Pith. sign in

REVIEW 3 major objections 5 minor 78 references

AlterAtlas claims that AI travel planning should shift from one-shot generation to an iterative loop of persona-based simulation, inspection, and revision.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 20:34 UTC pith:X3FQ7RHV

load-bearing objection The simulation-vs-revision confound is real, but the system and the validation paradigm are the contribution — worth a serious referee. the 3 major comments →

arxiv 2607.16565 v1 pith:X3FQ7RHV submitted 2026-07-18 cs.HC

AlterAtlas: Shifting Travel Planning from AI Generation to Validation via Persona-Driven Simulations

William Huang , Ruofei Du , Yang Zhang This is my paper
classification cs.HC
keywords human-AI interactionstravel planningpersona-based simulationitinerary validationlarge language modelsAI simulationsgeospatial groundinginteractive planning
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper tries to establish that AI travel planning fails less at generating routes than at helping people see whether a route will actually work for them. Its solution, AlterAtlas, replaces one-shot itinerary output with an iterative loop: build editable personas for each traveler, simulate how those personas would experience a candidate day step by step, show the resulting perceptions and state changes (fatigue, hunger, happiness), and let users revise both the plan and the persona. The main quantitative evidence is an expert evaluation of 51 matched plan pairs, where simulation-guided revisions raised average plan-persona alignment from 7.784 to 8.275 on a 10-point scale, a statistically significant improvement, while feasibility stayed high. A separate within-subjects study with 11 planners found the workflow surfaced hidden constraints, supported comparison of alternatives, and increased confidence in group plans. If the claim holds, the practical lesson is that travel-planning AI should be built around inspectable simulation artifacts and iterative validation, not one-shot generation.

Core claim

The core discovery, stated on the paper's own terms, is that persona-driven simulation can serve as a transparent validation layer for AI-generated travel plans. AlterAtlas takes an editable persona, derives typed simulation variables such as fatigue, hunger, and caffeine level, and then traverses a geospatially grounded itinerary node by node, generating what the person would perceive and how their state would change at each stop and segment. The resulting trace is turned into an itinerary judgment with localized suggestions for improvement, and users may accept or reject the edits, update the persona, and re-run simulation. Expert raters found the simulation-revised plans better matched th

What carries the argument

The central mechanism is the persona-based route simulator. Each traveler is represented by an editable natural-language persona, which is converted into typed simulation variables with explicit update rules and dependencies (for example, hunger rises with walking time and falls with meal size). The itinerary is grounded as a sequence of nodes — stops with metadata, images, reviews, and walking segments sampled every 400 meters with elevation, distance, and streetview context. A simulation agent steps through these nodes, updates the variables according to the rules, writes a localized perception at each node, and compiles the trace into an itinerary judgment and suggested improvements. This

Load-bearing premise

The whole mechanism rests on the assumption that the simulated perceptions and state updates — fatigue, hunger, happiness — closely match how a real traveler would experience the route; the paper itself notes that simulations may misrepresent geospatial conditions or over-literalize preferences, and only four participants returned for post-travel follow-up.

What would settle it

Have travelers follow an AlterAtlas plan while recording their actual fatigue, hunger, and mood, and compare those traces to the simulated ones at each stop and segment; systematic divergence would mean the suggested improvements are built on uncalibrated states. A cheaper check is a blind expert panel rating initial and revised plans without knowing which was simulation-revised — if the alignment gap disappears, the reported improvement is an artifact of the revision framing.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Travel AI systems should treat generated itineraries as draft artifacts, not endpoints, and provide a visible validation stage before a plan is accepted.
  • Users can express spatial preferences more easily by reacting to a simulated route than by writing a perfect prompt; a simulation trace doubles as a preference-elicitation tool.
  • Because the same route generates different revision suggestions for different personas, validation is inherently a function of who the plan is for, not only of the route geometry.
  • Editable personas can become shareable planning artifacts, supporting group trips where one person tracks everyone's constraints.
  • The same simulation-based validation layer could extend to other spatial planning tasks, such as personalized indoor navigation, where route suitability depends on the traveler's state.

Where Pith is reading between the lines

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

  • The alignment gain mixes two effects: the simulation's diagnostic value and the act of revision itself. A head-to-head comparison against human- or prompt-driven revisions with the same amount of edit effort would isolate the simulation's specific contribution.
  • Since the mechanism only models single-day walking trips, a natural stress test is multi-day or transit-based trips, where fatigue and hunger carry over across days and where route failures compound differently.
  • The paper's own admission that simulations may misrepresent geospatial conditions implies a calibration path: compare simulated state trajectories against sensor or diary data from real travelers on the same route and use the discrepancies to tune the update rules.
  • If simulations are accurate enough, the system can be used backwards as a planning probe: instead of asking 'what route fits this persona?', a user can ask 'what would make this route work?' and get persona edits, which is a different interaction loop worth testing.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. AlterAtlas is an interactive travel-planning system that replaces one-shot AI itinerary generation with a validation-and-revision workflow. Users define editable personas, prioritize POIs, and run LLM-based simulations that produce step-by-step perceptions and state updates (fatigue, hunger, etc.) grounded in geospatial data. The system then generates itinerary judgments and suggestions for improvement. The paper reports a formative study (N=7), an expert evaluation of 51 matched initial/updated itinerary pairs, a within-subjects user study (N=11) against an ecological baseline, and post-travel interviews (N=4). The central quantitative claim is that simulation-guided revision significantly improves expert-rated plan-persona alignment (initial M=7.784 vs updated M=8.275, Wilcoxon p=0.00201, average change +0.492). The user study reports significant improvements in self-reported understanding, constraint identification, and group-plan confidence, with no significant difference in active planning time or usability.

Significance. If the central claim holds, the paper makes a valuable contribution to HCI and AI-assisted planning: shifting from generation to inspectable, persona-driven validation is a plausible and well-motivated interaction paradigm. The system is fully implemented, with prompts documented in an appendix, and the expert evaluation uses a matched-pair design with blinded, consensus-rated judgments and high inter-rater reliability. The user study provides rich qualitative evidence for the interactional value of simulations. However, the central causal claim — that the simulation component, rather than the act of revision, drives the alignment improvement — is not isolated by the current experimental design. The paper's significance therefore rests on a specific methodological gap that a revision control could close.

major comments (3)
  1. [§5.1, §5.2] The expert study is described as an 'ablation study' comparing one-shot generation with simulation-based verification, but it conflates simulation with the mere act of LLM-based revision. Each initial itinerary is revised by accepting all suggestions from the simulation judgment; there is no control arm that revises the same initial plan with the same persona but without simulated traces (e.g., a direct 'improve this plan for this persona' prompt). Since the consolidation agent is already an LLM, a second-pass revision could plausibly produce many of the same fixes (reordering, removing burdensome segments, substituting POIs). The observed +0.492 improvement may therefore be attributable to revision per se, and the paper's central claim that persona-driven simulation is the active mechanism is not isolated. A no-simulation revision control is required to support the title and abstract cl
  2. [§5.2] The alignment rating is performed against the same persona description that drives the simulation and the revision suggestions. This creates a mild self-referential component: the updated plan is improved to better match the persona, and is then rated on that match. While not circular in a formal statistical sense, it further underscores the need for a persona-aware but simulation-free control condition. Additionally, the effect size is modest (+0.492 on a 1–10 scale) and the rating comes from only two experts (with consensus resolution). The paper should explicitly discuss how much of the gain might be explained by regression to the mean or by the general benefit of a second LLM pass.
  3. [§6.3, §7] The post-travel interviews (N=4) are used to support claims about simulation fidelity ('simulations were reasonably accurate', 'simulation fidelity'). This is a small, self-selected sample with no control and retrospective self-report. The paper does acknowledge the small sample, but Discussion leans on this evidence more than warranted. Since the entire validation mechanism depends on the accuracy of LLM-generated simulation traces, the manuscript should either present this as purely anecdotal or add a more systematic fidelity check (e.g., comparing simulated states to traveler-reported states during or immediately after the trip).
minor comments (5)
  1. [§5.1] The sentence 'This study can simultaneously be viewed as an ablation study' is misleading because no component is ablated; the initial condition omits the entire revision stage. Reword to 'comparison against a one-shot baseline'.
  2. [§6.1] Typo: 'each participantsself-enhancedAI workflow' should read 'each participant's self-enhanced AI workflow'.
  3. [§7] In Limitations, 'Fourth, While' should be 'Fourth, while'.
  4. [§4.2] The mapping from stated simulation variables (e.g., caffeine_level in the walkthrough) to actual simulation outputs is not shown in Figure 4. Adding a side-by-side of variable update rules and the corresponding trace would improve reproducibility.
  5. [Figure 6] The histogram bins for 'Change in Rating' are not labeled with precise bin edges; please clarify whether the bins are inclusive of endpoints, and consider adding a vertical line at zero.

Circularity Check

0 steps flagged

No circularity found: the central claim is an empirical comparison validated by external expert raters, not a derivation that reduces to its own inputs.

full rationale

AlterAtlas makes no formal derivation chain; its main quantitative claim (Section 5.2, 'Simulation-guided revision improved personalization alignment,' M=7.784 vs M=8.275, Wilcoxon p=0.00201) is an experimental result obtained by having two independent, locally experienced expert raters score 51 matched initial/updated itinerary pairs, blind to condition. The revision pipeline and the rating criterion both reference the same persona, but the raters are external to the system and the improvement could have failed, so it is not forced by construction. The absence of a no-simulation revision control is a real confound (the +0.492 gain may partly reflect revision per se), but that is an alternative-explanation/validity threat, not a definitional circularity. Related-work citations to the authors' own prior systems (e.g., Accessibility Scout [28], Geollery [25], Bring Environments to People [22], Embodied Exploration [45]) are used for positioning, not as load-bearing justification; there is no imported uniqueness theorem or ansatz. The limitation statement that simulations 'may misrepresent geospatial conditions, over-literalize user preferences' is a fidelity caveat about the simulation mechanism, not evidence that the evaluation metric is identical to the system's internal objective. No quoted step exhibits the required reduction of a prediction to its input by construction, so the correct finding is no significant circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 3 axioms · 1 invented entities

The paper does not introduce physical constants or fitted scientific parameters. Its load-bearing assumptions are about the fidelity of LLM simulations, the validity of expert ratings, and the realism of synthetic personas. The only invented entity is the set of freeform simulation variables, which are user-customizable and lack independent validation.

axioms (3)
  • domain assumption LLM-based simulation traces accurately reflect how a traveler with a given persona would experience a route.
    The entire validation layer (Section 4.2, Simulation-based validation) relies on this; if simulations are inaccurate, suggested improvements are not meaningful.
  • domain assumption Expert ratings of feasibility and alignment are valid proxy measures for travel plan quality.
    The expert evaluation (Section 5) uses two local-area experts' Likert ratings; no objective outcome measure is available.
  • domain assumption Synthetic personas generated by Gemini are representative of real travelers' needs.
    Expert evaluation uses 51 synthetic personas (Section 5.1); if these are unrealistic, the alignment improvements may not generalize.
invented entities (1)
  • Simulation variables (e.g., fatigue, hunger, caffeine_level) with update rules and dependencies no independent evidence
    purpose: To model traveler state changes over an itinerary and expose constraints to the user.
    These variables are user-editable and generated by LLMs; they are not validated against physiological or behavioral measurements, and the paper provides no independent falsifiable predictions from them.

pith-pipeline@v1.3.0-alltime-deepseek · 170 in / 6683 out tokens · 126925 ms · 2026-08-01T20:34:07.721533+00:00 · methodology

0 comments
read the original abstract

Travel planning requires balancing interacting goals and constraints across time and space. Current AI travel tools provide limited support for encoding these constraints and understanding how generated travel plans may fail users. We present AlterAtlas, an interactive travel planning system that supports high-fidelity itinerary validation and revision through persona-based simulations grounded in geospatial information. AlterAtlas models travelers as editable personas, generates candidate itineraries from prioritized places of interest, and simulates how different personas would experience each plan. Simulations expose route-level tradeoffs, temporal user states (e.g., fatigue, hunger), and mismatches between plans and user preferences to allow users to iteratively refine both itineraries and user personas. An expert evaluation of 51 paired itineraries demonstrates that simulation-guided revisions significantly improve plan-persona alignment. Furthermore, a within-subjects study (N=11) reveals that AlterAtlas empowers users to uncover hidden constraints, fluidly compare alternatives, and build trust in their final plans. Our results suggest that simulation-based validation is a powerful, transparent interaction layer for AI-assisted travel planning.

Figures

Figures reproduced from arXiv: 2607.16565 by Ruofei Du, William Huang, Yang Zhang.

Figure 1
Figure 1. Figure 1: AlterAtlas shifts AI travel planning from one-shot generation to interactive validation by simulating how different [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: AlterAtlas setup interfaces. a) Trip parameter setup [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Discovery user interface. Users are shown recom [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Simulation user interface. a) Users begin by generating travel plans and simulations which are shown in an interactive [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Example simulation trace for and persona states for the “Chris” persona described in Section 4.1 for a generated day [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Distribution of paired score changes between initial [PITH_FULL_IMAGE:figures/full_fig_p007_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Post-study survey comparison between baseline [PITH_FULL_IMAGE:figures/full_fig_p008_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Suggested improvements for a travel itinerary from the personas of P3, P6, P10, and P11 generated through simulations [PITH_FULL_IMAGE:figures/full_fig_p009_8.png] view at source ↗

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

78 extracted references · 19 canonical work pages · 3 internal anchors

  1. [1]

    [n. d.]. https://developers.google.com/maps

  2. [2]

    [n. d.]. Booking.Com Enhances Travel Planning with New AI-Powered Features for Easier, Smarter Decisions. https://news.booking.com/bookingcom-enhances- travel-planning-with-new-ai-powered-features–for-easier-smarter-decisions/

  3. [3]

    [n. d.]. FastAPI. https://fastapi.tiangolo.com/

  4. [4]

    [n. d.]. Meet Layla: AI Trip Planner 2025. https://layla.ai

    2025

  5. [5]

    [n. d.]. Pydantic AI. https://ai.pydantic.dev/

  6. [6]

    [n. d.]. Tavily. https://www.tavily.com/

  7. [7]

    [n. d.]. Trip Genie: Personalized Travel Itinerary Planner. https://www.trip- genie.com/

  8. [8]

    [n. d.]. Vue.Js. https://vuejs.org/

  9. [9]

    How We’re Reimagining Maps with Gemini

    2026. How We’re Reimagining Maps with Gemini. https://blog.google/products- and-platforms/products/maps/ask-maps-immersive-navigation/

    2026

  10. [10]

    Agag and Ahmed A

    Gomaa M. Agag and Ahmed A. El-Masry. 2017. Why Do Consumers Trust Online Travel Websites? Drivers and Outcomes of Consumer Trust toward Online Travel Websites.Journal of Travel Research56, 3 (March 2017), 347–369. doi:10.1177/0047287516643185

    work page doi:10.1177/0047287516643185 2017

  11. [11]

    Erke Aribas and Evren Daglarli. 2024. Transforming Personalized Travel Recom- mendations: Integrating Generative AI with Personality Models.Electronics13, 23 (Nov. 2024). doi:10.3390/electronics13234751

    work page doi:10.3390/electronics13234751 2024

  12. [12]

    Jinan Azem, Farhan Ahmed, Joni Salminen, and Jim Jansen. 2025. Interviewing AI-Generated Personas: Talking To Your Data to Generate Qualitative Text from Users. 77–87. doi:10.1109/FLLM67465.2025.11390880

    arXiv 2025

  13. [13]

    Nikolas Belle, Dakota Barnes, Alfonso Amayuelas, Ivan Bercovich, Xin Eric Wang, and William Wang. 2025. Agents of Change: Self-Evolving LLM Agents for Strategic Planning. arXiv:2506.04651 [cs] doi:10.48550/arXiv.2506.04651

    doi:10.48550/arxiv.2506.04651 2025

  14. [14]

    Karim Benharrak, Tim Zindulka, Florian Lehmann, Hendrik Heuer, and Daniel Buschek. 2024. Writer-Defined AI Personas for On-Demand Feedback Generation. InProceedings of the CHI Conference on Human Factors in Computing Systems. 1–18. arXiv:2309.10433 [cs] doi:10.1145/3613904.3642406

    Pith/arXiv arXiv 2024

  15. [15]

    Kenny Andres Bermejo-Xiao and Ixent Galpin. 2025. From Photos to Places: Generating Personalized Tourist Itineraries with Large Language Models and Image Analysis. InProceedings of the 6th International Workshop on Applied Artificial Intelligence (W AAI 2025), Workshops at the 8th International Conference on Applied Informatics (ICAIW 2025) (CEUR Workshop...

    2025

  16. [16]

    Nicholas Bolten, Amirhossein Amini, Yun Hao, Vaishnavi Ravichandran, Andre Stephens, and Anat Caspi. 2016. Urban Sidewalks: Visualization and Routing for Individuals with Limited Mobility. arXiv:1602.04281 [cs] doi:10.48550/arXiv. 1602.04281

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1602.04281 2016

  17. [17]

    Nicolas Bougie and Narimasa Watanabe. 2025. CitySim: Modeling Urban Be- haviors and City Dynamics with Large-Scale LLM-Driven Agent Simulation. arXiv:2506.21805 [cs] doi:10.48550/arXiv.2506.21805

    Pith/arXiv arXiv doi:10.48550/arxiv.2506.21805 2025

  18. [18]

    Virginia Braun and Victoria Clarke. 2019. Reflecting on Reflexive Thematic Analysis.Qualitative Research in Sport, Exercise and Health11, 4 (Aug. 2019), 589–597. doi:10.1080/2159676X.2019.1628806

    arXiv 2019

  19. [19]

    Aili Chen, Xuyang Ge, Ziquan Fu, Yanghua Xiao, and Jiangjie Chen. 2024. Trave- lAgent: An AI Assistant for Personalized Travel Planning. arXiv:2409.08069 [cs] doi:10.48550/arXiv.2409.08069

    Pith/arXiv arXiv doi:10.48550/arxiv.2409.08069 2024

  20. [20]

    Chaoran Chen, Leyang Li, Luke Cao, Yanfang Ye, Tianshi Li, Yaxing Yao, and Toby Jia-jun Li. 2025. Why Am I Seeing This: Democratizing End User Auditing for Online Content Recommendations. arXiv:2410.04917 [cs] doi:10.48550/arXiv. 2410.04917

    Pith/arXiv arXiv doi:10.48550/arxiv.2410.04917 2025

  21. [21]

    Xiang Cheng, Yulan Hu, Xiangwen Zhang, Lu Xu, Zheng Pan, Xin Li, and Yong Liu. 2026. TravelBench: A Broader Real-World Benchmark for Multi-Turn and Tool-Using Travel Planning. arXiv:2512.22673 [cs] doi:10.48550/arXiv.2512.22673

    Pith/arXiv arXiv doi:10.48550/arxiv.2512.22673 2026

  22. [22]

    Hao-Yun Chi, Jingzhen Sha, and Yang Zhang. 2023. Bring Environments to People – A Case Study of Virtual Tours in Accessibility Assessment for People with Limited Mobility. InProceedings of the 20th International Web for All Conference (W4A ’23). Association for Computing Machinery, New York, NY, USA, 96–103. doi:10.1145/3587281.3587292

    arXiv 2023

  23. [23]

    Youngjoon Choi, Jaeseok Lee, and Yeongbae Choe. 2021. Online Travel In- formation Filtering: Role of Commercial Cues in Trust and Distrust Mech- anisms.Journal of Travel & Tourism Marketing38, 7 (Sept. 2021), 710–724. Huang et al. doi:10.1080/10548408.2021.1985037

    arXiv 2021

  24. [24]

    Jieren Deng, Zhizhang Hu, Ziyan He, Aleksandar Cvetkovic, Pak Kiu Chung, Dragomir Yankov, and Chiqun Zhang. 2025. IMAIA: Interactive Maps AI As- sistant for Travel Planning and Geo-Spatial Intelligence. arXiv:2507.06993 [cs] doi:10.48550/arXiv.2507.06993

    work page doi:10.48550/arxiv.2507.06993 2025

  25. [25]

    Ruofei Du, David Li, and Amitabh Varshney. 2019. Geollery: A Mixed Reality Social Media Platform. InProceedings of the 2019 CHI Conference on Human Factors in Computing Systems. ACM, Glasgow Scotland Uk, 1–13. doi:10.1145/ 3290605.3300915

    arXiv 2019

  26. [26]

    Ankit Garg, Deepti Verma, Laxmi Pandey, K.Suresh Kumar, Rashmi Singh, and Anubhav Sharma. 2025. Enhancing Tailored Travel by Integrating Generative AI with Insights Driven by Personality. In2025 International Conference on Intelligent Control, Computing and Communications (IC3). 404–409. doi:10.1109/IC363308. 2025.10957442

    arXiv 2025

  27. [27]

    Juhye Ha, Hyeon Jeon, Daeun Han, Jinwook Seo, and Changhoon Oh. 2024. CloChat: Understanding How People Customize, Interact, and Experience Per- sonas in Large Language Models. InProceedings of the 2024 CHI Conference on Human Factors in Computing Systems(Honolulu, HI, USA)(CHI ’24). Associ- ation for Computing Machinery, New York, NY, USA, Article 305, 2...

    arXiv 2024

  28. [28]

    Froehlich, and Yang Zhang

    William Huang, Xia Su, Jon E. Froehlich, and Yang Zhang. 2025. Accessibility Scout: Personalized Accessibility Scans of Built Environments. InProceedings of the 38th Annual ACM Symposium on User Interface Software and Technology. 1–18. arXiv:2507.23190 [cs] doi:10.1145/3746059.3747624

    arXiv 2025

  29. [29]

    Zhuoxuan Huang, Yunshan Ma, Hongyu Zhang, Hua Ma, and Zhu Sun. 2026. iTIMO: An LLM-empowered Synthesis Dataset for Travel Itinerary Modification. arXiv:2601.10609 [cs] doi:10.48550/arXiv.2601.10609

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2601.10609 2026

  30. [30]

    Soon-Gyo Jung, Joni Salminen, Kholoud Khalil Aldous, and Bernard J. Jansen

  31. [31]

    Azem, João M

    Ilkka Kaate, Joni Salminen, Soon-Gyo Jung, Trang Thi Thu Xuan, Jinan Y. Azem, João M. Santos, and Bernard J Jansen. 2025. When Personas Talk to You: Eval- uating the Evolution of User Personas from Static Profiles to Conversational User Interfaces. InProceedings of the 2025 ACM Designing Interactive Systems Conference. ACM, Madeira Portugal, 2350–2372. do...

    arXiv 2025

  32. [32]

    Joshi, Hanna Moazam, Heather Miller, Matei Zaharia, and Christopher Potts

    Omar Khattab, Arnav Singhvi, Paridhi Maheshwari, Zhiyuan Zhang, Keshav San- thanam, Sri Vardhamanan, Saiful Haq, Ashutosh Sharma, Thomas T. Joshi, Hanna Moazam, Heather Miller, Matei Zaharia, and Christopher Potts. 2023. DSPy: Compiling Declarative Language Model Calls into Self-Improving Pipelines. https://arxiv.org/abs/2310.03714v1

    Pith/arXiv arXiv 2023

  33. [33]

    Hikaru Kumamoto and Hideo Joho. 2025. Exploring the Role of Large Language Model in Collaborative Travel Planning Task.Information Research: An Interna- tional Electronic Journal30, iConf (March 2025). doi:10.47989/ir30iconf46966

    work page doi:10.47989/ir30iconf46966 2025

  34. [34]

    Simon Lämmer, Mark Colley, and Patrick Ebel. 2026. GTA: Generative Traffic Agents for Simulating Realistic Mobility Behavior. arXiv:2601.16778 [cs] doi:10. 1145/3772318.3790772

    arXiv 2026

  35. [35]

    James R. Lewis. 2018. The System Usability Scale: Past, Present, and Future. International Journal of Human-Computer Interaction34, 7 (2018), 577–590. doi:10. 1080/10447318.2018.1455307

    arXiv 2018

  36. [36]

    Juan Liu, Chaohui Wang, Shimin Fang, and Tingting Zhang. 2019. Scale Devel- opment for Tourist Trust toward a Tourism Destination.Tourism Management Perspectives31 (July 2019), 383–397. doi:10.1016/j.tmp.2019.07.001

    work page doi:10.1016/j.tmp.2019.07.001 2019

  37. [37]

    Tianhui Liu, Jie Feng, Hetian Pang, Xin Zhang, Tianjian Ouyang, Zhiyuan Zhang, and Yong Li. 2025. CityLens: Benchmarking Large Language-Vision Models for Urban Socioeconomic Sensing. arXiv:2506.00530 [cs] doi:10.48550/arXiv.2506. 00530

    work page doi:10.48550/arxiv.2506.00530 2025

  38. [38]

    Yuxuan Lu, Bingsheng Yao, Hansu Gu, Jing Huang, Jessie Wang, Yang Li, Jiri Gesi, Qi He, Toby Jia-Jun Li, and Dakuo Wang. 2025. UXAgent: A System for Simulating Usability Testing of Web Design with LLM Agents. arXiv:2504.09407 [cs] doi:10. 48550/arXiv.2504.09407

    doi:10.48550/arxiv.2504.09407 2025

  39. [39]

    Rashid Mushkani. 2025. Do Vision-Language Models See Urban Scenes as People Do? An Urban Perception Benchmark. arXiv:2509.14574 [cs] doi:10.48550/arXiv. 2509.14574

    doi:10.48550/arxiv.2509.14574 2025

  40. [40]

    Keisuke Otaki and Yukino Baba. 2025. Travel Itinerary Recommendation Using Interaction-Based Augmented Data.Expert Systems with Applications269 (April 2025), 126294. doi:10.1016/j.eswa.2024.126294

    arXiv 2025

  41. [41]

    Song, Xiaojuan Ma, and Juho Kim

    Jeongeon Park, Bryan Min, Kihoon Son, Jean Y. Song, Xiaojuan Ma, and Juho Kim. 2026. ChoiceMates: Supporting Unfamiliar Online Decision-Making with Multi-Agent Conversational Interactions. arXiv:2310.01331 [cs] doi:10.48550/ arXiv.2310.01331

    work page doi:10.48550/arxiv.2310.01331 2026

  42. [42]

    Jae Eun Park, Alei Fan, and Kevin Kam Fung So. 2025. Enhancing the Effective- ness of Generative AI Travel Recommendations: Balancing Source Credibility and Cognitive Load.Journal of Hospitality & Tourism Research(Nov. 2025), 10963480251401822. doi:10.1177/10963480251401822

    work page doi:10.1177/10963480251401822 2025

  43. [43]

    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. arXiv:2304.03442 [cs] doi:10.48550/arXiv.2304.03442

    Pith/arXiv arXiv doi:10.48550/arxiv.2304.03442 2023

  44. [44]

    Zou, Aaron Shaw, Benjamin Mako Hill, Carrie Cai, Meredith Ringel Morris, Robb Willer, Percy Liang, and Michael S

    Joon Sung Park, Carolyn Q. Zou, Aaron Shaw, Benjamin Mako Hill, Carrie Cai, Meredith Ringel Morris, Robb Willer, Percy Liang, and Michael S. Bernstein

  45. [45]

    Siyou Pei, Alexander Chen, Chen Chen, Franklin Mingzhe Li, Megan Fozzard, Hao-Yun Chi, Nadir Weibel, Patrick Carrington, and Yang Zhang. 2023. Embod- ied Exploration: Facilitating Remote Accessibility Assessment for Wheelchair Users with Virtual Reality. InProceedings of the 25th International ACM SIGAC- CESS Conference on Computers and Accessibility (ASS...

    arXiv 2023

  46. [46]

    Joan Perez and Giovanni Fusco. 2025. Streetscape Analysis with Genera- tive AI (SAGAI): Vision-Language Assessment and Mapping of Urban Scenes. arXiv:2504.16538 [cs] doi:10.48550/arXiv.2504.16538

    work page doi:10.48550/arxiv.2504.16538 2025

  47. [47]

    Jinghua Piao, Yuwei Yan, Jun Zhang, Nian Li, Junbo Yan, Xiaochong Lan, Zhihong Lu, Zhiheng Zheng, Jing Yi Wang, Di Zhou, Chen Gao, Fengli Xu, Fang Zhang, Ke Rong, Jun Su, and Yong Li. 2025. AgentSociety: Large-Scale Simulation of LLM-Driven Generative Agents Advances Understanding of Human Behaviors and Society. arXiv:2502.08691 [cs] doi:10.48550/arXiv.2502.08691

    Pith/arXiv arXiv doi:10.48550/arxiv.2502.08691 2025

  48. [48]

    Riesbeck and Roger C

    Christopher K. Riesbeck and Roger C. Schank. 2013.Inside Case-Based Reasoning. Psychology Press, New York. doi:10.4324/9780203781821

    work page doi:10.4324/9780203781821 2013

  49. [49]

    Muhammed Rizwan, Lars Carlsson, and Mohammad Loni. 2025. PersonaBOT: Bringing Customer Personas to Life with LLMs and RAG. arXiv:2505.17156 [cs] doi:10.48550/arXiv.2505.17156

    Pith/arXiv arXiv doi:10.48550/arxiv.2505.17156 2025

  50. [50]

    Glynn, and Mei-Ling T

    Bernard Rosner, Robert J. Glynn, and Mei-Ling T. Lee. 2006. The Wilcoxon Signed Rank Test for Paired Comparisons of Clustered Data.Biometrics62, 1 (March 2006), 185–192. doi:10.1111/j.1541-0420.2005.00389.x

    arXiv 2006

  51. [51]

    S, Nivash

    Nithyasri Prabha. S, Nivash. V, Yogitha. R, G. Kalaiarasi, and M. Selvi. 2024. Personalized Travel Itinerary Generator System. In2024 International Conference on Intelligent Systems for Cybersecurity (ISCS). 1–6. doi:10.1109/ISCS61804.2024. 10581212

    arXiv 2024

  52. [52]

    Manaswi Saha, Michael Saugstad, Hanuma Teja Maddali, Aileen Zeng, Ryan Holland, Steven Bower, Aditya Dash, Sage Chen, Anthony Li, Kotaro Hara, and Jon Froehlich. 2019. Project Sidewalk: A Web-based Crowdsourcing Tool for Collecting Sidewalk Accessibility Data At Scale. InProceedings of the 2019 CHI Conference on Human Factors in Computing Systems (CHI ’19...

    arXiv 2019

  53. [53]

    Yuanzhe Shen, Kaimin Wang, Changze Lv, Xiaoqing Zheng, and Xuanjing Huang

  54. [54]

    Hakseung Shin, Juan Luis Nicolau, Juhyun Kang, Abhinav Sharma, and Hoon Lee. 2022. Travel Decision Determinants during and after COVID-19: The Role of Tourist Trust, Travel Constraints, and Attitudinal Factors.Tourism Management 88 (Feb. 2022), 104428. doi:10.1016/j.tourman.2021.104428

    arXiv 2022

  55. [55]

    Hedderich, Bartłomiej Jakub Rey, Andrés Lucero, and Antti Oulasvirta

    Joongi Shin, Michael A. Hedderich, Bartłomiej Jakub Rey, Andrés Lucero, and Antti Oulasvirta. 2024. Understanding Human-AI Workflows for Generating Per- sonas. InDesigning Interactive Systems Conference. ACM, Copenhagen Denmark, 757–781. doi:10.1145/3643834.3660729

    arXiv 2024

  56. [56]

    arXiv:2508.01432 [cs] doi:10.48550/arXiv.2508.01432

    TripTailor: A Real-World Benchmark for Personalized Travel Planning. arXiv:2508.01432 [cs] doi:10.48550/arXiv.2508.01432

    Pith/arXiv arXiv doi:10.48550/arxiv.2508.01432

  57. [57]

    Vogt Susan I

    Christine A. Vogt Susan I. Stewart. 1999. A Case-Based Approach to Un- derstanding Vacation Planning.Leisure Sciences21, 2 (June 1999), 79–95. doi:10.1080/014904099273165

    work page doi:10.1080/014904099273165 1999

  58. [58]

    Vikranth Udandarao, Noel Abraham Tiju, Muthuraj Vairamuthu, Harsh Mis- try, and Dhruv Kumar. 2025. Roamify: Designing and Evaluating an LLM Based Google Chrome Extension for Personalised Itinerary Planning. arXiv:2504.10489 [cs] doi:10.48550/arXiv.2504.10489

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2504.10489 2025

  59. [59]

    Akshat Singh, Raksha Madhogaria, Abhishek Misra, and E. Elakiya. 2024. Au- tomated Travel Planning Via Multi-Agent Systems and Real-Time Intelligence. social science research network:5089025 doi:10.2139/ssrn.5089025

    work page doi:10.2139/ssrn.5089025 2024

  60. [60]

    Dakuo Wang, Ting-Yao Hsu, Yuxuan Lu, Hansu Gu, Limeng Cui, Yaochen Xie, William Headean, Bingsheng Yao, Akash Veeragouni, Jiapeng Liu, Sreyashi Nag, and Jessie Wang. 2025. AgentA/B: Automated and Scalable Web A/BTesting with Interactive LLM Agents. arXiv:2504.09723 [cs] doi:10.48550/arXiv.2504.09723

    doi:10.48550/arxiv.2504.09723 2025

  61. [61]

    Lei Wang, Chen Ma, Xueyang Feng, Zeyu Zhang, Hao Yang, Jingsen Zhang, Zhiyuan Chen, Jiakai Tang, Xu Chen, Yankai Lin, Wayne Xin Zhao, Zhewei Wei, and Jirong Wen. 2024. A Survey on Large Language Model Based Autonomous Agents.Frontiers of Computer Science18, 6 (Dec. 2024), 186345. doi:10.1007/ s11704-024-40231-1

    2024

  62. [62]

    Katerina Volchek and Stanislav Ivanov. 2024. ChatGPT as a Travel Itinerary Plan- ner. InInformation and Communication Technologies in Tourism 2024, Katerina Berezina, Lyndon Nixon, and Aarni Tuomi (Eds.). Springer Nature Switzerland, Cham, 365–370. doi:10.1007/978-3-031-58839-6_38

    work page doi:10.1007/978-3-031-58839-6_38 2024

  63. [63]

    Jian Xie, Kai Zhang, Jiangjie Chen, Tinghui Zhu, Renze Lou, Yuandong Tian, Yanghua Xiao, and Yu Su. 2024. TravelPlanner: A Benchmark for Real-World Planning with Language Agents. arXiv:2402.01622 [cs] doi:10.48550/arXiv.2402. 01622 AlterAtlas: Shifting Travel Planning from AI Generation to Validation via Persona-Driven Simulations

    Pith/arXiv arXiv doi:10.48550/arxiv.2402 2024

  64. [64]

    Zamfirescu-Pereira, Richmond Y

    J.D. Zamfirescu-Pereira, Richmond Y. Wong, Bjoern Hartmann, and Qian Yang

  65. [65]

    Yaobin Wang, Haoyang Miao, Meiling Xiong, and Yuxiao Wang. 2025. How Does ChatGPT-generated Information Influence Travel Planning? The Mediating Role of Inspiration.Tourism Review81, 2 (June 2025), 694–717. doi:10.1108/TR-03- 2025-0235

    work page doi:10.1108/tr-03- 2025

  66. [66]

    Qianyu Zhou, Jiaxin Zhang, and Zehong Zhu. 2025. Evaluating Urban Visual Attractiveness Perception Using Multimodal Large Language Model and Street View Images.Buildings15, 16 (Aug. 2025). doi:10.3390/buildings15162970

    work page doi:10.3390/buildings15162970 2025

  67. [67]

    Zhangxiang Zhu and Kening Yang. 2023. Virtual Community Users’ Travel Information-Seeking Continuance Intention: An Empirical Study Based on the Cognitive Load Theory and Flow Theory.Journal of Systems and Information Technology25, 1 (March 2023), 91–108. doi:10.1108/JSIT-07-2022-0178

    work page doi:10.1108/jsit-07-2022-0178 2023

  68. [68]

    Jianke Zou, Kai Zhang, Yiping Sun, and Jianke Zou. 2024. Research on Travel Route Planning Optimization Based on Large Language Model. preprints:2024071398 doi:10.20944/preprints202407.1398.v1 Huang et al. A System Prompts This section documents the prompts AlterAtlas uses for its multi- agent workflow. Prompts are documented as either DSPy signatures or ...

    arXiv 2024

  69. [69]

    Mills, and Hui-Wen Huang

    Jianhao Zhang, Daniel J. Mills, and Hui-Wen Huang. 2024. Enhancing Travel Planning and Experiences with Multimodal ChatGPT 4.0. InProceedings of the 2024 International Conference on Innovation in Artificial Intelligence (ICIAI ’24). Association for Computing Machinery, New York, NY, USA, 12–19. doi:10.1145/ 3655497.3655529

    arXiv 2024

  70. [75]

    food in Rome

    tavily_extract : Extracts information from a web page . It accepts a URL and returns extracted information from the page . CRITICAL - For all recommendations , you MUST return : - Justification grounded in tool results - a short summary of the place - The specific web search and place search queries used - All relevant URLs used to craft the query for thi...

  71. [76]

    search_places : Searches for real - world places ( e . g . , restaurants , attractions ) . It returns place details and a pre - written summary

  72. [77]

    It accepts a search query and returns a list of results

    tavily_search : Searches the web for information about a specific topic . It accepts a search query and returns a list of results

  73. [78]

    It accepts a URL and returns extracted information from the page

    tavily_extract : Extracts information from a web page . It accepts a URL and returns extracted information from the page

  74. [79]

    It accepts an optional'travelMode'parameter ('DRIVE','WALK', or'TWO_WHEELER')

    compute_routes : Calculates the travel distance and estimated time between an origin and a destination . It accepts an optional'travelMode'parameter ('DRIVE','WALK', or'TWO_WHEELER') . You must always use'WALK'. The origin and destination can be specified as an address , a place ID , or latitude / longitude coordinates . Huang et al

  75. [80]

    Plan a walk including the Colosseum and Trevi Fountain

    plot_map : Plots markers and walking paths on a map for visualization . It accepts longitude and latitudes and optional waypoint coordinates and returns a rendered map reference . CRITICAL - You must not invent place_ids . All places must originate from tool calls . CRITICAL - For any multi - tool query ( e . g . , search_places + compute_routes , or tavi...

  76. [2023]

    InProceedings of the 2023 CHI Conference on Human Factors in Computing Systems (CHI ’23)

    Why Johnny Can’t Prompt: How Non-AI Experts Try (and Fail) to Design LLM Prompts. InProceedings of the 2023 CHI Conference on Human Factors in Computing Systems (CHI ’23). Association for Computing Machinery, New York, NY, USA, 1–21. doi:10.1145/3544548.3581388

    arXiv 2023

  77. [2024]

    arXiv:2411.10109 [cs] doi:10.48550/arXiv.2411.10109

    Generative Agent Simulations of 1,000 People. arXiv:2411.10109 [cs] doi:10.48550/arXiv.2411.10109

    Pith/arXiv arXiv doi:10.48550/arxiv.2411.10109

  78. [2025]

    doi:10.1016/j.ijhcs.2025.103445

    PersonaCraft: Leveraging Language Models for Data-Driven Persona Development.International Journal of Human-Computer Studies197 (March 2025), 103445. doi:10.1016/j.ijhcs.2025.103445

    arXiv 2025