Pith. sign in

REVIEW 3 major objections 6 minor 100 references

Same Stories, Different Journeys: From Social Comparison to Sensemaking in AI-Mediated Peer Career Exploration

T0 review · 3 major / 6 minor · reviewed 2026-07-14 · grok-4.5

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

Pith's one-line read When peer career posts stay fixed, shifting from passive scrolling to AI persona dialogue lowers cognitive effort and redirects upward comparison into self-reframing.

desk verdict Solid CHI system paper with a real design pattern; the modality claim is confounded with the full package, but the work still earns a serious referee. read the letter →

arxiv 2607.11039 v1 pith:M2YWBAB5 submitted 2026-07-13 cs.HC cs.AI

Pengping Tan , Baoquan Zhao , Zhenhui Peng This is my paper · ORCID
classification cs.HCcs.AI
keywords CareerexplorationSocialcomparisonSensemakingPersona-groundedagentsmediaUGCRetrieval-augmentedgenerationSelf-determinationtheory
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

Young job seekers use authentic peer posts for career grounding, but passive feeds also produce overload and upward comparison anxiety. This paper claims the tension is driven more by interaction modality than by the content itself. JobMate converts real social-media career posts into structured persona cards and Self-Determination-Theory-guided conversational agents, keeping the original posts visible as authenticity anchors. In a between-subjects study of 24 students across three disciplines, both native browsing and JobMate reduced career-decision difficulties, yet JobMate did so at significantly lower effort and shifted users from “I’m not as good as others” toward “what should I do next,” while still relying on real posts for emotional grounding. The result matters because it shows designers can keep the value of authentic peer stories while redesigning how people encounter them.

What carries the argument

JobMate: a four-stage pipeline that cleans and classifies real career posts, extracts person-centric fields (background, outcome, challenges tags, summary), builds dual-track retrieval-augmented personas, and runs dialogue under Self-Determination Theory rules (relatedness via empathic self-disclosure, competence via reframing, autonomy via non-directive options). The mechanism converts passive feed consumption into active, grounded conversation while leaving the original post visible.

What would settle it

A larger multi-week field study comparing JobMate-style dialogue with native browsing that found no NASA-TLX Effort difference, no qualitative shift from upward-comparison language to next-step self-reframing, and no continued reliance on real posts for emotional grounding would falsify the claim that modality, not content, drives the outcome.

Watch

Extended reading notes

Core claim

Holding authentic peer career content fixed, AI-mediated persona dialogue reduces cognitive cost relative to native social-media browsing and redirects social comparison from potentially detrimental upward comparison toward constructive self-reframing and next-step sensemaking; users nevertheless continue to treat the real posts as the emotional and trust anchor.

Load-bearing premise

A single 30-minute laboratory exploration task with fixed objectives is assumed to capture the same social-comparison dynamics, sensemaking depth, and emotional grounding that arise in naturalistic multi-session job-seeking on social media.

Editorial extensions

If this is right

  • Interaction modality, not content alone, shapes whether authentic peer experiences produce anxiety or usable self-knowledge.
  • Foregrounding challenges rather than achievements on persona cards can steer comparison toward lateral normalization instead of upward threat.
  • AI systems that ground dialogue in real user-generated content can match the perceived support of human platforms while lowering screening cost.
  • Career and other high-comparison domains need adjustable density and emotion-versus-action modes for different cognitive styles.
  • Showing the real-post basis of each persona is required for users to trust AI-mediated peer experience.

Reading between the lines

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

  • The same modality shift—authentic posts kept, interaction changed to grounded dialogue—could reduce comparison harm in fitness, parenting, academic-grade, or chronic-illness peer content without removing the stories people trust.
  • Longitudinal deployment is needed to test whether short-term effort savings and self-reframing convert into actual career actions rather than rebound anxiety.
  • Coverage bias in the underlying posts (over-representation of tech or large-company paths) can make the method feel templated for underrepresented trajectories unless retrieval deliberately rebalances them.
  • Component ablations (cards alone vs dialogue alone vs SDT framing) would isolate which piece actually redirects comparison direction.
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

3 major / 6 minor

Summary. The paper presents JobMate, a system that converts real RedNote career posts into challenge-foregrounded persona cards and SDT-informed, dual-track RAG conversational agents, aiming to keep authentic peer content while shifting users from passive feed browsing to active dialogue. A formative survey/interview study (N=64/8) motivates four design goals; a between-subjects lab study (N=24; CS, Psychology, Chinese Literature) compares JobMate to unconstrained native RedNote browsing on a 30-minute career-exploration task. Both conditions reduce CDDQ scores with no between-group difference on total reduction; JobMate shows lower NASA-TLX Effort (p_t=0.012) and marginally lower Frustration, with qualitative themes that comparison shifts from upward threat toward self-reframing and next-step sensemaking while authenticity of real posts remains the emotional anchor. The authors argue that interaction modality, not content alone, shapes the value–harm tension in peer-experience consumption, and they offer design implications for support modes, transparency, and coverage.

Significance. If the core claim holds—that redesigning interaction around authentic UGC can lower cognitive cost and redirect social comparison without discarding peer authenticity—the work is a useful contribution to HCI systems for career exploration and, more broadly, high-comparison UGC settings (study, fitness, parenting, health). Strengths include a complete end-to-end pipeline (cleaning, classification, structured extraction, dual-track RAG, SDT dialogue rules), explicit design goals tied to formative findings, multi-instrument evaluation (CDDQ, NASA-TLX, SDS, logs, interviews), and honest limitations on sample size, short task, single-platform coverage, and missing ablations. The contribution is primarily design-and-evaluation rather than a tightly isolated causal mechanism; its value for the field depends on how carefully claims about “modality” versus the full JobMate package are scoped.

major comments (3)
  1. Abstract, §1, §6.1–6.2, and §7.1 attribute lower Effort and redirection of social comparison primarily to “AI-mediated dialogue” / interaction modality while “authentic peer content is held fixed.” The control is unconstrained native RedNote browsing (§5.2), not a structured-card-only or original-post-only arm. JobMate simultaneously de-noises ads, compresses posts into challenge-first cards, anchors the original post, and adds dual-track RAG + SDT coaching (§4.1–4.4). Qualitative evidence credits clean cards and challenge tags as much as dialogue (§6.1–6.2), and §7.4 states components were not ablated. The causal claim for modality alone is therefore not secured; either add ablation/control conditions or reframe claims as effects of the full JobMate package versus native feed browsing.
  2. §5.1–5.2 and §6: N=24 (n=4 per discipline×condition cell) with Mann–Whitney subgroup tests is underpowered for the boundary-condition claims in §6.3 and for treating disciplinary cognitive style as a robust moderator. Pre–post CDDQ improves in both arms with no between-group total difference (p_t=0.432); the quantitative headline rests on Effort (p_t=0.012) and a marginal Frustration result, while the comparison-redirection claim is almost entirely thematic. The manuscript should (i) center the primary confirmatory contrast, (ii) label discipline analyses as exploratory, and (iii) avoid overstating “redirected social comparison” relative to the mixed quantitative pattern.
  3. §5.2 procedure and §7.4: the 30-minute fixed-objective lab task is a weak proxy for multi-session naturalistic job-seeking comparison and sensemaking. The paper’s own limitations note that anxiety rebound and action conversion were not observed. Given that the strongest claim concerns how comparison is experienced and converted into next steps, the ecological-validity gap is load-bearing; either strengthen the discussion of what short-task evidence can and cannot support, or plan/report a longer field deployment as central rather than future work.
minor comments (6)
  1. Figure 5 caption says plots label the control as “Baseline” for native RedNote; keep terminology consistent with “RedNote” throughout text and figures to avoid confusion with a true baseline condition.
  2. Implementation (§4.5) states conversational agents use GPT-5.2 while supplementary materials list gpt-4o-mini defaults; reconcile model names and report the actual deployment model used in the user study.
  3. Table 1 footnote uses † for both “Higher is better” and “Marginal (p<.10)”; disambiguate symbols.
  4. §6.1 reports two exploration strategies (deep divers n=5, broad explorers n=4) from logs; a brief definition of how mixed users were classified would help reproducibility.
  5. Related Work §2.3 and contributions: clarify more sharply what is novel relative to PlanHelper, DesignQuizzer, ComViewer, and SDT career chatbots beyond grounding personas in real others’ posts.
  6. Supplementary participant tables and full prompts are valuable; ensure the camera-ready main text points to them and that any Chinese-to-English instrument rendering notes are explicit for CDDQ/SDS adaptations.

Circularity Check

0 steps flagged · score 0.0 of 10

Empirical HCI evaluation with no circular derivation: outcomes are measured questionnaire deltas and interview themes, not quantities defined by their own inputs.

full rationale

JobMate is an interaction-design and between-subjects evaluation paper (N=24), not a first-principles derivation. Design goals (DG1–DG4) are motivated by a formative survey/interviews and prior theory (social comparison, SDT, cognitive load, ICAP); they do not define the dependent measures. The load-bearing claims—lower NASA-TLX Effort (p_t=0.012), comparable CDDQ reduction, and qualitative redirection of comparison—are pre/post and between-group empirical scores plus thematic analysis of interviews and logs. No parameter is fitted to data and then re-presented as a prediction; no uniqueness theorem or self-citation chain forces the result; SDT and social-comparison theory are interpretive frames, not redefinitions of CDDQ/NASA-TLX/SDS items. Confounds noted by a skeptic (modality bundled with de-noising, challenge tags, and LLM coaching; no ablation) are causal-attribution and design-validity issues, not circularity by construction. The paper is self-contained against its own instruments and control condition; score 0 is the honest finding.

Assumptions & free parameters 3 free parameters · 5 assumptions · 2 invented entities

Load-bearing structure is empirical HCI plus standard social-psychology and learning theories, plus engineering choices (LLM models, RAG tracks, SDT prompt rules). No free physical constants; free parameters are implementation hyperparameters and design choices that shape agent behavior. Invented entities are the system artifacts under evaluation, not postulated natural kinds.

free parameters (3)
  • LLM sampling hyperparameters (temperature 0.7, top_p 0.9, frequency/presence penalties 0.3, max_tokens 500)
    Supplementary defaults for greeting and multi-turn chat; change agent tone and thus qualitative experience without being fit to outcome scales, but they are free design knobs the results depend on.
  • Dual-track RAG retrieval depths (Top-2 homogeneous stories; Top-3 heterogeneous knowledge items)
    Hand-chosen retrieval counts that define each persona’s supporting context (§4.3); not derived from a uniqueness theorem.
  • Reply length cap (~150 Chinese characters) and challenge-tag foregrounding on cards
    Design parameters intended to promote lateral comparison and colloquial tone (DG3, §4.4); outcomes partly depend on these choices.
assumptions (5)
  • domain assumption Social comparison theory: upward comparison often harms self-evaluation; lateral comparison can normalize (Festinger; Buunk; Wood).
    Used to motivate challenge-first cards and interpret interview themes as “redirected comparison” (§2.1, §6.2).
  • domain assumption Self-Determination Theory needs (relatedness, competence, autonomy) can be operationalized as conversational rules that deliver emotional plus informational support.
    Structures the agent prompt and SDS measures (§4.4, §5.3).
  • domain assumption Active constructive/interactive engagement (generation effect, ICAP) yields deeper processing than passive reception.
    Justifies shifting modality from browsing to dialogue as the mechanism for sensemaking (§2.2, §7.1).
  • ad hoc to paper GPT-class models with RAG over extracted post fields can role-play a specific poster consistently enough for a 30-minute career dialogue without inventing contradictory biography.
    Core engineering premise of persona agents; constrained by “stay consistent with main post” rules but not independently validated beyond user ratings (§4.4–4.5).
  • ad hoc to paper Native unconstrained RedNote browsing is a valid baseline for “passive authentic peer-content consumption.”
    Between-subjects control definition (§5.2); confounds structure, ads, and ranking with pure modality.
invented entities (2)
  • JobMate dual-track RAG persona (core post + homogeneous resonance stories + heterogeneous knowledge)
    purpose: Transform authentic career UGC into conversable agents while preserving an authenticity anchor.
    System construct under test; independent_evidence false outside this evaluation.
  • Challenge-first persona cards (background, outcome, challenges tags, summary)
    purpose: Reduce screening cost and bias attention toward lateral rather than upward comparison.
    UI invention tied to DG3; effects not separated from dialogue in an ablation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Same Stories, Different Journeys: From Social Comparison to Sensemaking in AI-Mediated Peer Career Exploration." pith.science (2026). https://pith.science/paper/M2YWBAB5

@misc{pith2026260711039,
  author       = {Pith},
  title        = {Pith review of: Same Stories, Different Journeys: From Social Comparison to Sensemaking in AI-Mediated Peer Career Exploration},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/M2YWBAB5}},
  note         = {Machine review of arXiv:2607.11039}
}
abstract

Young job seekers frequently turn to social media to compare themselves with peers and make sense of career possibilities. However, passive feed browsing creates a paradox: the authentic peer content that provides emotional grounding also triggers potentially detrimental upward social comparison and cognitive overload. Previous work has either structured online user-generated content to reduce noise without changing the passive browsing modality, or built AI-powered career exploration systems that disregard authentic human experiences. To address this gap, we developed JobMate, an interactive system that transforms real social media career posts into persona-grounded conversational AI agents, shifting the interaction from passive scrolling to active, personalized dialogue. We conducted a between-subjects study ($N$ = 24, three disciplines) comparing JobMate with native RedNote browsing. Our study shows that JobMate's AI-mediated dialogue redirected social comparison from potentially detrimental upward comparison toward constructive self-reframing, while promoting sensemaking through active conversational engagement. However, users still relied on the authenticity of real peer content for emotional grounding. We discuss design implications for AI systems that augment authentic online user-generated content consumption across social comparison contexts.

Figures

Figures reproduced from arXiv: 2607.11039 by the authors.

Figure 1
Figure 1. End-to-end experience: native RedNote-style career feeds give way to structured persona fields ( [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. JobMate interface. (a) Two-phase onboarding collecting demographic attributes (school, major, degree) and psycholog [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Split-view detail interface. Left: original RedNote post as an authenticity anchor. Right: persona-grounded chat with [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Data pipeline overview. Raw posts are cleaned and LLM-classified; valid content splits into a [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Pre–post CDDQ subscales for JobMate vs. RedNote [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 7
Figure 7. Figure 7: JobMate-only ratings for the seven interface and [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 6
Figure 6. Figure 6: Conversation turns in JobMate, stacked by persona: [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

100 extracted references · 17 canonical work pages

  1. [1]

    Eugene Agichtein, Carlos Castillo, Debora Donato, Aristides Gionis, and Gilad Mishne. 2008. Finding High-Quality Content in Social Media.Proceedings of the International Conference on Web Search and Data Mining(2008), 183–194. doi:10.1145/1341531.1341557

    work page doi:10.1145/1341531.1341557 2008

  2. [2]

    Markus Appel, Caroline Marker, and Timo Gnambs. 2023. A Meta-Analysis of the Effects of Social Media Exposure to Upward Comparison Targets on Self- Evaluations and Emotions.Media Psychology26, 5 (2023), 661–681. doi:10.1080/ 15213269.2023.2180647 Same Stories, Different Journeys: From Social Comparison to Sensemaking in AI-Mediated Peer Career Exploration...

    arXiv 2023

  3. [3]

    Berkelaar

    Brenda L. Berkelaar. 2017. Different Ways New Information Technologies In- fluence Conventional Organizational Practices and Employment Relationships: The Case of Cybervetting for Personnel Selection.Human Relations70, 9 (2017), 1115–1140. doi:10.1177/0018726716686400

    work page doi:10.1177/0018726716686400 2017

  4. [4]

    Virginia Braun and Victoria Clarke. 2006. Using Thematic Analysis in Psy- chology.Qualitative Research in Psychology3, 2 (2006), 77–101. doi:10.1191/ 1478088706qp063oa

    2006

  5. [5]

    Buunk, Rebecca L

    Bram P. Buunk, Rebecca L. Collins, Shelley E. Taylor, Nico W. VanYperen, and Gayle A. Dakof. 1990. The Affective Consequences of Social Comparison: Either Direction Has Its Ups and Downs.Journal of Personality and Social Psychology 59, 6 (1990), 1238–1249. doi:10.1037/0022-3514.59.6.1238

    work page doi:10.1037/0022-3514.59.6.1238 1990

  6. [6]

    Zhuoer Chen et al. 2024. DesignQuizzer: A Community-Powered Conversational Agent for Learning Visual Design. InProceedings of the 2024 ACM Conference on Computer-Supported Cooperative Work and Social Computing. doi:10.1145/3637366

    work page doi:10.1145/3637366 2024

  7. [7]

    Michelene T. H. Chi and Ruth Wylie. 2014. The ICAP Framework: Linking Cognitive Engagement to Active Learning Outcomes.Educational Psychologist 49, 4 (2014), 219–243. doi:10.1080/00461520.2014.965823

    doi:10.1080/00461520.2014.965823 2014

  8. [8]

    Collins and Lynn Carol Miller

    Nancy L. Collins and Lynn Carol Miller. 1994. Self-Disclosure and Liking: A Meta- Analytic Review.Psychological Bulletin116, 3 (1994), 457–475. doi:10.1037/0033- 2909.116.3.457

    doi:10.1037/0033- 1994

Show all 100 references
  1. [9]

    DeAndrea

    David C. DeAndrea. 2015. Testing the Proclaimed Affordances of Online Support Groups in a Nationally Representative Sample of Adults Seeking Mental Health Assistance.Journal of Health Communication20, 2 (2015), 147–156. doi:10.1080/ 10810730.2014.914606

    2015

  2. [10]

    Wenxuan Du, Ziyi Zhu, Xinyi Xu, Haoyue Che, and Si Chen. 2024. CareerSim: Gamification Design Leveraging LLMs For Career Development Reflection. In Extended Abstracts of the CHI Conference on Human Factors in Computing Systems. 1–7. doi:10.1145/3613905.3650928

    2024 doi

  3. [11]

    Leon Festinger. 1954. A Theory of Social Comparison Processes.Human Relations 7, 2 (1954), 117–140. doi:10.1177/001872675400700202

    1954 doi

  4. [12]

    Raffaele Filieri and Fraser McLeay. 2014. E-WOM and Accommodation: An Analysis of the Factors That Influence Travelers’ Adoption of Information from Online Reviews.Journal of Travel Research53, 1 (2014), 44–57. doi:10.1177/ 0047287513481274

    2014

  5. [13]

    Itamar Gati, Mina Krausz, and Samuel H. Osipow. 1996. A Taxonomy of Difficul- ties in Career Decision Making.Journal of Counseling Psychology43, 4 (1996), 510–526. doi:10.1037/0022-0167.43.4.510

    1996 doi

  6. [14]

    Taewan Ha, Changhoon Oh, Yoonseo Oh, Dain Kwak, Daeyeon Kim, Minjoon Lee, and Juho Kim. 2024. CloChat: Understanding How People Customize, Interact, and Experience Personas in Large Language Models. InProceedings of the 2024 CHI Conference on Human Factors in Computing Systems...

    2024 doi

  7. [15]

    Hyoungwook Han, Bohyun Park, and Kwangsu Seo. 2025. A Self-Determination Theory-based Career Counseling Chatbot: Motivational Interactions to Address Career Decision-Making Difficulties and Enhance Engagement. InProceedings of the Extended Abstracts of the CHI Conference on Hu...

    2025 doi

  8. [16]

    Hart and Lowell E

    Sandra G. Hart and Lowell E. Staveland. 1988. Development of NASA-TLX (Task Load Index): Results of Empirical and Theoretical Research.Advances in Psychology52 (1988), 139–183. doi:10.1016/S0166-4115(08)62386-9

    1988 doi

  9. [17]

    Andreas Hirschi, Anne Herrmann, and Anita C. Keller. 2015. Career Adaptivity, Adaptability, and Adapting: A Conceptual and Empirical Investigation.Journal of Vocational Behavior87 (2015), 1–10. doi:10.1016/j.jvb.2014.11.008

    2015 doi

  10. [18]

    Angie Ho, Jeff Hancock, and Adam S. Miner. 2018. Psychological, Relational, and Emotional Effects of Self-Disclosure After Conversations With a Chatbot. Journal of Communication68, 4 (2018), 712–733. doi:10.1093/joc/jqy026

    2018 doi

  11. [19]

    Tristram Hooley. 2012. How the Internet Changed Career: Framing the Rela- tionship Between Career Development and Online Technologies.Journal of the National Institute for Career Education and Counselling29, 1 (2012), 3–12. doi:10.20856/jnicec.2902

    2012 doi

  12. [20]

    Hayeon Jeon, Hajin Lim, Eun-mee Kim, John Zimmerman, and Laura Dabbish

  13. [21]

    InProceedings of the 2025 CHI Conference on Human Factors in Computing Systems

    Letters from Future Self: Augmenting the Letter-Exchange Exercise with LLM-based Agents to Enhance Young Adults’ Career Exploration. InProceedings of the 2025 CHI Conference on Human Factors in Computing Systems. 1–21. doi:10. 1145/3706598.3714206

    2025

  14. [22]

    Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, and Douwe Kiela. 2020. Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks. InAdvances in N...

    2020 doi arXiv

  15. [23]

    Chengbo Li, Ting-Hao Huang, Zhenhui Peng, et al. 2022. PlanHelper: Support- ing Activity Plan Construction with Answer Posts in Community-based QA Platforms. InProceedings of the 2022 ACM Conference on Computer-Supported Cooperative Work and Social Computing. doi:10.1145/3555625

    2022 doi

  16. [24]

    Mufan Luo and Jeffrey T. Hancock. 2020. Self-Disclosure and Social Media: Motivations, Mechanisms and Psychological Well-Being.Current Opinion in Psychology31 (2020), 110–115. doi:10.1016/j.copsyc.2019.08.019

    2020 doi

  17. [25]

    Yuxin Luo, Jing Zhang, and Xin Li. 2024. Effects of Social Media Use on Employ- ment Anxiety among Chinese Youth: The Roles of Upward Social Comparison, Online Social Support and Self-Esteem.Frontiers in Psychology15 (2024), 1398801. doi:10.3389/fpsyg.2024.1398801

    2024 doi

  18. [26]

    Mowbray and Hazel Hall

    John A. Mowbray and Hazel Hall. 2021. Using Social Media During Job Search: The Case of 16–24 Year Olds in Scotland.Journal of Information Science47, 4 (2021), 535–550. doi:10.1177/0165551520927657

    2021 doi

  19. [27]

    2011.The Filter Bubble: What the Internet Is Hiding from You

    Eli Pariser. 2011.The Filter Bubble: What the Internet Is Hiding from You. Penguin Press

    2011

  20. [28]

    O’Brien, Carrie J

    Joon Sung Park, Joseph C. O’Brien, Carrie J. 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. doi:10.1...

    2023 doi

  21. [29]

    Peter Pirolli and Stuart Card. 2005. The Sensemaking Process and Leverage Points for Analyst Technology as Identified Through Cognitive Task Analysis. InProceedings of International Conference on Intelligence Analysis, Vol. 5. 2–4

    2005

  22. [30]

    Rains, Emily B

    Stephen A. Rains, Emily B. Peterson, and Kevin B. Wright. 2015. Communicating Social Support in Computer-Mediated Contexts: A Meta-Analytic Review of Content Analyses Examining Support Messages Shared Online.Communication Research42, 6 (2015), 796–819. doi:10.1177/0093650215573125

    2015 doi

  23. [31]

    Rosner, Jeremy A

    Zachary A. Rosner, Jeremy A. Elman, and Arthur P. Shimamura. 2013. The Generation Effect: Activating Broad Neural Circuits During Memory Encoding. Cortex49, 7 (2013), 1901–1909. doi:10.1016/j.cortex.2012.09.009

    2013 doi

  24. [32]

    Ryan and Edward L

    Richard M. Ryan and Edward L. Deci. 2000. Self-Determination Theory and the Facilitation of Intrinsic Motivation, Social Development, and Well-Being. American Psychologist55, 1 (2000), 68–78. doi:10.1037/0003-066X.55.1.68

    2000 doi

  25. [33]

    Yunfan Shao, Linyang Li, Junqi Dai, and Xipeng Qiu. 2024. Character-LLM: A Trainable Agent for Role-Playing.Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing(2024). doi:10.18653/v1/2024.emnlp- main.774

    2024 doi

  26. [34]

    Vera Liao, and Ziang Xiao

    Nikhil Sharma, Q. Vera Liao, and Ziang Xiao. 2024. Generative Echo Chamber? Effects of LLM-Powered Search Systems on Diverse Information Seeking. In Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems. doi:10.1145/3613904.3641924

    2024 doi

  27. [35]

    Slamecka and Peter Graf

    Norman J. Slamecka and Peter Graf. 1978. The Generation Effect: Delineation of a Phenomenon.Journal of Experimental Psychology: Human Learning and Memory4, 6 (1978), 592–604. doi:10.1037/0278-7393.4.6.592

    1978 doi

  28. [36]

    Duda, and Nikos Ntoumanis

    Martyn Standage, Joan L. Duda, and Nikos Ntoumanis. 2005. A Test of Self- Determination Theory in School Physical Education.British Journal of Educa- tional Psychology75, 3 (2005), 411–433. doi:10.1348/000709904X22359

    2005 doi

  29. [37]

    Sangho Suh, Bryan Min, Srishti Palani, and Haijun Xia. 2023. Sensecape: En- abling Multilevel Exploration and Sensemaking with Large Language Models. In Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology. doi:10.1145/3586183.3606756

    2023 doi

  30. [38]

    John Sweller. 1988. Cognitive Load During Problem Solving: Effects on Learning. Cognitive Science12, 2 (1988), 257–285. doi:10.1207/s15516709cog1202_4

    1988 doi

  31. [39]

    Emma van Zandvoort, Karel Vredenburg, and Marit Bentvelzen. 2025. Unhealthy Comparisons to Promote Healthy Behavior? Exploring the Impact of Social Comparison Strategies in Personal Informatics. InProceedings of the 2025 CHI Conference on Human Factors in Computing Systems. 1–...

    2025 doi

  32. [40]

    Philippe Verduyn, Oscar Ybarra, Maxime Résibois, John Jonides, and Ethan Kross

  33. [41]

    Do Social Network Sites Enhance or Undermine Subjective Well-Being? A Critical Review.Social Issues and Policy Review11, 1 (2017), 274–302. doi:10. 1111/sipr.12033

    2017

  34. [42]

    Vogel, Jason P

    Erin A. Vogel, Jason P. Rose, Lindsay R. Roberts, and Katheryn Eckles. 2014. Social Comparison, Social Media, and Self-Esteem.Psychology of Popular Media Culture3, 4 (2014), 206–222. doi:10.1037/ppm0000047

    2014 doi

  35. [43]

    Ziyuan Wang, Zhaoying Zeng, Yuwei Li, and Zhi Ding. 2025. CareerPooler: AI-Powered Metaphorical Pool Simulation Improves Experience and Outcomes in Career Exploration.arXiv preprint arXiv:2509.11461(2025). doi:10.48550/arXiv. 2509.11461

    2025 doi

  36. [44]

    Thomas A. Wills. 1981. Downward Comparison Principles in Social Psychology. Psychological Bulletin90, 2 (1981), 245–271. doi:10.1037/0033-2909.90.2.245

    1981 doi

  37. [45]

    Joanne V. Wood. 1989. Theory and Research Concerning Social Comparisons of Personal Attributes.Psychological Bulletin106, 2 (1989), 231–248. doi:10.1037/ 0033-2909.106.2.231

    1989

  38. [46]

    Shiwei Wu, Chuhan Shi, Zhenhui Peng, et al . 2025. ComViewer: An Interac- tive Visual Tool to Help Viewers Seek Social Support in Online Mental Health Communities. InProceedings of the 2025 ACM Conference on Computer-Supported Cooperative Work and Social Computing. doi:10.4855...

    2025 doi arXiv

  39. [47]

    Shangui Yang, Yun Liu, and Ye Liu. 2021. Social Comparison on Social Media Increases Career Frustration: A Focus on the Mitigating Effect of Companionship. Frontiers in Psychology12 (2021), 720960. doi:10.3389/fpsyg.2021.720960

    2021 doi

  40. [48]

    Saizheng Zhang, Emily Dinan, Jack Urbanek, Arthur Szlam, Douwe Kiela, and Jason Weston. 2018. Personalizing Dialogue Agents: I Have a Dog, Do You Have Pets Too?. InProceedings of the 56th Annual Meeting of the Association for Computational Linguistics. 2204–2213. doi:10.18653/...

    2018 doi

  41. [49]

    Yu Zhang, Jingwei Sun, Li Feng, Cen Yao, Mingming Fan, Liuxin Zhang, Qiany- ing Wang, Xin Geng, and Yong Rui. 2024. See Widely, Think Wisely: Toward Preprint, 2026, Tan et al. Designing a Generative Multi-agent System to Burst Filter Bubbles. InPro- ceedings of the 2024 CHI Co...

    2024 doi

  42. [50]

    I know I must choose a career, but right now I do not have the motivation to decide (I do not want to do it)

  43. [51]

    Work is not the most important thing in life, so choosing a career does not worry me much

  44. [52]

    4.For me, making decisions is usually very difficult

    I believe I do not need to choose a career now, because time will naturally lead me to the right career choice. 4.For me, making decisions is usually very difficult

  45. [53]

    6.I usually fear failure

    I usually feel my decisions need confirmation and support from professionals or other people I trust. 6.I usually fear failure. 7.I like to do things my own way. Preprint, 2026. 8.I hope entering the career I choose will also solve my personal problems. 9.I believe there is on...

    2026

  46. [54]

    I find making a career decision difficult because I do not know what steps to take

  47. [55]

    I find making a career decision difficult because I do not know what factors to consider

  48. [56]

    I find making a career decision difficult because I do not know how to combine what I know about myself with the different career information I have

  49. [57]

    I find making a career decision difficult because I still do not know which careers interest me

  50. [58]

    I find making a career decision difficult because I am still unsure about my career preferences (e.g., what relationships I want with people, what decision environment I prefer)

  51. [59]

    I find making a career decision difficult because I know too little about my abilities or personality traits

  52. [60]

    I find making a career decision difficult because I do not know how my abilities or personality traits will change in the future

  53. [61]

    I find making a career decision difficult because I know too little about existing occupations or training programs

  54. [62]

    I find making a career decision difficult because I know too little about the characteristics of occupations or training programs I am interested in

  55. [63]

    I find making a career decision difficult because I do not know what occupations will be like in the future

  56. [64]

    I find making a career decision difficult because I do not know how to obtain more information about myself

  57. [65]

    I find making a career decision difficult because I do not know how to obtain ac- curate, up-to-date information about existing occupations and training programs

  58. [66]

    I find making a career decision difficult because I keep changing my career preferences

  59. [67]

    I find making a career decision difficult because information about my abilities or personality traits is contradictory

  60. [68]

    I find making a career decision difficult because information about specific occu- pations or training programs is contradictory

  61. [69]

    I find making a career decision difficult because several occupations are equally attractive and it is hard to choose among them

  62. [70]

    I find making a career decision difficult because I do not like any occupation or training program I could enter

  63. [71]

    I find making a career decision difficult because the occupation I am interested in has a feature that troubles me

  64. [72]

    I find making a career decision difficult because my preferences cannot all be realized in a single occupation

  65. [73]

    I find making a career decision difficult because my skills and abilities do not match the requirements of occupations I am interested in

  66. [74]

    I find making a career decision difficult because people important to me disagree with my career choice

  67. [75]

    I find making a career decision difficult because different important people rec- ommend different careers. 3.2 Workload (NASA–TLX-style dimensions) Items 35–40 mirror NASA-TLX dimensions (mental demand, physical demand, temporal demand, performance, effort, frustration) with ...

  68. [76]

    I felt the mental and perceptual demands of the task (e.g., thinking, deciding, remembering)

  69. [77]

    I felt the physical demands of the task (e.g., clicking, typing, how often I had to operate the interface)

  70. [78]

    38.I felt how successful I was in completing the task (performance)

    I felt how hurried or relaxed the pace of completing the task was (temporal demand; bipolar endpoints in Chinese). 38.I felt how successful I was in completing the task (performance). 39.I felt how hard I had to work to complete the task (effort). 40.I felt frustrated, irritat...

    2026

  71. [79]

    During exploration, I receivedinformationalsupport (e.g., useful information to understand different career paths)

  72. [80]

    During exploration, I receivedemotionalsupport (e.g., feeling understood or accompanied rather than facing job-search problems alone)

  73. [81]

    The career experiences shown in the system’s recommended ordering were relevant and helpful

  74. [82]

    The card presentation helped me quickly understand others’ career experiences

  75. [83]

    job-search experience sharing

    Dialogue with the AI companion helped me understand these experiences more deeply. 46.Chatting with the AI felt natural and easy. 47.The related recommended readings were useful. 48.This system helped me explore careers more effectively. 49.If I had the chance, I would use thi...

    2026

  76. [84]

    Personal narrative (e.g., job-search journey, internship diary, fall recruiting recap, story of how an offer was obtained)

  77. [85]

    Interview experience (e.g., specific interview questions, written-test experience, interview flow debrief)

  78. [86]

    Industry knowledge (e.g., HCI industry trends, portfolio tips, role explainers)

  79. [87]

    is_valid

    Recruiting information (e.g., referral codes, urgent intern hiring, campus recruiting announcements) [Invalid] Pure institutional course-selling ads, low-information spam, or content unrelated to job search, further education, or the HCI field. [Output format] Output one and o...

  80. [88]

    background_info

    "background_info": concise core background (e.g., "psychology major at a highly selective university", "STEM new grad with weaker grades", "QS top-30 bachelor's + master's"). If not mentioned, output ""

  81. [89]

    final_outcome

    "final_outcome": the final outcome in very few words. In deployment, length was capped at roughly 20 Chinese characters; keep the English string comparably short (e.g., "one big-tech offer in hand", "rejected by two major firms", "multiple QS top-100 admits"). If no clear outcome, ""

  82. [90]

    background_tags

    "background_tags": extract 1--3 tags. Prioritize difficulties, disadvantages, or pain points (e.g., ["non-prestige undergrad", "zero internships", "field switcher", "late-cycle search", "no research output"]). If none apply, use distinctive traits

  83. [91]

    struggle_summary

    "struggle_summary": one short sentence summarizing the most anxious or hardest phase of their search and how they got through it. If no struggle is described, one sentence summarizing their profile. [Example output] { "background_info": "Industrial design undergrad from a non-...

  84. [92]

    Thank them for choosing to chat with you

  85. [93]

    In first person, warm and colloquial, greet them

  86. [94]

    I've been there too

    Draw on your own struggle story to express empathy ("I've been there too") so they feel you are a peer

  87. [95]

    Offer one supportive line tailored to their mood and difficulty

  88. [96]

    Infer what they might want to talk about; from their perspective, give 3 short prompts (questions or topics) to start the conversation

  89. [97]

    greeting

    Keep the greeting under ~150 Chinese characters in deployment (keep English concise here). Do not use Markdown. Return JSON only, no other text: { "greeting": "...", "suggested_replies": ["...", "...", "..."] } // Fixed user message paired with the above system prompt: // "Ple...

    2026

  90. [98]

    Don't panic---my resume got rejected so many times I could wrap the planet; even stray dogs side-eyed my code, and I'm still here (even landed an offer)

    Relatedness - Strong empathy; self-disclose from your own story; connect their situation to yours. - Accept their feelings without judgment. - Example tone: "Don't panic---my resume got rejected so many times I could wrap the planet; even stray dogs side-eyed my code, and I'm ...

  91. [99]

    You call that grunt work? That's 'end-to-end coordination and delivery'at a big tech firm---you just need packaging; the base is solid

    Competence --- surface their strengths (priority) - If they sound insecure or lost, avoid lofty lectures; help them notice their own strengths. - Invite one or two small things they did (course project, club, even organizing a game guild) and reframe those as workplace-relevan...

  92. [100]

    you must

    Autonomy - Never command ("you must", "you should"). - Offer options; let them decide. It's OK if they vent or want to check out. - Stay curious; do not steer them to a single "right" decision. - Example tone: "This is exhausting---if you truly can't face the resume today, shu...

Pith tools

Reviewed July 14, 2026 · model on record in the stance chip above.