Pith. sign in

REVIEW 2 major objections 5 minor 81 references

From Conversation to Orchestration: HCI Challenges and Opportunities in Interactive Multi-Agentic Systems

T0 review · 2 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Pith's one-line read When AI becomes a team, the user becomes a composer, and HCI must answer six design challenges.

desk verdict A timely, clearly written agenda paper that recontextualizes familiar HCI challenges for hierarchical multi-agentic systems; the six challenges are expert-selected, but the paper is honest about that and the agenda is worth taking seriously. read the letter →

arxiv 2506.20091 v1 pith:TB64WXZ4 submitted 2025-06-25 cs.HC

classification cs.HC
keywords multi-agenticsystemshierarchicalarchitecturehuman-agentinteractionHCIorchestrationtransparencytrustdesignchallenges
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper argues that when AI assistants become teams of specialized agents, the human-computer interaction problem changes in kind, not just in scale. Focusing on hierarchical multi-agentic systems, in which a supervisor or orchestrator agent mediates between the user and a set of subagents, it identifies six design challenges: opaque agent teamwork, parallel execution, emergent complexity, conflict resolution, mental-model formation, and trust and explainability. The paper presents these as a research agenda with illustrative examples and preliminary design considerations, such as orchestration panels, agent group cards, and interrupt mechanisms. If the claim is right, designers should stop treating multi-agentic systems as extensions of single-chatbot interfaces and instead build interfaces around delegation, oversight, and conflict resolution.

What carries the argument

The central object is the hierarchical supervisor-subagent architecture, in which an orchestrator agent sits between the user and specialised subagents, translating goals, delegating subtasks, and monitoring progress. This architecture is the load-bearing structure of the paper's argument: the authors justify focusing on it because it is the common default in current frameworks such as AutoGen and OpenAI Agents and because it is supposed to reduce cognitive load by hiding distributed complexity. The argument's machinery is the six-challenge taxonomy, which turns that architecture into concrete interface-design problems, together with proposed scaffolding ideas — orchestration panels, organigram-style hierarchy visualisation, agent and group cards, roundtable conflict views, and interrupt mechanisms.

What would settle it

A survey of deployed end-user-facing multi-agentic systems finding that most use decentralized, multi-body, or peer-to-peer architectures would undercut the paper's focus; so would a controlled experiment in which users interacting through a supervisor agent show equal or higher cognitive load and lower situation awareness than users delegating directly to specialised agents.

Watch

Extended reading notes

Core claim

The paper's central claim is that the hierarchical architecture, where a top-level orchestrator agent is the user's gateway to specialised subagents, fundamentally changes the user's role from direct operator to what the authors call 'the composer' of an agentic team. From that shift, the paper derives six design challenges: reducing opaqueness in agentic teamwork, interacting in parallel, designing for emergent complexity, resolving conflicts, understanding multi-agentic systems, and navigating trust and explainability. Each challenge is illustrated with a marketplace scenario and paired with design considerations and open questions; the paper positions this set as a starting research agenda for human-multi-agent interaction rather than a finished set of solutions.

Load-bearing premise

The paper's load-bearing premise is that the hierarchical supervisor/subagent architecture is and will remain the dominant form of multi-agentic systems that end-users encounter, so design work should be organised around it.

Editorial extensions

If this is right

  • Chat-only interfaces will be insufficient: users need high-level orchestration views plus optional low-level access to agent roles, constraints, and inter-agent messages.
  • Interfaces must represent parallel execution explicitly, with mechanisms to pause, prioritise, or interrupt a subset of agents without stalling the whole system.
  • Systems should visualise agent hierarchies and architecture changes, and provide debug, intervene, and guide controls to address cascading failures.
  • Trust design must move from a single-agent explainability model to layered trust: users may trust the orchestrator while subagents fail, so explanations and recovery mechanisms must work at multiple levels.
  • Researchers should develop new evaluation instruments for collective reliability, delegation confidence, and emotional responses to agent teams, since traditional usability testing does not capture asynchronous emergent multi-agent behaviour.

Reading between the lines

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

  • Beyond the paper's scope, the six challenges likely apply to decentralized multi-agent architectures as well, but in inverted form: without a supervisor gateway, opacity shifts to direct many-to-many interaction, making cognitive load the first-order problem.
  • A testable extension is a controlled comparison of agent 'group cards' versus individual 'agent cards' for mental-model formation and trust calibration, since the paper proposes both but does not test either.
  • If the supervisor-agent abstraction fails to deliver the assumed cognitive-load reduction, the design recommendations tied to hierarchy would need reworking; measuring workload under hierarchical versus direct delegation would settle this.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 5 minor

Summary. This position paper argues that the shift from single-agent chatbots to multi-agentic systems, specifically the hierarchical architecture with a supervisor or orchestrator agent, introduces a distinct set of end-user-facing design challenges. Drawing on a review of current frameworks (AutoGen, CrewAI, LangGraph, OpenAI Agents, MetaGPT, Magentic-One) and the authors' expertise, the paper identifies six design challenges: reducing opaqueness, interacting in parallel, designing for emergent complexity, resolving conflicts, understanding multi-agentic systems, and navigating trust and explainability. For each challenge it offers illustrative scenarios, design considerations, and open research questions. The paper also outlines opportunities for HCI researchers in visualization, control/delegation, user experience and ethics, and design infrastructure and research methods. The central claim is that these six challenges constitute a research agenda that can drive future work in Human-Multi-Agent interaction.

Significance. If the proposed agenda is taken as a starting point rather than a final taxonomy, the paper makes a timely and useful contribution. Its strengths are its clear focus on the end-user perspective, its consistent use of a marketplace scenario to ground abstract challenges, and its connection of multi-agentic interaction to established HCI concepts such as mental models, trust calibration, and meaningful human control. The paper also appropriately acknowledges that literature and use cases are still sparse, and it builds on current developer tools and frameworks, which gives the challenges some empirical grounding. The main value is as a discussion piece that could help structure early research in human interaction with multi-agentic systems. However, the contribution's load-bearing element, the set of six challenges, is presented without a transparent selection protocol, which limits the claim that these are the overarching challenges worthy of a research agenda.

major comments (2)
  1. [Sections 1 and 3] The central contribution is a set of six design challenges presented as a research agenda, but the paper does not describe any systematic method for their identification. The introduction states that the authors "build on this knowledge and our expertise" (Section 1), and Section 3 introduces "our identified six design challenges" without reporting a literature search, selection criteria, coding scheme, or any other derivation protocol. Section 3.3 hedges that the challenges are "not as a comprehensive list," but this hedge is local to emergent complexity, whereas the abstract and Section 1 describe the set as "overarching challenges" that can "drive future research." Because the six challenges determine the scope of all later design considerations, the absence of a transparent method is a load-bearing issue: a reader cannot tell whether a challenge was deliberately excluded or simply overlooked, and the design considerations in Sections 3.1-3.6 inherit that hidden selection. I recommend either describing a systematic analysis process or explicitly reframing the contribution as a non-exhaustive set of plausible challenges grounded in the authors' expertise and current developer frameworks.
  2. [Section 1, architectural scope] The paper's focus on hierarchical architecture is justified by two claims: that it reduces cognitive burden and that it is "emerging as a popular architecture" or "often proposed as the default structure in current frameworks." The second claim is supported only by reference [70] (AutoGen) and a single example of an end-user interface (Magentic-One [26]). AutoGen is a developer framework, and its popularity among developers does not by itself demonstrate that hierarchical structures dominate end-user-facing multi-agentic systems. Since all six challenges are derived specifically for the hierarchical architecture, this assumption is load-bearing. I suggest strengthening this motivation with additional evidence of adoption (e.g., a survey of current frameworks, documentation, or examples of deployed systems) or softening the claim to state that hierarchical architecture is one common and illustrative architecture, not necessarily the dominant one.
minor comments (5)
  1. [Section 2.1 / Figure 1] Figure 1 is described as illustrating four architectures, but the four labels (a)-(d) should be checked to ensure they match the order and naming used in the text (single-agent, multi-body, hierarchical, decentralised/organisational).
  2. [Table 1] Table 1 lists Magentic-One as targeting "Prompter & developer," but Section 2.2 describes Magentic-One as a ready-to-use team, with Magentic-UI being the end-user interface; please align the table row more precisely with the text or clarify the distinction between the team and its UI.
  3. [Section 2.2] The phrase "on canva" should read "on a canvas," and the description of AutoGen Studio could be tightened for clarity (e.g., "drag-and-drop representations" and "node graph visualisation" are both mentioned, and the relationship between them is not fully clear).
  4. [Section 3.2] The reference to Magentic-UI is given as a blog post [51]; consider also citing the technical report on Magentic-One [26] so readers can locate the system description in a peer-reviewed or archival source.
  5. [Section 2.1] The definition of "agentic system" is introduced via Chan et al. [18] but the term "agentic" is used frequently before that; consider defining it early in Section 1 or at the first use to avoid ambiguity.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper is a qualitative research agenda whose challenges are explicitly presented as expert-identified and non-exhaustive, with no fitted inputs, predictions, or load-bearing self-citation.

full rationale

This is a position/vision paper in HCI, not a derivation or empirical prediction paper. There are no equations, no fitted parameters, no benchmark evaluations, and no quantities that are 'predicted' from data. The central contribution is a set of six design challenges for end-user interaction with hierarchical multi-agentic systems. The authors state that they 'build on this knowledge and our expertise, and use the characteristics of the current tools already available for developers to identify a set of design challenges' (Section 1), and Section 3.3 explicitly cautions that the challenges are 'not as a comprehensive list, but to initiate conversation within HCI and to guide future work in this space.' Thus the challenges are presented as an expert synthesis, not as a derived, exhaustive, or falsifiable result, and the paper's own framing prevents the agenda from being circular in a self-definitional sense. The paper does contain several self-citations (e.g., Schömbs et al. 2023, 2024a, 2024b, 2024c; Zhang et al. 2025), but none is load-bearing for the central claim: they are used as supporting examples in literature reviews (e.g., anthropomorphism in HRI, uncertainty visualisation, ROSAnnotator as an agentic system example), not as a 'uniqueness theorem' or as the sole justification for the six challenges. No argument reduces to its own input: identifying challenges by expert judgement may raise questions of comprehensiveness or selection bias, but that is a methodological limitation, not circular reasoning. The paper does not rename a known result, does not smuggle in an ansatz via citation, and does not fit a parameter and then call it a prediction. A search-based audit of how the six challenges relate to documented HCI issues would be a useful validity check, but its absence is a correctness-risk concern, not evidence of circularity. Therefore, the appropriate circularity score is 0.

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

No free parameters or invented entities. The analysis rests on several domain assumptions about the prevalence of hierarchical architectures and end-user roles, which the paper itself partly acknowledges.

assumptions (5)
  • domain assumption The hierarchical architecture is emerging as a popular or default structure in current multi-agentic frameworks (AutoGen, OpenAI Agents, Magentic-One).
    Used to justify focusing on hierarchical architectures (Section 1); if decentralized or multi-body architectures dominate instead, the challenge set may be misaligned.
  • domain assumption Users will act as end-users ('composers') interacting with multi-agentic systems at a high level, rather than as developers.
    Section 3, the 'composer' framing assumes end-users delegate and supervise rather than micro-manage; this assumption drives the entire challenge list.
  • domain assumption LLM-based agents can communicate, exchange information, and exhibit emergent behaviours that end-users must navigate.
    The paper's definition of multi-agentic systems (Section 2.1) relies on this; if agents are merely parallel LLM calls without emergent interaction, the challenges (DC2, DC3, DC4) would be less relevant.
  • domain assumption A supervisor/orchestrator agent reduces the user's cognitive burden by abstracting complexity.
    Section 1 states this as a key rationale for focusing on hierarchical architectures; it is cited from prior work (e.g., Magentic-One) but not empirically tested.
  • domain assumption The end-user will carry responsibilities spanning the developer-user spectrum (configuring, deploying, supervising agents).
    Section 3.6 explicitly acknowledges this as 'the inherent assumption that end-users bear the burden of operating across the developer-user spectrum'.

how reviews work

0 comments
Cite this review

Pith. "Pith review of From Conversation to Orchestration: HCI Challenges and Opportunities in Interactive Multi-Agentic Systems." pith.science (2026). https://pith.science/paper/TB64WXZ4

@misc{pith2026250620091,
  author       = {Pith},
  title        = {Pith review of: From Conversation to Orchestration: HCI Challenges and Opportunities in Interactive Multi-Agentic Systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TB64WXZ4}},
  note         = {Machine review of arXiv:2506.20091}
}
read the original abstract

Recent advances in multi-agentic systems (e.g. AutoGen, OpenAI Swarm) allow users to interact with a group of specialised AI agents rather than a single general-purpose agent. Despite the promise of this new paradigm, the HCI community has yet to fully examine the opportunities, risks, and user-centred challenges it introduces. We contribute to research on multi-agentic systems by exploring their architectures and key features through a human-centred lens. While literature and use cases remain limited, we build on existing tools and frameworks available to developers to identify a set of overarching challenges, e.g. orchestration and conflict resolution, that can guide future research in HCI. We illustrate these challenges through examples, offer potential design considerations, and provide research opportunities to spark interdisciplinary conversation. Our work lays the groundwork for future exploration and offers a research agenda focused on user-centred design in multi-agentic systems.

Figures

Figures reproduced from arXiv: 2506.20091 by the authors.

Figure 1
Figure 1. This figure illustrates four agentic system architec [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

81 extracted references · 42 canonical work pages

  1. [70]

    White, Doug Burger, and Chi Wang

    Qingyun Wu, Gagan Bansal, Jieyu Zhang, Yiran Wu, Beibin Li, Erkang Zhu, Li Jiang, Xiaoyun Zhang, Shaokun Zhang, Jiale Liu, Ahmed Hassan Awadallah, Ryen W. White, Doug Burger, and Chi Wang. 2023. AutoGen: Enabling Next-Gen LLM Appli- cations via Multi-Agent Conversation. https://doi.org/10.48550/arXiv.2308.08155 arXiv:2308.08155 [cs]

  2. [26]

    Adam Fourney, Gagan Bansal, Hussein Mozannar, Cheng Tan, Eduardo Salinas, Erkang, Zhu, Friederike Niedtner, Grace Proebsting, Griffin Bassman, Jack Gerrits, Jacob Alber, Peter Chang, Ricky Loynd, Robert West, Victor Dibia, Ahmed Awadallah, Ece Kamar, Rafah Hosn, and Saleema Amershi. 2024. Magentic-One: A Generalist Multi-Agent System for Solving Complex T...

  3. [1]

    [n. d.]. LangGraph. https://langchain-ai.github.io/langgraph/

  4. [2]

    Agentforce: Create Powerful AI Agents

    2025. Agentforce: Create Powerful AI Agents. Salesforce website. https://www.salesforce.com/au/agentforce/

  5. [3]

    langchain-ai/langgraph-studio

    2025. langchain-ai/langgraph-studio. https://github.com/langchain- ai/langgraph-studio original-date: 2024-07-29T22:11:00Z

  6. [4]

    AgentOps . 2025. AgentOps. https://github.com/AgentOps-AI/agentops original-date: 2023-08-15T23:26:23Z

  7. [5]

    CrewAI . [n. d.]. CrewAI. https://www.crewai.com/

  8. [6]

    CrewAI-Studio . 2025. CrewAI-Studio. https://github.com/strnad/CrewAI-Studio original-date: 2024-05-30T08:15:40Z

Show all 81 references
  1. [7]

    OpenAI-Agent . 2025. openai-agent. https://github.com/openai/openai-agents- python/tree/main

  2. [8]

    Gavin Abercrombie, Amanda Cercas Curry, Tanvi Dinkar, Verena Rieser, and Zeerak Talat. 2023. Mirages: On Anthropomorphism in Dialogue Systems. https://doi.org/10.48550/arXiv.2305.09800 arXiv:2305.09800 [cs]

  3. [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. https://doi.org/10.48550/arXiv.2412.10380 arXiv:2412.10380 [cs]

  4. [10]

    Joseph Bates et al. 1994. The role of emotion in believable agents.Commun. ACM 37, 7 (1994), 122–125

  5. [11]

    Alexandra Bejarano and Tom Williams. 2023. No name, no voice, less trust: Robot group identity performance, entitativity, and trust distribution. In 2023 32nd IEEE International Conference on Robot and Human Interactive Communication (RO-MAN). IEEE, 1339–1346

  6. [12]

    Wünderlich, and Christian Brock

    Markus Blut, Cheng Wang, Nancy V. Wünderlich, and Christian Brock. 2021. Understanding anthropomorphism in service provision: a meta-analysis of physical robots, chatbots, and other AI. Journal of the Academy of Marketing Science 49, 4 (July 2021), 632–658. https://doi.org/10....

  7. [13]

    Stephen Brade, Bryan Wang, Mauricio Sousa, Sageev Oore, and Tovi Gross- man. 2023. Promptify: Text-to-image generation through interactive prompt exploration with large language models. InProceedings of the 36th Annual ACM Symposium on User Interface Software and Technology. 1–14

  8. [14]

    Zana Buçinca, Maja Barbara Malaya, and Krzysztof Z. Gajos. 2021. To Trust or to Think: Cognitive Forcing Functions Can Reduce Overreliance on AI in AI-assisted Decision-making. Proceedings of the ACM on Human-Computer Interaction 5, CSCW1 (April 2021), 1–21. https://doi.org/10...

  9. [15]

    Cristiano Castelfranchi. 1994. Guarantees for autonomy in cognitive agent architecture. In International Workshop on Agent Theories, Architectures, and Languages. Springer, 56–70

  10. [16]

    Jonker, Jeroen Van Den Hoven, Deborah Forster, and Reginald L

    Luciano Cavalcante Siebert, Maria Luce Lupetti, Evgeni Aizenberg, Niek Beckers, Arkady Zgonnikov, Herman Veluwenkamp, David Abbink, Elisa Giaccardi, Geert-Jan Houben, Catholijn M. Jonker, Jeroen Van Den Hoven, Deborah Forster, and Reginald L. Lagendijk. 2023. Meaningful human ...

  11. [18]

    Alan Chan, Rebecca Salganik, Alva Markelius, Chris Pang, Nitarshan Rajkumar, Dmitrii Krasheninnikov, Lauro Langosco, Zhonghao He, Yawen Duan, Micah Carroll, Michelle Lin, Alex Mayhew, Katherine Collins, Maryam Molamoham- madi, John Burden, Wanru Zhao, Shalaleh Rismani, Konstan...

  12. [19]

    Olanubi, Joseph M

    Michelle Cohn, Mahima Pushkarna, Gbolahan O. Olanubi, Joseph M. Moran, Daniel Padgett, Zion Mengesha, and Courtney Heldreth. 2024. Believing Anthropomorphism: Examining the Role of Anthropomorphic Cues on Trust in Large Language Models. InExtended Abstracts of the CHI Conferen...

  13. [20]

    Hai Dang, Sven Goller, Florian Lehmann, and Daniel Buschek. 2023. Choice over control: How users write with large language models using diegetic and non-diegetic prompting. In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems. 1–17

  14. [21]

    Kate Darling. 2015. ’Who’s Johnny?’ Anthropomorphic Framing in Human-Robot Interaction, Integration, and Policy. https://doi.org/10.2139/ssrn.2588669

  15. [22]

    Ameet Deshpande, Tanmay Rajpurohit, Karthik Narasimhan, and Ash- win Kalyan. 2023. Anthropomorphization of AI: Opportunities and Risks. https://doi.org/10.48550/arXiv.2305.14784 arXiv:2305.14784 [cs]

  16. [23]

    Victor Dibia, Jingya Chen, Gagan Bansal, Suff Syed, Adam Fourney, Erkang Zhu, Chi Wang, and Saleema Amershi. 2024. AUTOGEN STUDIO: A No-Code Developer Tool for Building and Debugging Multi-Agent Systems. InProceedings of the 2024 Conference on Empirical Methods in Natural Lang...

  17. [24]

    Ali Dorri, Salil S Kanhere, and Raja Jurdak. 2018. Multi-agent systems: A survey. Ieee Access 6 (2018), 28573–28593

  18. [25]

    Wen Duan, Naomi Yamashita, Yoshinari Shirai, and Susan R Fussell. 2021. Bridging fluency disparity between native and nonnative speakers in multilingual multiparty collaboration using a clarification agent. Proceedings of the ACM on Human-Computer Interaction 5, CSCW2 (2021), 1–31

  19. [27]

    Adam Fourney, Gagan Bansal, Hussein Mozannar, Cheng Tan, Eduardo Salinas, Friederike Niedtner, Grace Proebsting, Griffin Bassman, Jack Gerrits, Jacob Alber, et al. 2024. Magentic-one: A generalist multi-agent system for solving complex tasks. arXiv preprint arXiv:2411.04468 (2024)

  20. [28]

    Yao Fu, Hao Peng, Ashish Sabharwal, Peter Clark, and Tushar Khot. 2022. Complexity-based prompting for multi-step reasoning. In The Eleventh International Conference on Learning Representations

  21. [29]

    Dawei Gao, Zitao Li, Xuchen Pan, Weirui Kuang, Zhijian Ma, Bingchen Qian, Fei Wei, Wenhao Zhang, Yuexiang Xie, Daoyuan Chen, et al. 2024. Agentscope: A flexible yet robust multi-agent platform.arXiv preprint arXiv:2402.14034 (2024)

  22. [30]

    Jie Gao, Simret Araya Gebreegziabher, Kenny Tsu Wei Choo, Toby Jia-Jun Li, Simon Tangi Perrault, and Thomas W Malone. 2024. A Taxonomy for Human-LLM Interaction Modes: An Initial Exploration. In Extended Abstracts of the CHI Conference acronym ’XX, June 03–05, 2018, Woodstock,...

  23. [32]

    Donna S Haverkamp and Susan Gauch. 1998. Intelligent information agents: review and challenges for distributed information sources.Journal of the American society for information science 49, 4 (1998), 304–311

  24. [33]

    Sirui Hong, Xiawu Zheng, Jonathan Chen, Yuheng Cheng, Jinlin Wang, Ceyao Zhang, Zili Wang, Steven Ka Shing Yau, Zijuan Lin, Liyang Zhou, et al . 2023. Metagpt: Meta programming for multi-agent collaborative framework. arXiv preprint arXiv:2308.00352 (2023)

  25. [34]

    Bowen Jiang, Yangxinyu Xie, Xiaomeng Wang, Weijie J Su, Camillo Jose Taylor, and Tanwi Mallick. 2024. Multi-modal and multi-agent systems meet rationality: A survey. InICML 2024 Workshop on LLMs and Cognition

  26. [35]

    Ellen Jiang, Edwin Toh, Alejandra Molina, Kristen Olson, Claire Kayacik, Aaron Donsbach, Carrie J Cai, and Michael Terry. 2022. Discovering the syntax and strate- gies of natural language programming with generative language models. InPro- ceedings of the 2022 CHI Conference o...

  27. [36]

    Tae Soo Kim, Yoonjoo Lee, Minsuk Chang, and Juho Kim. 2023. Cells, generators, and lenses: Design framework for object-oriented interaction with large language models. In Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology. 1–18

  28. [37]

    Hannah Rose Kirk, Bertie Vidgen, Paul Röttger, and Scott A. Hale. 2024. The benefits, risks and bounds of personalizing the alignment of large language models to individuals. Nature Machine Intelligence 6, 4 (April 2024), 383–392. https://doi.org/10.1038/s42256-024-00820-y

  29. [38]

    Harsh Kumar, Yiyi Wang, Jiakai Shi, Ilya Musabirov, Norman AS Farb, and Joseph Jay Williams. 2023. Exploring the use of large language models for improving the awareness of mindfulness. InExtended Abstracts of the 2023 CHI Conference on Human Factors in Computing Systems. 1–7

  30. [39]

    John D. Lee. 2001. Emerging challenges in cognitive ergonomics: Managing swarms of self-organizing agent-based automation. Theoretical Issues in Ergonomics Science 2, 3 (Jan. 2001), 238–250. https://doi.org/10.1080/14639220110104925 Publisher: Taylor & Francis _eprint: https:/...

  31. [40]

    Lee and Katrina A

    John D. Lee and Katrina A. See. 2004. Trust in Automation: Design- ing for Appropriate Reliance. Human Factors 46, 1 (March 2004), 50–80. https://doi.org/10.1518/hfes.46.1.50_30392 Publisher: SAGE Publications Inc

  32. [41]

    Bran, Sam Cox, Oliver Schilter, Carlo Baldassari, Andrew D White, and Philippe Schwaller

    Andres M. Bran, Sam Cox, Oliver Schilter, Carlo Baldassari, Andrew D White, and Philippe Schwaller. 2024. Augmenting large language models with chemistry tools. Nature Machine Intelligence (2024), 1–11

  33. [42]

    Pattie Maes. 1995. Agents that reduce work and information overload. InReadings in human–computer interaction. Elsevier, 811–821

  34. [43]

    Maja J. Matarić. 1995. Issues and approaches in the design of collective autonomous agents. Robotics and Autonomous Systems16, 2-4 (Dec. 1995), 321–331. https://doi.org/10.1016/0921-8890(95)00053-4

  35. [44]

    Dan Milmo. 2023. ChatGPT reaches 100 million users two months after launch. The Guardian (Feb. 2023). https://www.theguardian.com/technology/2023/feb/ 02/chatgpt-100-million-users-open-ai-fastest-growing-app

  36. [45]

    Margaret Mitchell, Simone Wu, Andrew Zaldivar, Parker Barnes, Lucy Vasserman, Ben Hutchinson, Elena Spitzer, Inioluwa Deborah Raji, and Timnit Gebru. 2019. Model Cards for Model Reporting. InProceedings of the Conference on Fairness, Ac- countability, and Transparency (FAT* ’1...

  37. [46]

    Odell, H

    James J. Odell, H. Van Dyke Parunak, and Mitchell Fleischer. 2003. The Role of Roles in Designing Effective Agent Organizations. InSoftware Engineering for Large-Scale Multi-Agent Systems, Gerhard Goos, Juris Hartmanis, Jan Van Leeuwen, Alessandro Garcia, Carlos Lucena, Franco...

  38. [47]

    Eugenio Oliveira, Klaus Fischer, and Olga Stepankova. 1999. Multi-agent systems: which research for which applications. Robotics and Autonomous Systems 27, 1-2 (1999), 91–106

  39. [48]

    Jun Ota. 2006. Multi-agent robot systems as distributed autonomous systems. Advanced engineering informatics 20, 1 (2006), 59–70

  40. [49]

    Saumya Pareek, Eduardo Velloso, and Jorge Goncalves. 2024. Trust Development and Repair in AI-Assisted Decision-Making during Complementary Expertise. In The 2024 ACM Conference on Fairness, Accountability, and Transparency. ACM, Rio de Janeiro Brazil, 546–561. https://doi.org...

  41. [50]

    Gaëtan Podevijn, Rehan O’grady, Nithin Mathews, Audrey Gilles, Carole Fantini-Hauwel, and Marco Dorigo. 2016. Investigating the effect of increasing robot group sizes on the human psychophysiological state in the context of human–swarm interaction. Swarm Intelligence 10 (2016)...

  42. [51]

    Brenda Potts. 2025. Magentic-UI, an experimental human-centered web agent. https://www.microsoft.com/en-us/research/blog/magentic-ui-an- experimental-human-centered-web-agent/

  43. [52]

    Vera Liao, and Nikola Banovic

    Snehal Prabhudesai, Leyao Yang, Sumit Asthana, Xun Huan, Q. Vera Liao, and Nikola Banovic. 2023. Understanding Uncertainty: How Lay Decision- makers Perceive and Interpret Uncertainty in Human-AI Decision Making. In Proceedings of the 28th International Conference on Intellige...

  44. [53]

    Eileen Roesler. 2023. Anthropomorphic framing and failure comprehensibility influence different facets of trust towards industrial robots.Frontiers in Robotics and AI 10 (Sept. 2023). https://doi.org/10.3389/frobt.2023.1235017 Publisher: Frontiers

  45. [54]

    Roesler, D

    E. Roesler, D. Manzey, and L. Onnasch. 2021. A meta-analysis on the effectiveness of anthropomorphism in human-robot interaction. Science Robotics 6, 58 (Sept. 2021), eabj5425. https://doi.org/10.1126/scirobotics.abj5425 Publisher: American Association for the Advancement of Science

  46. [55]

    Stuart Russell, Peter Norvig, and Artificial Intelligence. 1995. A modern approach. Artificial Intelligence. Prentice-Hall, Egnlewood Cliffs 25, 27 (1995), 79–80

  47. [56]

    Sarah Schömbs, Jacobe Klein, and Eileen Roesler. 2023. Feeling with a robot—the role of anthropomorphism by design and the tendency to anthropomorphize in human-robot interaction. Frontiers in Robotics and AI 10 (June 2023). https://doi.org/10.3389/frobt.2023.1149601 Publisher...

  48. [57]

    Sarah Schömbs, Jiahe Pan, Yan Zhang, Jorge Goncalves, and Wafa Johal. 2024. FaceVis: Exploring a Robot’s Face for Affective Visualisation Design. InExtended Abstracts of the CHI Conference on Human Factors in Computing Systems (CHI EA ’24) . Association for Computing Machinery...

  49. [58]

    Sarah Schömbs, Saumya Pareek, Jorge Goncalves, and Wafa Johal. 2024. Robot- Assisted Decision-Making: Unveiling the Role of Uncertainty Visualisation and Embodiment. In Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems (CHI ’24). Association for Comp...

  50. [59]

    Chuhan Shi, Yicheng Hu, Shenan Wang, Shuai Ma, Chengbo Zheng, Xiaojuan Ma, and Qiong Luo. 2023. RetroLens: A Human-AI Collaborative System for Multi-step Retrosynthetic Route Planning. In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems. 1–20

  51. [60]

    Masahiro Shiomi and Norihiro Hagita. 2019. Do the number of robots and the participant’s gender influence conformity effect from multiple robots?Advanced Robotics 33, 15-16 (2019), 756–763

  52. [61]

    Yoav Shoham. 1993. Agent-oriented programming. Artificial intelligence 60, 1 (1993), 51–92

  53. [62]

    Michael Terry, Chinmay Kulkarni, Martin Wattenberg, Lucas Dixon, and Mered- ith Ringel Morris. 2024. Interactive AI Alignment: Specification, Process, and Evalu- ation Alignment. https://doi.org/10.48550/arXiv.2311.00710 arXiv:2311.00710 [cs]

  54. [63]

    2005.Conflicting Agents: Conflict Management in Multi-Agent Systems

    Cathérine Tessier, Laurent Chaudron, and Heinz-Jürgen Müller. 2005.Conflicting Agents: Conflict Management in Multi-Agent Systems. Springer Science & Business Media. Google-Books-ID: YODlBwAAQBAJ

  55. [64]

    Bernstein, and Ranjay Krishna

    Helena Vasconcelos, Matthew Jörke, Madeleine Grunde-McLaughlin, Tobias Ger- stenberg, Michael S. Bernstein, and Ranjay Krishna. 2023. Explanations Can Reduce Overreliance on AI Systems During Decision-Making.Proc. ACM Hum.-Comput. Interact. 7, CSCW1 (April 2023), 129:1–129:38....

  56. [65]

    Xinru Wang and Ming Yin. 2021. Are Explanations Helpful? A Comparative Study of the Effects of Explanations in AI-Assisted Decision-Making. In 26th International Conference on Intelligent User Interfaces. ACM, College Station TX USA, 318–328. https://doi.org/10.1145/3397481.3450650

  57. [66]

    Xinru Wang and Ming Yin. 2022. Effects of Explanations in AI-Assisted Decision Making: Principles and Comparisons. ACM Transactions on Interactive Intelligent Systems 12, 4 (Dec. 2022), 1–36. https://doi.org/10.1145/3519266

  58. [67]

    Adam Waytz, Joy Heafner, and Nicholas Epley. 2014. The mind in the machine: Anthropomorphism increases trust in an autonomous vehi- cle. Journal of Experimental Social Psychology 52 (May 2014), 113–117. https://doi.org/10.1016/j.jesp.2014.01.005

  59. [68]

    Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, et al. 2022. Chain-of-thought prompting elicits reasoning in large language models. Advances in neural information processing systems 35 (2022), 24824–24837

  60. [69]

    Choon Yue Wong and Gerald Seet. 2017. Workload, awareness and automation in multiple-robot supervision. International Journal of Advanced Robotic Systems 14, 3 (May 2017), 1729881417710463. https://doi.org/10.1177/1729881417710463 Publisher: SAGE Publications

  61. [71]

    Qingyun Wu, Gagan Bansal, Jieyu Zhang, Yiran Wu, Shaokun Zhang, Erkang Zhu, Beibin Li, Li Jiang, Xiaoyun Zhang, and Chi Wang. 2023. Autogen: Enabling next-gen llm applications via multi-agent conversation framework.arXiv preprint arXiv:2308.08155 (2023)

  62. [72]

    Tongshuang Wu, Michael Terry, and Carrie Jun Cai. 2022. Ai chains: Transparent and controllable human-ai interaction by chaining large language model prompts. HCI Challenges and Opportunities in Interactive Multi-Agentic Systems Conference acronym ’XX, June 03–05, 2018, Woodst...

  63. [73]

    Ziang Xiao, Xingdi Yuan, Q Vera Liao, Rania Abdelghani, and Pierre-Yves Oudeyer

  64. [74]

    Ann Yuan, Andy Coenen, Emily Reif, and Daphne Ippolito. 2022. Wordcraft: story writing with large language models. InProceedings of the 27th International Conference on Intelligent User Interfaces. 841–852

  65. [75]

    Edward N Zalta, Uri Nodelman, Colin Allen, and John Perry. 1995. Stanford encyclopedia of philosophy

  66. [76]

    Wenshuo Zang, Mengsha Hu, and Rui Liu. 2024. Large Language Model Driven Interactive Learning for Real-Time Cognitive Load Prediction in Human-Swarm Systems. In 2024 33rd IEEE International Confer- ence on Robot and Human Interactive Communication (ROMAN) . 97–102. https://doi...

  67. [77]

    Andong Zhang and Pei-Luen Patrick Rau. 2023. Tools or peers? Impacts of anthropomorphism level and social role on emotional attachment and disclosure tendency towards intelligent agents. Computers in Human Behavior 138 (Jan. 2023), 107415. https://doi.org/10.1016/j.chb.2022.107415

  68. [78]

    Yan Zhang, Haoqi Li, Ramtin Tabatabaei, and Wafa Johal. 2025. ROSAnnotator: A Web Application for ROSBag Data Analysis in Human-Robot Interaction.arXiv preprint arXiv:2501.07051 (2025)

  69. [79]

    Vera Liao, and Rachel K

    Yunfeng Zhang, Q. Vera Liao, and Rachel K. E. Bellamy. 2020. Effect of confidence and explanation on accuracy and trust calibration in AI-assisted decision making. In Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency (FAT* ’20). Association for C...

  70. [80]

    Chengbo Zheng, Yuheng Wu, Chuhan Shi, Shuai Ma, Jiehui Luo, and Xiaojuan Ma

  71. [81]

    Qingxiaoyang Zhu and Hao-Chuan Wang. 2023. Leveraging Large Language Model as Support for Human Problem Solving: An Exploration of Its Appropri- ation and Impact. InCompanion Publication of the 2023 Conference on Computer Supported Cooperative Work and Social Computing. 333–337

  72. [82]

    InProceedings of the 2023 CHI Conference on Human Factors in Computing Systems

    Competent but Rigid: Identifying the Gap in Empowering AI to Participate Equally in Group Decision-Making. InProceedings of the 2023 CHI Conference on Human Factors in Computing Systems. 1–19

  73. [2023]

    InCompanion proceedings of the 28th international conference on intelligent user interfaces

    Supporting qualitative analysis with large language models: Combining codebook with GPT-3 for deductive coding. InCompanion proceedings of the 28th international conference on intelligent user interfaces. 75–78

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.