Pith. sign in

REVIEW 4 major objections 6 minor 50 references

Observable Social Life Spaces: Exploring User Interpretations of agent-side life context in human-agent interaction

T0 review · 4 major / 6 minor · reviewed 2026-08-02 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Only visible social lives, not hidden memories, raise users' sense of equality with agents.

desk verdict Interesting design variable, but the experiment confounds visual observability with a real-time influence channel, so the necessity claim goes too far. read the letter →

arxiv 2603.21505 v2 pith:KYLGCYZY submitted 2026-03-23 cs.HC

Zihong He , Shuqin Wang , Songchen Zhou , Qinghui Lin , Jialin Wang , Chen Liang , Hai-Ning Liang This is my paper
classification cs.HC
keywords human-agentinteractionperceivedequalityobservablesociallifespacesgenerativeagentsdual-trackmemoryvisualobservabilityrelationalframingmixed-methodsstudy
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

Most AI assistants are treated as tools because users only ever see command-and-response. This paper claims that the missing ingredient for perceived equality is visual access to an agent's own social life: an observable world where the agent works, moves, and talks with other agents. In a three-condition experiment (Baseline, Unobservable dual-track memory, Observable life space), only the Observable condition significantly raised users' perceived equality (p=0.015). The finding suggests designers should render an agent's off-command existence rather than just giving it memories of one. The authors caution that the quantitative effect is modest and should be read cautiously at the participant level.

What carries the argument

The 'Observable Social Life Spaces' paradigm: a continuous 2D virtual environment where LLM-driven agents move via A* path planning, initiate dialogues on spatial proximity, and accumulate a dual-track memory (an interaction track with the user and a social-life-space track with their autonomous events). The load-bearing mechanism is the rendering step that turns the hidden memory track into visible, real-time animation; the dual-track memory alone is the control condition that does not move equality.

What would settle it

Re-run the study with an Observable condition that displays the exact same life-space feed but removes interactive affordances (user cannot influence or address the agent inside the view). If perceived equality no longer exceeds baseline, the effect is due to interactivity or agency, not visual observability; if it does, the claim is supported. Alternatively, log and compare agent action distributions across conditions to confirm behavioral equivalence.

Watch

Extended reading notes

Core claim

The central discovery is that visual observability, not mere possession of an autonomous memory, is the threshold that shifts users' relational stance. Agents with a hidden dual-track memory of an independent life scored no better than baseline on perceived equality; only when the same life was rendered as a visible continuous environment did equality ratings rise. Qualitatively, 6/8 participants in the Observable condition spontaneously framed the agent as a peer or friend without prompting, versus 2/8 baseline and 5/8 unobservable. The authors interpret this as evidence that seeing the agent's life substantiates its independent existence and disrupts the master-servant frame.

Load-bearing premise

The paper assumes the Observable and Unobservable conditions differ only in visual access, but the Observable condition also let users speak to agents inside the visualization and see their suggestions change the agent's behavior in real time, and no check verified the two conditions produced identical agent behavior.

Editorial extensions

If this is right

  • Showing an agent's autonomous life in the interface, rather than only storing it in memory, is what changes users' sense of equality.
  • Agents that keep an independent, visible life outside user commands can shift interaction from instrumental service toward partnership.
  • Designers of relational agents should budget for visual rendering of off-command activity, not just memory architectures.
  • Users with relational needs respond more strongly to visible life contexts than users with instrumental needs, suggesting personalization of observability.
  • The dual-track memory design provides a controlled way to test observability effects, useful for future human-agent interaction studies.

Reading between the lines

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

  • A direct implication the authors leave implicit: if 'life space' visibility is what triggers peer framing, then even simple idle animations or activity logs might suffice in low-fidelity settings, without full sandbox simulation.
  • The study's short sessions cannot tell whether the equality boost decays with familiarity; a longitudinal test would show whether visible life sustains perceived equality over repeated use.
  • Because the Observable condition also let users influence the agent's behavior in real time, a cleaner test would separate watching from interacting; the paper's claim about observability may actually bundle in agency.
  • The 5-cm circular touchscreen and chef persona may prime service hierarchies; testing with peer-like identities (e.g., fellow traveler) could push equality scores further.
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

4 major / 6 minor

Summary. The paper introduces the Observable Social Life Spaces (OSLS) paradigm, in which an LLM-driven agent inhabits a continuously rendered virtual world with daily routines and social relationships, alongside a dual-track memory mechanism that separates user-interaction memories from the agent's autonomous life events. The authors report a between-subjects study (N=24, n=8 per condition) comparing a Baseline (interaction-only), an Unobservable condition (dual-track memory but no visual access), and an Observable condition (same memory architecture plus visual access to the agent's life space). The central claim is that only when the agent's social life space is visually observable does perceived equality during interaction significantly increase (ANOVA p=0.015; pairwise Observable vs Baseline p=0.006, Observable vs Unobservable p=0.031, Baseline vs Unobservable n.s.). Qualitative thematic analysis supports the idea that visibility helps users treat the agent as a peer rather than a tool. The abstract explicitly cautions that participant-level analysis suggests a conservative interpretation, but this analysis is not reported in the results.

Significance. The proposed OSLS architecture is a concrete, implementable design that connects autonomous social simulation with a conversational interface, and the study addresses an underexplored question in HAI: whether visualizing an agent's independent life can shift users' relational stance from instrumental to equal. If the central claim were well supported, the work would offer a practical design guideline for relational agents and digital companions. The paper also contributes a clearly specified dual-track memory design, a mixed-methods instrument, and a set of application scenarios. However, the quantitative evidence for the 'visual observability is the necessary threshold' claim is currently undermined by a confound in the experimental design and by statistical reporting gaps. The qualitative data are suggestive but do not by themselves establish the causal role of visual observability.

major comments (4)
  1. [§3.3, §4.1] The Observable condition does not differ from Unobservable only in rendering. Section 3.3 states that users can speak to agents inside the visualization and, if the agent accepts a user-suggested action, 'behavior planning is triggered immediately, allowing the user's influence to be visible in real time.' This is an interactive influence channel that is absent in the Unobservable condition. The assertion in §4.1 that 'identical dual-track memory generation' was maintained is therefore not credible: user-suggested actions alter the agent's behavior and hence the Social Life Space Track memories. Without a manipulation check or equivalence test showing that the two conditions were behaviorally matched, the significant Equality effect cannot be attributed to visual observability alone; it may reflect visible user agency or co-construction.
  2. [§5.1] The post-hoc t-tests are uncorrected for multiple comparisons. With three pairwise tests, the Observable–Unobservable contrast (p=0.031) does not survive a Bonferroni correction (adjusted alpha ≈ 0.017). Moreover, no effect sizes, standard deviations, or confidence intervals are reported, and the 'only when visually observable' claim rests on accepting the null Baseline–Unobservable difference with n=8 per condition—a very low-powered test. The paper should report corrected post-hocs, effect sizes, and ideally a mixed-effects model with participant as a random effect to support the central claim.
  3. [Abstract vs. §5.1] The abstract cautions that 'participant-level analysis suggests that the quantitative effect should be interpreted cautiously,' but Section 5.1 does not present any participant-level analysis (e.g., mixed-effects regression or a random-intercept model). If such analysis was performed, it must be reported so readers can judge whether the condition effect survives when participant-level variance is modeled. This is especially important because the study uses repeated measures (Turn 1 and Turn 2) and between-subjects groups of n=8.
  4. [§5.2.2, Table 4] The qualitative theme 'Perceived mutual influence' directly evidences the confound raised in §3.3. Participants in the Observable condition explicitly say 'Because of your influence, something may change in its world' (P3). This confirms that Observable users experienced an interactive agency that Unobservable users did not, further undermining the claim that the observed effect is purely due to visual observability. The authors should either redesign the experiment to isolate visual access (e.g., a fourth condition with observation but no influence) or substantially soften the claim to 'observable interactive life space.'
minor comments (6)
  1. [§5.1, Tables 1–2] Report standard deviations and effect sizes (e.g., partial eta-squared) for all means. The current tables provide only group means, making it impossible to assess variability.
  2. [§4.1] The ANOVA summary is incomplete: 'F(2, N) = 4.62' should give the actual denominator degrees of freedom and the error term (e.g., between-subject error, within-subject error). Also state whether the mixed-design ANOVA included Gender and Turn as factors in the same model and how the Equality dependent variable was derived from the questionnaire.
  3. [§4.2] The perceived-equality scale is described as 'adapted from the HALIE framework' with an additional dimension, but no items or reliability information are provided. Include the adapted items and any psychometric properties (e.g., Cronbach's alpha).
  4. [Abstract] There is a typo: 'ObservableLife Spaces' is missing a space.
  5. [§6.2] The limitations section should explicitly acknowledge the observability/influence confound identified above, not just the short-term nature of the study.
  6. [§5.1] The phrase 'no significant difference was found between the Baseline and Unobservable conditions' is used to claim that unobservable life space 'did not suffice.' With n=8 per condition, this is an absence of evidence, not evidence of absence. Please rephrase and temper the inference.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: empirical user study; the central result is measured, not derived from fitted inputs or self-citations.

full rationale

This paper is an empirical perception study, not a derivation chain. There are no fitted parameters, no normalizations, and no equations whose output is predetermined by their inputs. The central claim—that visual observability of an agent's life space enhances perceived equality (p=0.015)—rests on user ratings collected across conditions; the ratings are not a logical consequence of the system design. The 'social factor equality' principles in Section 3.1 are design inspirations, not mathematical constraints that force the questionnaire responses. No load-bearing self-citation appears: prior work on generative agents (Park et al.) and HALIE (Lee et al.) are external to the authors, and no 'uniqueness theorem' is invoked. The main validity concern is a potential confound: Section 3.3 states that in the visualized mode 'if the agent accepts user-suggested actions... behavior planning is triggered immediately, allowing the user's influence to be visible in real time,' while Section 4.1 claims backend equivalence was 'strictly held... constant' without reporting a manipulation check. This could mean the Observable condition differed not only in visual observability but also in an interactive agency channel. However, a confound is a threat to internal/construct validity, not circularity: the equality ratings could have come out insignificant, and they are not identical to the manipulation by construction. The abstract's participant-level caution and the absence of a reported participant-level mixed-effects analysis are reporting/statistical concerns, not circular steps. Given the rubric, the honest finding is no significant circularity.

Assumptions & free parameters 3 free parameters · 4 assumptions · 1 invented entities

This is an empirical HCI study, not a mathematical derivation, so the central claim rests on experimental-control and measurement assumptions rather than free parameters fitted to equations. The hand-chosen design constants (thresholds, persona, LLM choices) affect the stimulus but are not fitted values in the statistical model; they are listed because they are required for replication.

free parameters (3)
  • Spatial proximity threshold for agent social trigger = not reported
    Section 3.3 says dialogues are initiated when proximity criteria are met, but the actual threshold is unspecified; it shapes the agent's social life content.
  • Long-term memory summarization threshold = not reported
    Section 3.3 states short-term memories are compressed once a threshold is reached; the threshold controls how agent memory evolves and what users can observe.
  • Chef persona and five-agent roster = primary agent as restaurant chef
    Chosen by hand to control experimental variables; the authors note in Limitations that the chef persona carries service-oriented undertones that may influence perceived equality.
assumptions (4)
  • domain assumption Self-reported Likert items adapted from HALIE (plus an added equality dimension) validly measure perceived equality.
    Section 4.2 describes the questionnaire but no validated equality scale is used; Section 6.2 explicitly calls for more rigorous scales.
  • domain assumption Dual-track backend makes Unobservable and Observable conditions behaviorally identical except for the visual layer.
    Asserted in Section 4.1, but Section 3.3 gives Observable users an additional visible-influence affordance; no manipulation check verifies the equivalence.
  • domain assumption Sociological theories of friendship (equal positioning, mutual investment) transfer to human-agent relationships.
    Sections 1 and 3.1 ground the design in social identity theory and friendship theories; the transfer from human-human to human-agent relations is assumed, not tested.
  • domain assumption LLM behavior is consistent enough across sessions and conditions to avoid confounding.
    DeepSeek Chat and GPT-4.1-mini are used with no reproducibility or consistency checks; prompt details and generation parameters are not disclosed.
invented entities (1)
  • Observable Social Life Spaces (OSLS) paradigm
    purpose: Design construct coupling a visible persistent virtual world with dual-track memory to cultivate perceived equality
    Defined and evaluated only within this paper; no external benchmark or falsifiable quantitative target; effectiveness rests on self-report and qualitative interpretation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Observable Social Life Spaces: Exploring User Interpretations of agent-side life context in human-agent interaction." pith.science (2026). https://pith.science/paper/KYLGCYZY

@misc{pith2026260321505,
  author       = {Pith},
  title        = {Pith review of: Observable Social Life Spaces: Exploring User Interpretations of agent-side life context in human-agent interaction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KYLGCYZY}},
  note         = {Machine review of arXiv:2603.21505}
}
abstract

Many AI agents are organized around instrumental "command-execution" interactions, where users primarily encounter agents through task requests and responses. Recent work on generative agents and agent life worlds has drawn attention to agents that maintain social contexts beyond direct user commands. In this paper, we study how observable social life spaces shape users' subjective experience, relational interpretations, and perceived equality during human-agent interaction. We introduce the \textit{Observable Social Life Spaces} paradigm, where agents inhabit a continuous virtual environment, engage in daily activities, and form social relationships that users can directly observe. Through an exploratory mixed-methods study ($N=24$), we found that the Observable condition yielded higher perceived-equality ratings and more frequent equality-related role descriptions than the Baseline and Unobservable conditions, but participant-level analysis suggests that the quantitative effect should be interpreted cautiously. We discuss perceived equality as a user-perception signal shaped by this design, with attention to boundary conditions including visual richness, novelty, and person-like attribution from visible agent cues.

Figures

Figures reproduced from arXiv: 2603.21505 by the authors.

Figure 1
Figure 1. Overview of the Observable Social Life Spaces paradigm aimed at cultivating perceived equality in human-agent [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. System architecture of the Observable Social Life Spaces for Interactive Agents. In the lower panel, the system is [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Four progressive application scenarios for the Observable Social Life Spaces paradigm. The upper left panel visualizes [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

50 extracted references · 3 linked inside Pith

  1. [1]

    Petter Bae Bae Brandtzæg, Marita Skjuve, Kim Kristoffer Kristoffer Dysthe, and Asbjørn Følstad. 2021. When the social becomes non-human: young people’s perception of social support in chatbots. InProceedings of the 2021 CHI conference on human factors in computing systems. 1–13

    2021

  2. [2]

    Timothy Bickmore and Justine Cassell. 2001. Relational agents: a model and implementation of building user trust. InProceedings of the SIGCHI conference on Human factors in computing systems. 396–403

    2001

  3. [3]

    Timothy W Bickmore, Lisa Caruso, and Kerri Clough-Gorr. 2005. Acceptance and usability of a relational agent interface by urban older adults. InCHI’05 extended abstracts on Human factors in computing systems. 1212–1215

    2005

  4. [4]

    Timothy W Bickmore and Rosalind W Picard. 2004. Towards caring machines. InCHI’04 extended abstracts on Human factors in computing systems. 1489–1492

    2004

  5. [5]

    Petter Bae Brandtzaeg, Marita Skjuve, and Asbjørn Følstad. 2022. My AI friend: How users of a social chatbot understand their human–AI friendship.Human Communication Research48, 3 (2022), 404–429. He et al

    2022

  6. [6]

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

    2006

  7. [7]

    Hyunsung Cho, Jacqui Fashimpaur, Naveen Sendhilnathan, Jonathan Browder, David Lindlbauer, Tanya R Jonker, and Kashyap Todi. 2025. Persistent Assistant: Seamless Everyday AI Interactions via Intent Grounding and Multimodal Feed- back. InProceedings of the 2025 CHI Conference on Human Factors in Computing Systems. 1–19

    2025

  8. [8]

    Marek Chojnacki. 2018. Time and meaning. Alfred schutz’s concept of intersub- jectivity and its cognitive corollaries.Humanities and Social Sciences25, 2 (2018), 65–80

    2018

Show all 50 references
  1. [9]

    Roberto Cipriani. 2013. The many faces of social time: A sociological approach. Time & Society22, 1 (2013), 5–30

    2013

  2. [10]

    Dean Cocking and Jeanette Kennett. 1998. Friendship and the Self.ethics108, 3 (1998), 502–527

    1998

  3. [11]

    Samuel Rhys Cox, Rune Møberg Jacobsen, and Niels van Berkel. 2025. The Impact of a Chatbot’s Ephemerality-Framing on Self-Disclosure Perceptions. In Proceedings of the 7th ACM Conference on Conversational User Interfaces. 1–17

    2025

  4. [12]

    Samuel Rhys Cox, Yi-Chieh Lee, and Wei Tsang Ooi. 2023. Comparing how a chatbot references user utterances from previous chatting sessions: an inves- tigation of users’ privacy concerns and perceptions. InProceedings of the 11th International Conference on Human-Agent Interact...

    2023

  5. [13]

    Fred D Davis. 1989. Perceived usefulness, perceived ease of use, and user accep- tance of information technology.MIS quarterly13, 3 (1989), 319–340

    1989

  6. [14]

    David DeVault, Ron Artstein, Grace Benn, Teresa Dey, Ed Fast, Alesia Gainer, Kallirroi Georgila, Jon Gratch, Arno Hartholt, Margaux Lhommet, et al . 2014. SimSensei Kiosk: A virtual human interviewer for healthcare decision support. InProceedings of the 2014 international conf...

    2014

  7. [15]

    1996.Friendship processes

    Beverley Fehr. 1996.Friendship processes. Vol. 12. Sage

    1996

  8. [16]

    Susan T Fiske, Amy JC Cuddy, and Peter Glick. 2007. Universal dimensions of social cognition: Warmth and competence.Trends in cognitive sciences11, 2 (2007), 77–83

    2007

  9. [17]

    Erving Goffman. 2023. The presentation of self in everyday life. InSocial theory re-wired. Routledge, 450–459

    2023

  10. [18]

    my agent understands me better

    Yuki Hou, Haruki Tamoto, and Homei Miyashita. 2024. " my agent understands me better": Integrating dynamic human-like memory recall and consolidation in llm-based agents. InExtended Abstracts of the CHI Conference on Human Factors in Computing Systems. 1–7

    2024

  11. [19]

    Jin Jeong and Tak Yeon Lee. 2025. LIGS: Developing an LLM-infused Game System for Emergent Narrative. InProceedings of the Extended Abstracts of the CHI Conference on Human Factors in Computing Systems. 1–12

    2025

  12. [20]

    Kangsoo Kim, Divine Maloney, Gerd Bruder, Jeremy N Bailenson, and Gregory F Welch. 2017. The effects of virtual human’s spatial and behavioral coherence with physical objects on social presence in AR.Computer Animation and Virtual Worlds28, 3-4 (2017), e1771

    2017

  13. [21]

    Tae Kyun Kim. 2015. T test as a parametric statistic.Korean journal of anesthesi- ology68, 6 (2015), 540–546

    2015

  14. [22]

    Mikko Korkiakoski, Saeid Sheikhi, Jesper Nyman, Jussi Saariniemi, Kalle Tapio, and Panos Kostakos. 2025. An Empirical Evaluation of AI-Powered Non-Player Characters’ Perceived Realism and Performance in Virtual Reality Environments. arXiv preprint arXiv:2507.10469(2025)

    2025 arXiv

  15. [23]

    Mina Lee, Megha Srivastava, Amelia Hardy, John Thickstun, Esin Durmus, Ash- win Paranjape, Ines Gerard-Ursin, Xiang Lisa Li, Faisal Ladhak, Frieda Rong, et al. 2022. Evaluating human-language model interaction.arXiv preprint arXiv:2212.09746(2022)

    2022 arXiv

  16. [24]

    Yuan Li, Lichao Sun, and Yixuan Zhang. 2025. Metaagents: large language model based agents for decision-making on teaming.Proceedings of the ACM on Human-Computer Interaction9, 2 (2025), 1–27

    2025

  17. [25]

    Micah Lott and William Hasselberger. 2025. With Friends Like These: Love and Friendship with AI Agents: M. Lott and W. Hasselberger.Topoi(2025), 1–13

    2025

  18. [26]

    Joon Sung Park, Joseph O’Brien, Carrie Jun Cai, Meredith Ringel Morris, Percy Liang, and Michael S Bernstein. 2023. Generative agents: Interactive simulacra of human behavior. InProceedings of the 36th annual acm symposium on user interface software and technology. 1–22

    2023

  19. [27]

    Chen Qian, Wei Liu, Hongzhang Liu, Nuo Chen, Yufan Dang, Jiahao Li, Cheng Yang, Weize Chen, Yusheng Su, Xin Cong, et al. 2024. Chatdev: Communicative agents for software development. InProceedings of the 62nd annual meeting of the association for computational linguistics (vol...

    2024

  20. [28]

    1988.Autobiographical memory

    David C Rubin. 1988.Autobiographical memory. Cambridge University Press

    1988

  21. [29]

    Marita Skjuve, Asbjørn Følstad, Knut Inge Fostervold, and Petter Bae Brandtzaeg

  22. [30]

    Mel Slater. 2018. Immersion and the illusion of presence in virtual reality.British journal of psychology109, 3 (2018), 431–433

    2018

  23. [31]

    Michael R Solomon, Carol Surprenant, John A Czepiel, and Evelyn G Gutman

  24. [32]

    Lars St, Svante Wold, et al. 1989. Analysis of variance (ANOVA).Chemometrics and intelligent laboratory systems6, 4 (1989), 259–272

    1989

  25. [33]

    Bas Stronks, Anton Nijholt, Paul van Der Vet, Dirk Heylen, and Aaron Machado

  26. [34]

    Sheldon Stryker. 1980. Symbolic interactionism: A social structural version.(No Title)(1980)

    1980

  27. [35]

    Mitsuhiro Tada. 2019. Time as sociology’s basic concept: A perspective from Al- fred Schutz’s phenomenological sociology and Niklas Luhmann’s social systems theory.Time & Society28, 3 (2019), 995–1012

    2019

  28. [36]

    Henri Tajfel, John Turner, William G Austin, Stephen Worchel, et al. 2001. An integrative theory of intergroup conflict.Intergroup relations: Essential readings (2001), 94–109

    2001

  29. [37]

    1978.Differentiation between social groups: Studies in the social psychology of intergroup relations.Academic press

    Henri Ed Tajfel. 1978.Differentiation between social groups: Studies in the social psychology of intergroup relations.Academic press

    1978

  30. [38]

    Lisa M Tillmann-Healy. 2003. Friendship as method.Qualitative inquiry9, 5 (2003), 729–749

    2003

  31. [39]

    Mike Treanor, Ben Samuel, and Mark J Nelson. 2025. Slice of Life: A Social Physics Game with Interactive Conversations using Symbolically Grounded LLM-Based Generative Dialogue. InProceedings of the 20th International Conference on the Foundations of Digital Games. 1–11

    2025

  32. [40]

    Siran Wang, Qiang Yan, and Lingli Wang. 2025. Task-oriented vs. social-oriented: chatbot communication styles in electronic commerce service recovery.Electronic Commerce Research25, 3 (2025), 1793–1825

    2025

  33. [41]

    Zihao Wang, Shaofei Cai, Anji Liu, Yonggang Jin, Jinbing Hou, Bowei Zhang, Haowei Lin, Zhaofeng He, Zilong Zheng, Yaodong Yang, et al. 2024. Jarvis-1: Open-world multi-task agents with memory-augmented multimodal language models.IEEE Transactions on Pattern Analysis and Machin...

    2024

  34. [42]

    Jochen Wirtz, Paul G Patterson, Werner H Kunz, Thorsten Gruber, Vinh Nhat Lu, Stefanie Paluch, and Antje Martins. 2018. Brave new world: service robots in the frontline.Journal of service management29, 5 (2018), 907–931

    2018

  35. [43]

    Zhen Wu, Serkan Kumyol, Shing Yin Wong, Xiaozhu Hu, Xin Tong, and Tristan Braud. 2025. Orchid: A Creative Approach for Authoring LLM-Driven Interactive Narratives. InProceedings of the 2025 Conference on Creativity and Cognition. 774–791

    2025

  36. [44]

    Nima Zargham, Maximilian A Friehs, Leandro Tonini, Dmitry Alexandrovsky, Emma Grace Ruthven, Lennart E Nacke, and Rainer Malaka. 2025. Let’s talk games: An expert exploration of speech interaction with NPCs.International Journal of Human–Computer Interaction41, 5 (2025), 3592–3612

    2025

  37. [45]

    Nima Zargham, Leandro Tonini, Dmitry Alexandrovsky, Emma Grace Ruthven, Maximilian A Friehs, Leon Tristan Dratzidis, Bastian Dänekas, Ioannis Bikas, Lennart E Nacke, Sven Zebel, et al. 2026. Dialogs with GenAI NPCs: Exploring Player Interactions with Speech Agents in a VR Game...

    2026

  38. [46]

    Chi Zhang, Zhao Yang, Jiaxuan Liu, Yanda Li, Yucheng Han, Xin Chen, Zebiao Huang, Bin Fu, and Gang Yu. 2025. Appagent: Multimodal agents as smartphone users. InProceedings of the 2025 CHI Conference on Human Factors in Computing Systems. 1–20

    2025

  39. [47]

    Hongxin Zhang, Zheyuan Zhang, Zeyuan Wang, Zunzhe Zhang, Lixing Fang, Qinhong Zhou, and Chuang Gan. 2025. Ella: Embodied Social Agents with Lifelong Memory.arXiv preprint arXiv:2506.24019(2025)

    2025 arXiv

  40. [1985]

    Journal of marketing49, 1 (1985), 99–111

    A role theory perspective on dyadic interactions: the service encounter. Journal of marketing49, 1 (1985), 99–111

    1985

  41. [2002]

    Embodied conversational agents-let’s specify and evaluate them(2002), 91–97

    Designing for friendship: Becoming friends with your ECA.Proc. Embodied conversational agents-let’s specify and evaluate them(2002), 91–97

    2002

  42. [2021]

    My chatbot companion-a study of human-chatbot relationships.Interna- tional Journal of Human-Computer Studies149 (2021), 102601

    2021

Pith tools

Reviewed August 2, 2026 · model on record in the stance chip above.