes. In col- laborative music-making between musicians with different motor abilities, AI could generate rhythmic accompaniment patterns that complement each performer’s physical playing style. In a work- place team where a member with ADHD struggles to synthesise discussions into structured plans, AI could draft proposals that the whole team iterates on, contributing executive function support as a shared capability rather than an individual accommodation. In design prototyping where a collaborator with limited hand mobility works alongside a non-disabled partner, AI-driven fabrication tools could execute fine motor tasks under the disabled collaborator’s direction. Emerging work on human-AI collaboration in remote sighted assistance [20,29] explores related configurations, though not yet in co-creative contexts. The critical design requirement is that AI contributions are pro- posals, not decisions: transparent, modifiable, and clearly distin- guished from human-authored work. In Carlile’s [9] terms, this layer faces a pragmatic boundary: AI’s entry introduces novelty that creates new dependencies and potentially conflicting interests. What is at stake is deeply personal: one’s way of working, one’s creative authority, one’s professional identity. Research with peo- ple with disabilities consistently shows that they maintain creative control even in heavily collaborative workflows [26]. Each new tri- adic configuration requires its own iterative process of negotiating roles and authority, not the assumption that what worked before will transfer. Extending the ADC notion of “ability provider” to a non-human agent is a deliberate theoretical move: AI does not share abilities in the same relational, embodied sense that human collaborators do [7]. But the functional role, contributing a required capability toward a shared goal, maps onto the ability combining model, and naming it as such allows us to reason about triadic configurations that current CSCW frameworks do not address. CHI’26, April 13–17, 2026, Barcelona, SpainXiao et al. 4 Research Agenda The framework opens questions at the intersection of accessibility and human-agent collaboration. We highlight four that we believe are especially pressing. What counts as equivalent Channelling across different abilities? A text description of a video frame conveys different things than seeing it. A structured meeting summary conveys dif- ferent things than the live discussion. How should AI calibrate detail level and format, and who controls that calibration? How do cultural and linguistic contexts shape what counts as adequate access? How does reducing information asymmetry reshape col- laboration? If Channelling gives all collaborators equivalent access, do established roles shift? Does a collaborator who can now in- dependently access previously gatekept information take on new tasks, or do existing divisions of labour persist? What new tensions emerge? How is agency negotiated in triadic configurations? When does AI shift from supporting collaboration to transforming it? How do collaborators negotiate creative ownership when AI has gener- ated content that both humans refine? This question is especially sharp for people with disabilities who have fought to maintain their creative authority. Research on algorithm aversion [13] and com- plementary team performance [3] suggests that trust calibration in human-AI teams is already complex; ability-diverse settings add further dimensions. What extensions to CSCW theory are needed? The “remote collaborator” metaphor is productive, but a collaborator that per- ceives through multiple modalities simultaneously, never tires, and has no lived experience of disability is a fundamentally different kind of partner. How do common ground and workspace aware- ness need rethinking for triadic, ability-diverse settings [10,16], and what new concepts might be required? How should the notion of knowledge “at stake” [9] be extended when what is at stake is not domain expertise but embodied, ability-specific ways of working? Parasuraman et al.’s taxonomy [22] of automation levels may offer a complementary lens for specifying AI’s degree of involvement across the three layers. 5 Discussion The framework makes two claims. First, accessibility should be understood as collaborative infrastructure, not only individual as- sistive capability. When access is generated through collaboration, as empirical research consistently demonstrates [6,24,25], AI sys- tems must be designed to facilitate that collaboration rather than merely to address individual deficiencies. Clark and Brennan’s [10] grounding theory helps make this concrete: accessibility barriers are, at their core, asymmetries in the costs of establishing mutual understanding. Channelling addresses these asymmetries, but Co- ordinating and Co-Creating recognise that equitable access alone is not enough; the coordination structures and creative roles built around those asymmetries must also be renegotiated. Second, the maturity of AI in disability contexts should be eval- uated through socio-technical outcomes: role transparency, re- pairability of breakdowns, and fair distribution of coordination labour, not only individual assistive task performance. Carlile’s [9] framework adds a specific warning here: when the process used does not match the type of boundary faced, failures surface downstream. Evaluating AI in ability-diverse collaboration must therefore attend to whether the system supports the right kind of shared knowledge at each layer, not just whether it provides accurate outputs. These claims carry practical implications. System designers should make collaboration states visible: roles, intentions, uncertainties, handoff statuses, and escalation pathways. Based on current tools that deal with parts of this problem [11,12,19], we argue that future systems should surface collaboration metadata: who contributed what, what the AI is uncertain about, and where human judge- ment is needed, in ways that are appropriate for each participant’s abilities and modalities. Another implication has to do with governance. If triadic co- agency is a goal for design, organisations need policies and inter- faces that make it easy to negotiate and contest AI’s role. Without these, advanced AI may replicate paternalistic patterns using a novel lexicon, a concern frequently articulated by disability schol- ars regarding assistive technologies [17, 21]. We note several limitations. The framework is conceptual; the layers need to be tested through design research and user studies with people with disabilities who collaborate with AI systems. The mapping between our three layers and Carlile’s three boundaries is productive but not exact: ability-diverse collaboration may present boundary configurations that do not fit neatly into the syntactic- semantic-pragmatic progression, particularly when multiple disabil- ity types are present. The framework has not yet addressed how its layers interact with power asymmetries beyond ability differences (e.g., employer-employee, expert-novice). 6 Conclusion Human-agent collaboration research has made real progress on trust, coordination, and shared agency, but has largely overlooked a basic fact: for many people with disabilities, collaboration is not oc- casional but structural, the ordinary infrastructure through which complex work gets done. Accessibility research understands this deeply but has not yet engaged with AI agents as collaborative partners. The question that connects both fields is deceptively sim- ple: what happens when an AI agent joins a collaboration already shaped by difference, interdependence, and hard-won agency? Our three-layer framework proposes that AI’s involvement is a progression, not a single relationship. Channelling establishes shared informational ground across abilities, reconfiguring the grounding conditions so that mutual understanding no longer de- pends on a single gatekeeper. Coordinating mediates handoffs and reduces the invisible coordination labour that ability-diverse part- nerships absorb, building shared meanings across different ways of working. Co-Creating contributes as a bounded participant whose proposals human collaborators evaluate and refine on their own terms, navigating the pragmatic boundaries where roles, interests, and creative authority must be negotiated. The framework is conceptual and needs empirical grounding through design research with people with disabilities who collabo- rate with AI systems, engagement with power asymmetries beyond Channelling, Coordinating, Collaborating: A Three-Layer Framework for Disability-Centered Human-Agent CollaborationCHI’26, April 13–17, 2026, Barcelona, Spain ability differences, and extension across a wider range of disabil- ity experiences and professional domains. But we believe its core move, centring interdependence rather than individual deficit, and treating triadic ability-diverse collaboration as an explicit design target, opens productive ground for both fields to build on. References [1]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 (ASSETS ’24). Association for Computing Machinery, New York, NY, USA, 1–14. https://doi.org/10.1145/3663548.3675631 [2]Saleema Amershi, Dan Weld, Mihaela Vorvoreanu, Adam Fourney, Besmira Nushi, Penny Collisson, Jina Suh, Shamsi Iqbal, Paul N. Bennett, Kori Inkpen, Jaime Teevan, Ruth Kikin-Gil, and Eric Horvitz. 2019. Guidelines for Human- AI Interaction. In Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems (CHI ’19). Association for Computing Machinery, New York, NY, USA, 1–13. https://doi.org/10.1145/3290605.3300233 [3]Gagan Bansal, Tongshuang Wu, Joyce Zhou, Raymond Fok, Besmira Nushi, Ece Kamar, Marco Tulio Ribeiro, and Daniel Weld. 2021. Does the Whole Exceed its Parts? The Effect of AI Explanations on Complementary Team Performance. In Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems (CHI ’21). Association for Computing Machinery, New York, NY, USA, 1–16. https://doi.org/10.1145/3411764.3445717 [4]Giulia Barbareschi, Catherine Holloway, Katherine Arnold, Grace Magomere, Wycliffe Ambeyi Wetende, Gabriel Ngare, and Joyce Olenja. 2020. The Social Network: How People with Visual Impairment use Mobile Phones in Kibera, Kenya. In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems (CHI ’20). Association for Computing Machinery, New York, NY, USA, 1–15. https://doi.org/10.1145/3313831.3376658 [5] Giulia Barbareschi, Tarika Kumar, Christopher Changmok Kim, George Chernyshov, and Kai Kunze. 2024. "Speech is Silver, Silence is Golden " An- alyzing Micro-communication Strategies between Visually Impaired Runners and their Guides. Proceedings of the CHI Conference on Human Factors in Com- puting Systems 12 (2024), 1–16. https://doi.org/10.1145/3613904.3642388 [6]Cynthia L. Bennett, Erin Brady, and Stacy M. Branham. 2018. Interdependence as a Frame for Assistive Technology Research and Design. In Proceedings of the 20th International ACM SIGACCESS Conference on Computers and Accessibility. ACM, Galway Ireland, 161–173. https://doi.org/10.1145/3234695.3236348 [7] Cynthia L. Bennett, Daniela K. Rosner, and Alex S. Taylor. 2020. The Care Work of Access. In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems. ACM, Honolulu HI USA, 1–15. https://doi.org/10.1145/3313831.3376568 [8]Stacy M. Branham and Shaun K. Kane. 2015. The Invisible Work of Accessibility: How Blind Employees Manage Accessibility in Mixed-Ability Workplaces. In Proceedings of the 17th International ACM SIGACCESS Conference on Computers & Accessibility - ASSETS ’15. ACM Press, Lisbon, Portugal, 163–171.https: //doi.org/10.1145/2700648.2809864 [9]Paul R. Carlile. 2004. Transferring, Translating, and Transforming: An Integrative Framework for Managing Knowledge Across Boundaries. Organization Science 15, 5 (Oct. 2004), 555–568. https://doi.org/10.1287/orsc.1040.0094 [10]Herbert H. Clark and Susan E. Brennan. 1991. Grounding in communication. In Perspectives on socially shared cognition. American Psychological Association, Washington, DC, US, 127–149. https://doi.org/10.1037/10096-006 [11]Maitraye Das, Darren Gergle, and Anne Marie Piper. 2023. Simphony: Enhancing Accessible Pattern Design Practices among Blind Weavers. In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems. ACM, Hamburg Germany, 1–19. https://doi.org/10.1145/3544548.3581047 [12]Maitraye Das, Thomas Barlow McHugh, Anne Marie Piper, and Darren Gergle. 2022. Co11ab: Augmenting Accessibility in Synchronous Collaborative Writing for People with Vision Impairments. In CHI Conference on Human Factors in Computing Systems. ACM, New Orleans LA USA, 1–18. https://doi.org/10.1145/ 3491102.3501918 [13]Berkeley J. Dietvorst, Joseph P. Simmons, and Cade Massey. 2015. Algorithm aversion: People erroneously avoid algorithms after seeing them err. Journal of Experimental Psychology: General 144, 1 (2015), 114–126. https://doi.org/10.1037/ xge0000033 Place: US. [14]Paul Dourish and Victoria Bellotti. 1992. Awareness and coordination in shared workspaces. In Proceedings of the 1992 ACM conference on Computer-supported cooperative work (CSCW ’92). Association for Computing Machinery, New York, NY, USA, 107–114. https://doi.org/10.1145/143457.143468 [15]Jonathan Grudin. 2018. From Tool to Partner: The Evolution of Human-Computer Interaction. In Extended Abstracts of the 2018 CHI Conference on Human Factors in Computing Systems (CHI EA ’18). Association for Computing Machinery, New York, NY, USA, 1–3. https://doi.org/10.1145/3170427.3170663 [16]Carl Gutwin and Saul Greenberg. 2002. A Descriptive Framework of Workspace Awareness for Real-Time Groupware. Computer Supported Cooperative Work (CSCW) 11, 3 (Sept. 2002), 411–446. https://doi.org/10.1023/A:1021271517844 [17] Megan Hofmann, Devva Kasnitz, Jennifer Mankoff, and Cynthia L Bennett. 2020. Living Disability Theory: Reflections on Access, Research, and Design. The 22nd International ACM SIGACCESS Conference on Computers and Accessibility (2020), 1–13. https://doi.org/10.1145/3373625.3416996 [18]Eric Horvitz. 1999. Principles of mixed-initiative user interfaces. In Proceedings of the SIGCHI conference on Human Factors in Computing Systems (CHI ’99). Association for Computing Machinery, New York, NY, USA, 159–166. https: //doi.org/10.1145/302979.303030 [19]Cheuk Yin Phipson Lee, Zhuohao Zhang, Jaylin Herskovitz, JooYoung Seo, and Anhong Guo. 2022. CollabAlly: Accessible Collaboration Awareness in Document Editing. In CHI Conference on Human Factors in Computing Systems. ACM, New Orleans LA USA, 1–17. https://doi.org/10.1145/3491102.3517635 [20]Sooyeon Lee, Rui Yu, Jingyi Xie, Syed Masum Billah, and John M. Carroll. 2022. Opportunities for Human-AI Collaboration in Remote Sighted Assistance. In Proceedings of the 27th International Conference on Intelligent User Interfaces (IUI ’22). Association for Computing Machinery, New York, NY, USA, 63–78. https://doi.org/10.1145/3490099.3511113 [21]Jennifer Mankoff, Gillian R. Hayes, and Devva Kasnitz. 2010. Disability studies as a source of critical inquiry for the field of assistive technology. In Proceedings of the 12th international ACM SIGACCESS conference on Computers and accessibility (ASSETS ’10). Association for Computing Machinery, New York, NY, USA, 3–10. https://doi.org/10.1145/1878803.1878807 [22] R. Parasuraman, T.B. Sheridan, and C.D. Wickens. 2000. A model for types and levels of human interaction with automation. IEEE Transactions on Systems, Man, and Cybernetics - Part A: Systems and Humans 30, 3 (May 2000), 286–297. https://doi.org/10.1109/3468.844354 [23] 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 (CHI ’25). Association for Computing Machinery, New York, NY, USA, 1–17. https://doi.org/10.1145/3706598.3713433 [24] Emily Q. Wang and Anne Marie Piper. 2018. Accessibility in Action: Co-Located Collaboration among Deaf and Hearing Professionals. Proceedings of the ACM on Human-Computer Interaction 2, CSCW (Nov. 2018), 1–25. https://doi.org/10. 1145/3274449 [25] Lan Xiao, Maryam Bandukda, Katrin Angerbauer, Weiyue Lin, Tigmanshu Bhat- nagar, Michael Sedlmair, and Catherine Holloway. 2024. A Systematic Review of Ability-diverse Collaboration through Ability-based Lens in HCI. Proceed- ings of the CHI Conference on Human Factors in Computing Systems (2024), 1–21. https://doi.org/10.1145/3613904.3641930 [26] Lan Xiao, Maryam Bandukda, Franklin Mingzhe Li, Mark Colley, and Catherine Holloway. 2025. Understanding the Video Content Creation Journey of Creators with Sensory Impairment in Kenya. In Proceedings of the 27th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS ’25). Association for Computing Machinery, New York, NY, USA, 1–14. https://doi.org/10.1145/ 3663547.3746356 [27]Jingyi Xie, Rui Yu, Kaiming Cui, Sooyeon Lee, John M. Carroll, and Syed Masum Billah. 2023. Are Two Heads Better than One? Investigating Remote Sighted Assistance with Paired Volunteers. In Proceedings of the 2023 ACM Designing Interactive Systems Conference (DIS ’23). Association for Computing Machinery, New York, NY, USA, 1810–1825. https://doi.org/10.1145/3563657.3596019 [28]Bingsheng Yao, Chaoran Chen, April Yi Wang, Sherry Tongshuang Wu, Toby Jia-jun Li, and Dakuo Wang. 2026. From Human-Human Collaboration to Human- Agent Collaboration: A Vision, Design Philosophy, and an Empirical Framework for Achieving Successful Partnerships Between Humans and LLM Agents. https: //doi.org/10.48550/arXiv.2602.05987 arXiv:2602.05987 [cs]. [29]Rui Yu, Sooyeon Lee, Jingyi Xie, Syed Masum Billah, and John M. Carroll. 2024. Human–AI Collaboration for Remote Sighted Assistance: Perspectives from the LLM Era. Future Internet 16, 7 (July 2024), 254. https://doi.org/10.3390/fi16070254