pith. sign in

arxiv: 2605.01610 · v1 · submitted 2026-05-02 · 💻 cs.HC · cs.AI

Less Interaction But More Explanation: A Communication Perspective on Agentic AI Interfaces

Eunchae Jang , S. Shyam Sundar This is my paper

Pith reviewed 2026-05-09 17:36 UTC · model grok-4.3

classification 💻 cs.HC cs.AI
keywords agentic AIhuman-AI interactionexplanationstrustcommunication perspectiveuser agencyoversightcustomization
0
0 comments X

The pith

Agentic AI reduces routine user interactions but demands more explanatory communication for oversight and sustained trust.

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

The paper argues that agentic AI systems, which proactively execute workflows on behalf of users rather than merely responding, alter the dynamics of human-AI communication. While this reduces the need for constant back-and-forth exchanges, it increases the requirement for targeted communication that allows users to oversee actions and understand decisions. Drawing on a communication perspective, the authors examine how users interpret the AI as either the source of actions or simply a channel through which actions occur, and how this interpretation influences trust. They propose that agentic AI must incorporate three specific types of explanations and offer customization options so that rising AI autonomy does not erode human agency.

Core claim

Agentic AI prioritizes workflow execution rather than conversation, which means users need less routine interaction but more communication for oversight and explanation. Because the system can occupy multiple communicative roles, users may struggle to determine whether the AI originates actions or merely transmits them, complicating trust. To mitigate risks, agentic AI should deliver action-process explanations that detail how decisions are reached, uncertainty explanations that convey reliability, and coordination explanations that clarify interactions with other agents or systems, while also providing customization affordances that let users choose when and which explanations appear.

What carries the argument

The source-versus-channel distinction in how users perceive AI communicative roles, which drives the need for action-process, uncertainty, and coordination explanations plus user-controlled customization affordances.

If this is right

  • Routine conversational turns with AI will decrease as proactive execution takes over.
  • Trust will hinge on users correctly identifying whether the AI originates actions or serves as a conduit.
  • Action-process explanations will let users trace decision steps and maintain oversight.
  • Uncertainty explanations will help users assess when to intervene in AI-driven workflows.
  • Coordination explanations will clarify how multiple agents or systems align on tasks.
  • Customization affordances will allow users to adjust explanation volume and timing to retain a sense of control.

Where Pith is reading between the lines

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

  • Designers of personal assistants or workflow tools could apply the source-channel lens to decide which explanation type to surface first in different contexts.
  • Without built-in customization, even well-explained agentic systems might still reduce users' feeling of agency over time.
  • The same communication-role framework might extend to multi-agent setups where users must distinguish among several autonomous systems.

Load-bearing premise

Users' perception of the AI as source or channel of actions will shape trust in a way that specifically requires the three proposed explanation types and customization affordances.

What would settle it

A user study measuring trust and perceived agency after interacting with an agentic system that provides none of the three explanation types versus one that provides all three plus customization options, checking whether the predicted differences in oversight needs and trust emerge.

read the original abstract

AI systems have long been expected to interact with users, answering questions, generating content, and continuing (social) conversations. Agentic AI, however, breaks from this expectation, as its primary objective is workflow execution on behalf of the users. If a system becomes more agentic, do users need less interaction with the system? Our answer is: less routine back-and-forth, but more communication for oversight and explanation, as agentic AI proactively acts, not just responds. Grounded in a communication perspective, we discuss how users perceive the communicative roles of AI systems (whether as the source of actions or merely a channel), and how this can shape trust. Because agentic AI can play multiple communicative roles, it can complicate this source perception and introduce potential risks. To address this, we propose three types of explanations that agentic AI needs to incorporate (action-process, uncertainty, and coordination), and suggest that customization affordances that allow users to decide when and which explanations they see may be key to preserving human agency as AI autonomy increases.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

0 major / 1 minor

Summary. The paper claims that as AI systems become more agentic—proactively executing workflows rather than merely responding—users require less routine interaction but more explanatory communication for oversight. Grounded in a communication perspective, it explores how users' perceptions of AI as the source versus channel of actions affect trust, notes complications from multiple roles, and proposes three explanation types (action-process, uncertainty, and coordination) plus customization affordances to preserve human agency.

Significance. This conceptual contribution could be significant for the HCI community by providing a theoretical basis for designing trustworthy agentic AI interfaces. It integrates communication theory to address trust and agency issues, offering specific design suggestions that may influence future interface development, although empirical studies would be needed to confirm the proposed mechanisms.

minor comments (1)
  1. The discussion of source and channel perceptions would benefit from a brief recap of the underlying communication theory to make the argument more accessible to a broader HCI audience.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive and constructive review, which accurately captures the core argument of our conceptual paper. The recommendation for minor revision is appreciated, and we will incorporate a brief note acknowledging the value of future empirical validation of the proposed mechanisms.

Circularity Check

0 steps flagged

No significant circularity in conceptual framework

full rationale

The paper is a purely conceptual discussion that applies established communication theory (source vs. channel perceptions and their effect on trust) to the distinction between responsive and proactive/agentic AI behavior. The central claim—that greater agency reduces routine interaction but increases the need for oversight-oriented explanations—follows directly from that premise without any equations, fitted parameters, or self-referential definitions. The three proposed explanation types and customization affordances are offered as design implications rather than derived necessities that loop back to the inputs. No load-bearing self-citations reduce the argument to unverified prior claims by the same authors; the derivation remains self-contained against external communication concepts.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on domain assumptions from communication theory applied to AI without new empirical grounding or formalization.

axioms (2)
  • domain assumption Users perceive AI systems as either the source of actions or merely a communication channel, and this perception shapes trust.
    Invoked in the discussion of communicative roles and trust formation.
  • domain assumption Providing action-process, uncertainty, and coordination explanations will address risks from multiple communicative roles.
    Basis for the proposed explanation types.

pith-pipeline@v0.9.0 · 5482 in / 1197 out tokens · 32644 ms · 2026-05-09T17:36:29.428343+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

21 extracted references · 21 canonical work pages

  1. [1]

    Deepak Bhaskar Acharya, Karthigeyan Kuppan, and B Divya. 2025. Agentic AI: Autonomous intelligence for complex goals—A comprehensive survey.IEEe Access13 (2025), 18912–18936

    work page 2025

  2. [2]

    Vijay Arya, Rachel KE Bellamy, Pin-Yu Chen, Amit Dhurandhar, Michael Hind, Samuel C Hoffman, Stephanie Houde, Q Vera Liao, Ronny Luss, Aleksandra Mojsilović, et al . 2019. One explanation does not fit all: A toolkit and taxonomy of ai explainability techniques.arXiv preprint arXiv:1909.03012(2019)

    work page arXiv 2019

  3. [3]

    Mohamed Amine Chatti, Mouadh Guesmi, and Arham Muslim. 2024. Visualization for recommendation explainability: A survey and new perspectives. ACM Transactions on Interactive Intelligent Systems14, 3 (2024), 1–40

    work page 2024

  4. [4]

    Cheng Chen, Eunchae Jang, and S Shyam Sundar. 2025. Racial Bias in AI Training Data: Do Laypersons Notice?Media Psychology(2025), 1–28

    work page 2025

  5. [5]

    Cheng Chen, Mengqi Liao, and S Shyam Sundar. 2024. When to explain? Exploring the effects of explanation timing on user perceptions and trust in AI systems. InProceedings of the Second International Symposium on Trustworthy Autonomous Systems. 1–17

    work page 2024

  6. [6]

    Serena Chen, Kimberly Duckworth, and Shelly Chaiken. 1999. Motivated heuristic and systematic processing.Psychological inquiry10, 1 (1999), 44–49

    work page 1999

  7. [7]

    Marco Dehnert and Paul A Mongeau. 2022. Persuasion in the age of artificial intelligence (AI): Theories and complications of AI-based persuasion. Human Communication Research48, 3 (2022), 386–403

    work page 2022

  8. [8]

    Hangzhi Guo, Pranav Narayanan Venkit, Eunchae Jang, Mukund Srinath, Wenbo Zhang, Bonam Mingole, Vipul Gupta, Kush R Varshney, S Shyam Sundar, and Amulya Yadav. 2025. Exposing AI Bias by Crowdsourcing: Democratizing Critique of Large Language Models. InProceedings of the 6 Jang & Sundar AAAI/ACM Conference on AI, Ethics, and Society, Vol. 8. 1169–1180

    work page 2025

  9. [9]

    Soodeh Hosseini and Hossein Seilani. 2025. The role of agentic ai in shaping a smart future: A systematic review.Array26 (2025), 100399

    work page 2025

  10. [10]

    Yongnam Jung, Cheng Chen, Eunchae Jang, and S Shyam Sundar. 2024. Do we trust chatgpt as much as google search and Wikipedia?. InExtended Abstracts of the CHI Conference on Human Factors in Computing Systems. 1–9. doi:10.1145/3613905.3650862

    work page doi:10.1145/3613905.3650862 2024

  11. [11]

    Emma Kallina and Jatinder Singh. 2025. Towards Responsible AI: Explainable AI as a Catalyst for Participatory Development Practices. In Human-centered Explainable AI Workshop (HCXAI) @ CHI 2025. Zenodo. doi:10.5281/zenodo.15170408 Workshop paper presented at HCXAI @ CHI 2025

    work page doi:10.5281/zenodo.15170408 2025

  12. [12]

    Hui Min Lee, Peixin Hua, Rehab Alayoubi, and S Shyam Sundar. 2025. Shorter and Simpler: Customizing Generative AI Responses Can Increase Satisfaction, Credibility, and Fact-Checking Intentions. InProceedings of the Extended Abstracts of the CHI Conference on Human Factors in Computing Systems. 1–7

    work page 2025

  13. [13]

    Hui Min Lee, Davis Yadav, Sangwook Lee, Keerthana Govindarazan, Cheng Chen, and S Shyam Sundar. 2025. While We Wait... How Users Perceive Waiting Times and Generation Cues during AI Image Generation. InProceedings of the Extended Abstracts of the CHI Conference on Human Factors in Computing Systems. 1–8

    work page 2025

  14. [14]

    Q Vera Liao and S Shyam Sundar. 2022. Designing for responsible trust in AI systems: A communication perspective. InProceedings of the 2022 ACM conference on fairness, accountability, and transparency. 1257–1268

    work page 2022

  15. [15]

    Mohammad Naiseh, Dena Al-Thani, Nan Jiang, and Raian Ali. 2023. How the different explanation classes impact trust calibration: The case of clinical decision support systems.International Journal of Human-Computer Studies169 (2023), 102941

    work page 2023

  16. [16]

    P Jonathon Phillips, P Jonathon Phillips, Carina A Hahn, Peter C Fontana, Amy N Yates, Kristen Greene, David A Broniatowski, and Mark A Przybocki. 2021. Four principles of explainable artificial intelligence. (2021)

    work page 2021

  17. [17]

    Ranjan Sapkota, Konstantinos I Roumeliotis, and Manoj Karkee. 2025. Ai agents vs. agentic ai: A conceptual taxonomy, applications and challenges. Information Fusion(2025), 103599

    work page 2025

  18. [18]

    Yuan Sun, Eunchae Jang, Fenglong Ma, and Ting Wang. 2024. Generative AI in the wild: Prospects, challenges, and strategies. InProceedings of the 2024 chi conference on human factors in computing systems. 1–16

    work page 2024

  19. [19]

    2008.The MAIN model: A heuristic approach to understanding technology effects on credibility

    S Shyam Sundar et al. 2008.The MAIN model: A heuristic approach to understanding technology effects on credibility. MacArthur Foundation Digital Media and Learning Initiative Cambridge, MA

    work page 2008

  20. [20]

    S Shyam Sundar and Eun-Ju Lee. 2022. Rethinking communication in the era of artificial intelligence.Human Communication Research48, 3 (2022), 379–385

    work page 2022

  21. [21]

    S Shyam Sundar and Clifford Nass. 2000. Source orientation in human-computer interaction: Programmer, networker, or independent social actor. Communication research27, 6 (2000), 683–703

    work page 2000