pith. sign in

arxiv: 2604.13996 · v1 · submitted 2026-04-15 · 💻 cs.HC

Acts of Configuration: Rethinking Provenance, Temporality and Legitimacy in Post-Mortem Agents

Kellie Yu Hui Sim , Pin Sym Foong , Darryl Lim , John-Henry Lim , Kenny Tsu Wei Choo This is my paper

Pith reviewed 2026-05-10 12:26 UTC · model grok-4.3

classification 💻 cs.HC
keywords post-mortem agentsadvance care planningAI delegationtemporalityprovenancelegitimacyimpaired capacitypersona persistence
0
0 comments X

The pith

Configuring AI agents for use after capacity loss reshapes how we understand their origins, timing of changes, and legitimacy for post-mortem roles.

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

Work on persona-persistent post-mortem agents has centered on a life-or-death divide, but this paper focuses on the overlooked phase when a person is still alive yet lacks decisional capacity. In a multi-phase workshop, participants trained AI agents for advance care planning and reflected on delegation: they preferred bounded agents tied closely to the user's own inputs rather than independent or expanding ones. Temporality led to a new pattern where agents adapt while the user has capacity, then stay fixed after capacity ends, yet still support later post-mortem applications. This act of configuring the agent for that post-capacity window alters standard ideas of provenance, temporality, and legitimacy. A sympathetic reader cares because it points toward designs that better match how people actually want their digital stand-ins to behave across changing life stages.

Core claim

Participants initially favoured bounded agents grounded in first-party authorship and representational fidelity over autonomous or evolving stand-ins. Temporality introduced novel ideas like adjacent use driven by persona persistence over functional expansion: agents should evolve while users retain capacity, remain static once capacity is lost, but somehow inform adjacent post-mortem uses. The configuration of agents for post-capacity use thus reshapes our understanding of provenance, temporality, and legitimacy for post-mortem agents.

What carries the argument

The configuration of agents for post-capacity use, which draws on participant reflections to set bounds on autonomy, fix timing of change, and anchor authority in original user inputs.

If this is right

  • Agents should remain bounded and grounded in first-party authorship to preserve representational fidelity.
  • Agents can adapt and evolve while the user retains capacity but must become static once capacity is lost.
  • The fixed agent can still support adjacent post-mortem uses through persona persistence without adding new functions after death.
  • This configuration process directly redefines provenance as tied to original authorship, temporality as phased rather than continuous, and legitimacy as dependent on the user's pre-loss inputs.
  • Design for post-mortem agents must account for the impaired-capacity period rather than defaulting to a simple life/death binary.

Where Pith is reading between the lines

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

  • Similar configuration logic could apply to other AI delegation tools such as financial or medical decision agents where capacity may fluctuate.
  • Real-world testing in clinical advance care planning tools could check whether the workshop preferences translate into actual user behavior and satisfaction.
  • The phased approach raises questions about how to technically implement a 'lock-in' point without losing useful context for later uses.
  • This view connects to wider debates on AI agency by suggesting that delegation rights are not permanent but stage-dependent.

Load-bearing premise

The ways workshop participants described preferences and temporality for agent delegation reflect broader patterns that hold outside this specific group and setting.

What would settle it

A study with a broader or different participant group in actual advance care planning scenarios where most people instead prefer fully autonomous, continuously evolving agents even after capacity loss would undermine the claim.

Figures

Figures reproduced from arXiv: 2604.13996 by Darryl Lim, John-Henry Lim, Kellie Yu Hui Sim, Kenny Tsu Wei Choo, Pin Sym Foong.

Figure 1
Figure 1. Figure 1: Slides used to introduce and prompt discussion of the deployment timeline parameter (fourth parameter). [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
read the original abstract

Work on persona-persistent post-mortem agents typically frames design around a life/death binary. This framing neglects a consequential yet under-theorised condition: when individuals remain alive but have impaired decisional capacity. Drawing on a multi-phase workshop in which participants trained and reflected on an AI agent for Advance Care Planning, we examined how people reason about agentic delegation post-capacity loss. Initially, participants favoured bounded agents grounded in first-party authorship and representational fidelity over autonomous or evolving stand-ins. However, temporality introduced novel ideas like adjacent use driven by persona persistence over functional expansion: agents should evolve while users retain capacity, remain static once capacity is lost, but somehow inform adjacent post-mortem uses. We discuss the implications of these findings and propose that the configuration of agents for post-capacity use reshapes our understanding of provenance, temporality, and legitimacy for post-mortem agents.

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

2 major / 1 minor

Summary. The paper reports findings from a multi-phase workshop in which participants trained and reflected on an AI agent for Advance Care Planning. It claims that participants initially preferred bounded agents grounded in first-party authorship and representational fidelity over autonomous or evolving stand-ins. The study identifies novel temporality ideas, such as agents that evolve while users retain capacity, remain static after capacity loss, yet inform adjacent post-mortem uses. The authors propose that configuring agents for post-capacity use reshapes understandings of provenance, temporality, and legitimacy for post-mortem agents, moving beyond a life/death binary.

Significance. If the qualitative findings prove robust and generalizable, the work offers a meaningful contribution to HCI and AI ethics by highlighting an under-theorized condition (impaired decisional capacity while alive) and introducing temporal configurations that could inform more nuanced designs for persistent agents in sensitive domains like end-of-life planning. It provides conceptual scaffolding for rethinking agent legitimacy and provenance that could influence future design guidelines and empirical studies.

major comments (2)
  1. The abstract and methods description report directional findings on participant preferences and temporality ideas but supply no information on participant count, recruitment method, demographics, workshop design phases, data collection, or analysis protocol. Because the central claims derive directly from these reported reflections, the absence of this information prevents evaluation of whether the data support the proposed theoretical shift (see also the generalizability concern in the skeptic note).
  2. The proposal that configuration for post-capacity use 'reshapes our understanding' of provenance, temporality, and legitimacy rests on a single workshop sample without explicit discussion of limitations, alternative interpretations, or how the observed patterns would generalize beyond the specific participant group and setting. This weakens the load-bearing move from empirical observations to conceptual reframing.
minor comments (1)
  1. The abstract could more clearly distinguish between initial preferences and the novel temporality ideas that emerged during reflection.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive comments, which identify key opportunities to strengthen the transparency and scope of our qualitative study. We respond to each major comment below.

read point-by-point responses

  1. Referee: The abstract and methods description report directional findings on participant preferences and temporality ideas but supply no information on participant count, recruitment method, demographics, workshop design phases, data collection, or analysis protocol. Because the central claims derive directly from these reported reflections, the absence of this information prevents evaluation of whether the data support the proposed theoretical shift (see also the generalizability concern in the skeptic note).

    Authors: We agree that the abstract omits these details, which is common for brevity but hinders evaluation. The manuscript's methods section outlines the multi-phase workshop but does not provide the requested specifics on participant count, recruitment, demographics, data collection, or analysis protocol. We will revise the abstract to include key elements (e.g., participant numbers and recruitment approach) and expand the methods section with these details from the study records. This will enable readers to assess the grounding of the directional findings and the proposed theoretical shift. revision: yes

  2. Referee: The proposal that configuration for post-capacity use 'reshapes our understanding' of provenance, temporality, and legitimacy rests on a single workshop sample without explicit discussion of limitations, alternative interpretations, or how the observed patterns would generalize beyond the specific participant group and setting. This weakens the load-bearing move from empirical observations to conceptual reframing.

    Authors: The conceptual proposal is presented as emerging from the workshop observations rather than a broad empirical generalization. We acknowledge that the manuscript would be strengthened by an explicit limitations discussion. In revision, we will add a dedicated subsection addressing the single-workshop sample, potential limits on generalizability to other groups or settings, and alternative interpretations of the reflections (e.g., context-specific influences from advance care planning). This will clarify the scope of the reframing without overstating the empirical basis. revision: yes

Circularity Check

0 steps flagged

No circularity: claims grounded in workshop data, not self-referential derivation

full rationale

The paper's central proposal—that configuring agents for post-capacity use reshapes provenance, temporality, and legitimacy—derives directly from reported participant reflections in a multi-phase workshop on AI agents for Advance Care Planning. No equations, fitted parameters, or predictions are present; the text presents directional qualitative findings (preference for bounded agents, ideas of adjacent-use temporality) as empirical observations and then discusses their conceptual implications. No self-citations are invoked as load-bearing uniqueness theorems, no ansatzes are smuggled, and no known results are renamed as novel derivations. The derivation chain is self-contained against external benchmarks because it rests on primary data collection rather than reducing to prior assumptions or fits by construction. Generalizability concerns exist but fall outside circularity criteria.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

This is a qualitative HCI study drawing conclusions from workshop reflections. It relies on standard domain assumptions about the value of user-centered workshops for theory development but introduces no free parameters, mathematical derivations, or new postulated entities.

axioms (1)
  • domain assumption Workshop participant reflections on AI agent design can reveal generalizable patterns in reasoning about post-capacity delegation.
    The paper uses these reflections to support broader claims about agent configuration and legitimacy.

pith-pipeline@v0.9.0 · 5469 in / 1260 out tokens · 76343 ms · 2026-05-10T12:26:50.487617+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

15 extracted references · 15 canonical work pages

  1. [1]

    Ruben Albers, Shadan Sadeghian, Matthias Laschke, and Marc Hassenzahl. 2023. Dying, Death, and the Afterlife in Human-Computer Interaction. A Scoping Review.. InProceedings of the 2023 CHI Conference on Human Factors in Computing Systems. ACM, Hamburg Germany, 1–16. doi:10.1145/3544548.3581199

    work page doi:10.1145/3544548.3581199 2023

  2. [2]

    Max Bain, Jaesung Huh, Tengda Han, and Andrew Zisserman. 2023. WhisperX: Time-Accurate Speech Transcription of Long-Form Audio.INTERSPEECH 2023 (2023)

    work page 2023

  3. [3]

    Charisse Foo, Pin Sym Foong, Camille Nadal, Natasha Ureyang, Thant Naylin, and Gerald Choon Huat Koh. 2025. The Benefits and Risks of LLMs for Facilitating Medical Decision-Making Among Laypersons. InProceedings of the 2025 ACM Designing Interactive Systems Conference. ACM, Madeira Portugal, 3173–3191. doi:10.1145/3715336.3735779

    work page doi:10.1145/3715336.3735779 2025

  4. [4]

    Pin Sym Foong, Natasha Ureyang, Charisse Foo, Sajeban Antonyrex, and Ger- ald Ch Koh. 2024. Designing for Caregiver-facing Values Elicitation Tools. In Proceedings of the CHI Conference on Human Factors in Computing Systems. ACM, Honolulu HI USA, 1–20. doi:10.1145/3613904.3642214

    work page doi:10.1145/3613904.3642214 2024

  5. [5]

    Nicola K Gale, Gemma Heath, Elaine Cameron, Sabina Rashid, and Sabi Redwood

    work page

  6. [6]

    BMC Medical Research Methodology13(1), 117 (2013) https://doi

    Using the Framework Method for the Analysis of Qualitative Data in Rethinking Provenance, Temporality and Legitimacy in Post-Mortem Agents DIS Companion ’26, June 13–17, 2026, Singapore, Singapore Multi-Disciplinary Health Research.BMC Medical Research Methodology13, 1 (Sept. 2013), 117. doi:10.1186/1471-2288-13-117

    work page doi:10.1186/1471-2288-13-117 2026

  7. [7]

    Tomasz Hollanek and Katarzyna Nowaczyk-Basińska. 2024. Griefbots, Deadbots, Postmortem Avatars: On Responsible Applications of Generative AI in the Digital Afterlife Industry.Philosophy & Technology37, 2 (2024), 63. doi:10.1007/s13347- 024-00744-w Speculative design scenarios for deadbots, stakeholder roles, and ethical recommendations

    work page doi:10.1007/s13347- 2024

  8. [8]

    Jessica Huang, Ig-Jae Kim, and Dongwook Yoon. 2025. Mirror to Companion: Exploring Roles, Values, and Risks of AI Self-Clones through Story Completion. InProceedings of the 2025 CHI Conference on Human Factors in Computing Systems. ACM, Yokohama Japan, 1–15. doi:10.1145/3706598.3713587

    work page doi:10.1145/3706598.3713587 2025

  9. [9]

    Jurgis Karpus and Anna Strasser. 2025. Persons and Their Digital Replicas. Philosophy & Technology38, 1 (March 2025), 25. doi:10.1007/s13347-025-00854-z

    work page doi:10.1007/s13347-025-00854-z 2025

  10. [10]

    Celine Latulipe, Ronnie Dsouza, and Murray Cumbers. 2022. Unofficial Proxies: How Close Others Help Older Adults with Banking. InProceedings of the 2022 CHI Conference on Human Factors in Computing Systems (CHI ’22). Association for Computing Machinery, New York, NY, USA, 1–13. doi:10.1145/3491102.3501845

    work page doi:10.1145/3491102.3501845 2022

  11. [11]

    AI Afterlife

    Ying Lei, Shuai Ma, Yuling Sun, and Xiaojuan Ma. 2025. “AI Afterlife” as Digital Legacy: Perceptions, Expectations, and Concerns.arXiv(2025). https://arxiv. org/abs/2502.10924 Qualitative study on perceptions of AI afterlives as digital legacy and design implications

    work page arXiv 2025

  12. [12]

    Brubaker

    Meredith Ringel Morris and Jed R. Brubaker. 2025. Generative Ghosts: Anticipat- ing Benefits and Risks of AI Afterlives. InProceedings of the 2025 CHI Conference on Human Factors in Computing Systems (CHI ’25). Association for Computing Machinery, New York, NY, USA, 1–14. doi:10.1145/3706598.3713758

    work page doi:10.1145/3706598.3713758 2025

  13. [13]

    Kellie Yu Hui Sim, Pin Sym Foong, Chenyu Zhao, Melanie Yi Ning Quek, Swarangi Subodh Mehta, and Kenny Tsu Wei Choo. 2026. Words to Describe What I’m Feeling: Exploring the Potential of AI Agents for High Subjectivity Decisions in Advance Care Planning. InProceedings of the 2026 CHI Conference on Human Factors in Computing Systems (CHI ’26). Association fo...

    work page doi:10.1145/3772318.3791335 2026

  14. [14]

    Tianqi Song, Yugin Tan, Zicheng Zhu, Yibin Feng, and Yi-Chieh Lee. 2025. Greater than the Sum of Its Parts: Exploring Social Influence of Multi-Agents. InPro- ceedings of the Extended Abstracts of the CHI Conference on Human Factors in Computing Systems (CHI EA ’25). Association for Computing Machinery, New York, NY, USA, 1–11. doi:10.1145/3706599.3719973

    work page doi:10.1145/3706599.3719973 2025

  15. [15]

    Tianqi Song, Yugin Tan, Zicheng Zhu, Yibin Feng, and Yi-Chieh Lee. 2025. Multi- Agents Are Social Groups: Investigating Social Influence of Multiple Agents in Human-Agent Interactions.Proc. ACM Hum.-Comput. Interact.9, 7 (Oct. 2025), CSCW452:1–CSCW452:33. doi:10.1145/3757633

    work page doi:10.1145/3757633 2025