Pith. sign in

REVIEW 2 major objections 5 minor 40 references

A hybrid LLM-ABM framework that updates contact networks from group mobility decisions better matches COVID-19 peak timing and size than static ABM alone.

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 · grok-4.5

2026-07-10 21:56 UTC pith:6CFGIGJI

load-bearing objection Solid engineering of a million-agent hybrid ABM-LLM loop, but the headline comparison rests on an unmatched baseline and an unvalidated behavioral model. the 2 major comments →

arxiv 2607.06757 v1 pith:6CFGIGJI submitted 2026-07-07 cs.AI cs.MA

LLM-powered reasoning in agent-based modeling

Sifat Afroj Moon , Dakotah Maguire , Adam Spannaus , Joe Tuccillo , Maksudul Alam , Sudip K. Seal , John Gounley , Heidi Hanson This is my paper
classification cs.AI cs.MA
keywords large language modelsagent-based modelingactivity-based networkindividual-based network modeldigital twinCOVID-19epidemic simulationhybrid ABM-LLM
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.

Agent-based epidemic models can represent millions of people and their contacts, but they usually start from fixed activity schedules that never change when an outbreak worsens. The authors introduce HALE, a hybrid system that keeps a full-scale individual network simulation while periodically asking an LLM, for 1,552 demographic-spatial groups, whether members of that group will keep going out. The LLM answers with structured yes/no decisions that deactivate the corresponding outdoor links for the next week. On a COVID-19 simulation of Salt Lake County from September 2021 to February 2022, the resulting curves track observed peak timing more closely and produce a total attack rate consistent with known asymptomatic fractions, whereas pure ABM runs with the same parameters overestimate both peak and size even after random contact reductions. The work therefore offers a practical way to close the mobility-data gap without replacing the ABM engine.

Core claim

When LLM-generated group-level decisions to stay home are used as a feedback loop that prunes outdoor edges in an activity-based temporal contact network, the resulting epidemic trajectories for COVID-19 in Salt Lake County align more closely with observed weekly cases in both peak timing and cumulative size than otherwise identical ABM-only simulations.

What carries the argument

The HALE framework: a hybrid loop in which an individual-based SIR ABM advances on daily weighted activity networks while, every week, 1,552 LLM agents (grouped by municipality, sex, race and age) return structured yes/no mobility decisions that deactivate the corresponding outdoor contacts for that week.

Load-bearing premise

The zero-shot answers of a single 8-billion-parameter language model, at temperature 0.2, correctly capture how real demographic groups change their outdoor activity when local infection rates rise.

What would settle it

Compare the LLM-derived weekly stay-home probabilities for each age-sex-municipality group against independent mobility traces (cell-phone, survey, or transit data) for Salt Lake County over the same months; systematic mismatch would falsify the behavioral premise.

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

If this is right

  • Population-scale digital twins can incorporate near-real-time behavioral adaptation without requiring continuous high-resolution mobility surveys.
  • Public-health scenario analysis can test how different demographic groups respond to incidence signals rather than assuming uniform compliance.
  • The same hybrid pattern can be reused for other contagion or diffusion processes that depend on deliberate outdoor activity.
  • Network construction methods that fill missing activity locations with spatial kernels become more useful once those edges can be dynamically switched off by LLM feedback.

Where Pith is reading between the lines

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

  • If the LLM stay-home probabilities prove reliable, the same grouping scheme could be applied to other counties or countries with only modest re-prompting.
  • The gradual rise in decline probability across age groups suggests that LLMs may encode a form of social-memory dynamics that pure ABMs usually must hard-code.
  • The near-zero response for tiny unincorporated places implies that geographic specificity in the prompt is itself a first-order control on model fidelity.
  • Coupling weather or policy announcements into the same weekly LLM query would be a low-cost next experiment that the architecture already supports.

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

2 major / 5 minor

Summary. The paper introduces HALE, a hybrid framework that couples a large-scale activity-based temporal contact network ABM (SIR on ~1.1M synthetic agents for Salt Lake County) with weekly zero-shot Llama-3.1-8B inferences over 1,552 demographic-spatial groups. LLM yes/no decisions on outdoor activity selectively deactivate deliberate edges in G(t), allowing mobility to respond to simulated incidence. As a proof-of-concept for COVID-19 (Sept 2021–Feb 2022, Omicron parameters), the mean of 30 HALE runs better matches CDC peak timing and produces a total size ~2.21 imes observed incidence (argued to be consistent with asymptomatic under-ascertainment), while an ABM-only baseline with the same networks and parameters overestimates even after a post-hoc 35% random contact deactivation. Supporting analyses show age- and municipality-differentiated stay-home probabilities that rise gradually with incidence.

Significance. If the comparative claim holds, HALE offers a practical, scalable route to inject adaptive human behavior into population-scale digital twins without replacing every agent by a generative model—an important engineering contribution for policy-oriented epidemic ABMs that currently rely on static survey-derived activity schedules. Strengths that should be credited include the carefully specified network-construction pipeline (anchored vs. non-anchored activities, class-level school contacts, exponential kernels with literature λ values), the HPC server-client integration that keeps LLM inference tractable for millions of agents, the public Zenodo release of code and data, and the explicit ablation-style comparison of HALE vs. ABM-only. These elements make the work a concrete, reproducible step beyond purely conceptual LLM-ABM proposals.

major comments (2)
  1. Results §3.1 and Fig. 3: the central claim that HALE better captures peak timing and size rests on an unmatched ABM-only control. The paper states that 35% of contacts are randomly deactivated “for a fair comparison” so that average degree roughly matches HALE after LLM updates. Random thinning preserves the original activity mix and spatial structure; HALE instead removes whole outdoor-activity classes for entire demographic-spatial groups in an incidence-dependent, time-varying manner (Architecture §2.3.1–2.3.3). Consequently the improvement cannot be attributed to LLM reasoning rather than to any structured reduction of high-transmission activities or to the particular 35% level chosen after inspecting HALE output. A matched ablation that applies the same activity-type or group-level deactivation schedule without LLM input (or a sensitivity sweep over the deactivation fraction) is req
  2. §2.3.3 and Results §3.2: the load-bearing premise that zero-shot Llama-3.1-8B (temperature 0.2, structured yes/no) for 1,552 municipality×sex×race×age bins accurately reflects real group-level mobility responses is never validated against observed mobility, survey, or cell-phone data. The temperature choice itself is acknowledged to be decisive (temperature=0 yields near-certain stay-home; 0.7 yields ~0.5), yet no external calibration or sensitivity analysis is reported. Because every network update after each weekly LLM step depends on these outputs, the epidemic-curve improvement remains under-supported until the behavioral module is checked against independent mobility evidence or at least subjected to a systematic temperature/prompt ablation.
minor comments (5)
  1. Abstract and Introduction state the simulation window as “September 2020 to February 2022,” while Results §3 and Fig. 3 use September 2021–February 2022; the dates should be reconciled throughout.
  2. §2.2.1: the office degree distribution N(µ=21.154/8, σ²=10.58/8) and the out-of-class school average degree of 4 are taken from literature but never sensitivity-tested; a brief note on robustness would strengthen the network-construction claims.
  3. Fig. 5 caption and surrounding text: “Harriman” should be “Herriman”; several municipality names are inconsistently capitalized.
  4. The prompt template in §3.2 is given for a single illustrative agent; a short appendix listing the exact structured-output schema and any system prompt would improve reproducibility.
  5. References [23] (Zenodo) and the funding acknowledgment are present, but the main text never states the exact Llama-3.1-8B checkpoint or vLLM version used; adding these details would complete the methods.

Circularity Check

0 steps flagged

No significant circularity: HALE epidemic curves are generated from pretrained zero-shot LLM decisions plus ABM dynamics and compared to external CDC incidence; the 35% baseline adjustment and temperature choice do not force the main claim by construction.

full rationale

The paper's central empirical claim (HALE better matches observed COVID-19 peak timing and size in Salt Lake County than ABM-only under shared SIR parameters and networks) is an external comparison against CDC weekly case data, not a quantity derived from its own inputs. The LLM component uses a fixed pretrained Llama-3.1-8B model in zero-shot mode with structured yes/no outputs; its weights were not trained or fine-tuned on the Salt Lake County incidence series, so the mobility decisions that update G(t) after each weekly step are independent of the target curves. Network construction from NHTS/UrbanPop data and the SIR update rules are standard and do not define the observed incidence. The post-hoc random 35% edge deactivation in the ABM-only control (chosen to approximate HALE's average degree reduction) and the temperature=0.2 setting (selected for non-degenerate age-group patterns) are modeling choices that affect the strength of the comparative claim, but they do not make HALE's match to CDC data true by construction or reduce any reported prediction to a fitted parameter. No self-definitional equations, load-bearing self-citations of uniqueness results, or renamed known patterns appear. The derivation chain is therefore self-contained against the external benchmark.

Axiom & Free-Parameter Ledger

6 free parameters · 5 axioms · 2 invented entities

The central claim rests on a mixture of standard epidemic-network assumptions, literature-derived contact parameters, and several paper-specific modeling choices (LLM temperature, group granularity, 35% baseline deactivation). No new physical entities are postulated; the main invented constructs are the HALE coupling itself and the 1,552 LLM group agents. Free parameters that directly affect the reported curves are listed exhaustively below.

free parameters (6)
  • LLM temperature = 0.2
    Set to 0.2 after observing that 0 yields near-certain stay-home and 0.7 yields ~0.5 random answers; directly controls the mobility feedback that produces the better peak match.
  • ABM-only contact deactivation fraction = 0.35
    Chosen as 35% so that the baseline has roughly the same average degree reduction as HALE; not derived from independent mobility data.
  • Omicron R0 and recovery rate = R0=9.5, γ=0.2
    Taken from literature (R0=9.5, γ=1/5 day⁻¹) and then converted to per-edge transmissibility using the network’s average degree; any error propagates into both HALE and ABM-only curves.
  • Office contact degree distribution = μ≈2.64, σ²≈1.32
    Gaussian N(μ=21.154/8, σ²=10.58/8) taken from prior contact surveys and scaled to an 8-hour workday; controls workplace transmission.
  • Out-of-class school average degree = 4
    Fixed at 4 contacts per 8-hour school day; arbitrary but literature-motivated choice that affects school transmission.
  • Activity-specific travel distances λ = activity-dependent
    Exponential-kernel parameters (exercise 4 mi, church 11.25 mi, food 2.6 mi, etc.) taken from heterogeneous surveys; unknown activities use the mean of known λs.
axioms (5)
  • ad hoc to paper Zero-shot Llama-3.1-8B can produce demographically and spatially differentiated stay-home decisions that approximate real human responses to local incidence.
    Stated in Section 2.3.3 and used for every network update; never validated against mobility ground truth.
  • domain assumption SIR compartmental dynamics with constant β and γ adequately describe Omicron transmission on the constructed activity networks.
    Standard epidemic-modeling premise adopted in Section 2.3.1 and Results.
  • domain assumption Exponential distance kernel correctly generates contacts for non-anchored activities when exact locations are unknown.
    Invoked in Section 2.2.2 with literature λ values.
  • ad hoc to paper Grouping agents into 1,552 municipality × sex × race × age bins is a sufficient resolution for LLM behavioral inference.
    Defined in Section 2.3.2; smaller cities are not further subdivided.
  • domain assumption Weekly LLM time steps (Δt_LLM = 7 days) match the relevant behavioral adaptation timescale.
    Chosen because CDC data are weekly (Section 2.3.3).
invented entities (2)
  • HALE framework (hybrid ABM-LLM loop with deliberate-edge deactivation) no independent evidence
    purpose: Couples scalable activity-based ABM with batched LLM mobility decisions so that contact networks adapt to simulated incidence.
    The architecture itself is the paper’s main construct; independent evidence is limited to the single Salt Lake County case study.
  • 1,552 LLM group agents (one per demographic-spatial bin) no independent evidence
    purpose: Provide structured yes/no mobility decisions that update the corresponding subpopulation’s network edges.
    Introduced to keep LLM cost tractable while preserving spatial and demographic heterogeneity; no external validation of the grouping.

pith-pipeline@v1.1.0-grok45 · 16729 in / 3611 out tokens · 43972 ms · 2026-07-10T21:56:32.981367+00:00 · methodology

0 comments
read the original abstract

Agent-based modeling (ABM) has the capability to model millions of individuals and their interactions, which is useful for policy making. However, ABMs have traditionally relied on static prior, which prevents the models from adapting to real-time changes. Our research provides a novel approach to addressing this information gap. Large language models (LLMs) offer new opportunities to predict human decision-making. Here, we introduce a scalable Hybrid Agent-based and Language-driven Epidemic (HALE) modeling framework that leverages LLMs to predict human decision-making in an ABM simulation. As a proof-of-concept, we use HALE to simulate COVID-19 and its effects in Salt Lake County, UT.

Figures

Figures reproduced from arXiv: 2607.06757 by Adam Spannaus, Dakotah Maguire, Heidi Hanson, Joe Tuccillo, John Gounley, Maksudul Alam, Sifat Afroj Moon, Sudip K. Seal.

Figure 1
Figure 1. Figure 1: The HALE framework—scalable agent-based simulation with LLM reasoning—is applied [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Spatial LLM grouping for Salt Lake County, UT. We use race, sex, and age to further [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Weekly new COVID-19 cases from September 2021 to February 2022 in Salt Lake County, [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: LLM inference for different age groups in the HALE framework. The [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: LLM inference in the HALE framework across different municipalities. The [PITH_FULL_IMAGE:figures/full_fig_p012_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: LLM inference within the HALE framework across small municipalities. The [PITH_FULL_IMAGE:figures/full_fig_p013_6.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

40 extracted references · 40 canonical work pages · 5 internal anchors

  1. [1]

    Accessed: 2025-08-21

    Federal Highway Administration (FHWA).National Household Travel Survey.https : / / nhts.ornl.gov/. Accessed: 2025-08-21. 2025

    work page 2025

  2. [2]

    Are all underimmunized measles clusters equally critical?

    Sifat Afroj Moon, Achla Marathe, and Anil Vullikanti. “Are all underimmunized measles clusters equally critical?” In:Royal Society Open Science10.8 (2023)

    work page 2023

  3. [3]

    Travel for medical or dental care by race/ethnicity and rurality in the US: findings from the 2001, 2009 and 2017 National Household Travel Surveys

    Marvellous Akinlotan et al. “Travel for medical or dental care by race/ethnicity and rurality in the US: findings from the 2001, 2009 and 2017 National Household Travel Surveys”. In: Preventive Medicine Reports35 (2023), p. 102297. 14

    work page 2001

  4. [4]

    Accessed: 2024-01-

    Rachel Bachman.How Close Do You Need to Be to Your Gym?https://www.wsj.com/ articles/how-close-do-you-need-to-be-to-your-gym-1490111186. Accessed: 2024-01-

    work page 2024

  5. [5]

    Decisions & distance: The rela- tionship between child care access and child care travel

    Evelyn Blumenberg, Madeline Wander, and Zhiyuan Yao. “Decisions & distance: The rela- tionship between child care access and child care travel”. In:Journal of transport geography 114 (2024), p. 103756

    work page 2024

  6. [6]

    Accessed: 2025-08-21

    Isaac Brodsky.H3: Uber’s Hexagonal Hierarchical Spatial Index.https://www.uber.com/ blog/h3/. Accessed: 2025-08-21. 2025

    work page 2025

  7. [7]

    AgentTorch: Agent-based Modeling with Automatic Differentiation

    Ayush Chopra et al. “AgentTorch: Agent-based Modeling with Automatic Differentiation”. In:Second Agent Learning in Open-Endedness Workshop

    work page

  8. [8]

    Large language models are zero-shot recognizers for activities of daily living

    Gabriele Civitarese et al. “Large language models are zero-shot recognizers for activities of daily living”. In:ACM Transactions on Intelligent Systems and Technology16.4 (2025), pp. 1– 32

    work page 2025

  9. [9]

    The United States COVID-19 Forecast Hub dataset

    Estee Y Cramer et al. “The United States COVID-19 Forecast Hub dataset”. In:medRxiv (2021).doi:10.1101/2021.11.04.21265886.url:https://www.medrxiv.org/content/ 10.1101/2021.11.04.21265886v1

    work page doi:10.1101/2021.11.04.21265886.url:https://www.medrxiv.org/content/ 2021

  10. [10]

    Mixing patterns between age groups in social networks

    Sara Y Del Valle et al. “Mixing patterns between age groups in social networks”. In:Social Networks29.4 (2007), pp. 539–554

    work page 2007

  11. [11]

    The llama 3 herd of models

    Abhimanyu Dubey et al. “The llama 3 herd of models”. In:arXiv e-prints(2024), arXiv– 2407

    work page 2024

  12. [12]

    gov/surveys/ntps/estable/table/ntps/ntps2021_sflt07_t1s

    National Center for Education Statistics.Average public school class size: Average class size in public K–12 schools, by school level, class type, and state: 2020–21.https://nces.ed. gov/surveys/ntps/estable/table/ntps/ntps2021_sflt07_t1s. Accessed: 2025-06-01. 2025

    work page 2020

  13. [13]

    Agent-based modelling meets generative ai in social network simu- lations

    Antonino Ferraro et al. “Agent-based modelling meets generative ai in social network simu- lations”. In:International Conference on Advances in Social Networks Analysis and Mining. Springer. 2024, pp. 155–170

    work page 2024

  14. [14]

    JSONSchemaBench: A Rigorous Benchmark of Structured Outputs for Language Models

    Saibo Geng et al. “Jsonschemabench: A rigorous benchmark of structured outputs for lan- guage models”. In:arXiv preprint arXiv:2501.10868(2025)

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  15. [15]

    Generative Agent-Based Modeling: Unveiling Social System Dynamics through Coupling Mechanistic Models with Generative Artificial Intelligence

    Navid Ghaffarzadegan et al. “Generative agent-based modeling: Unveiling social system dy- namics through coupling mechanistic models with generative artificial intelligence”. In:arXiv preprint arXiv:2309.11456(2023)

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  16. [16]

    Utilizing large language models in infectious disease transmission mod- elling for public health preparedness

    Kin On Kwok et al. “Utilizing large language models in infectious disease transmission mod- elling for public health preparedness”. In:Computational and Structural Biotechnology Jour- nal23 (2024), pp. 3254–3257

    work page 2024

  17. [17]

    Efficient memory management for large language model serving with pagedattention

    Woosuk Kwon et al. “Efficient memory management for large language model serving with pagedattention”. In:Proceedings of the 29th symposium on operating systems principles. 2023, pp. 611–626

    work page 2023

  18. [18]

    Agentic AI for sustainable development: Leveraging large language model- enhanced agent-based modeling for complex policy strategies

    Jia’an Liu et al. “Agentic AI for sustainable development: Leveraging large language model- enhanced agent-based modeling for complex policy strategies”. In:Emerging Media3.3 (2025), pp. 401–413

    work page 2025

  19. [19]

    Beyond neighborhood food environments: distance traveled to food establishments in 5 US cities, 2009–2011

    Jodi L Liu, Bing Han, and Deborah A Cohen. “Beyond neighborhood food environments: distance traveled to food establishments in 5 US cities, 2009–2011”. In:Preventing chronic disease12 (2015), E126. 15

    work page 2009

  20. [20]

    The effective reproductive number of the Omicron variant of SARS-CoV-2 is several times relative to Delta

    Ying Liu and Joacim Rockl¨ ov. “The effective reproductive number of the Omicron variant of SARS-CoV-2 is several times relative to Delta”. In:Journal of travel medicine29.3 (2022), taac037

    work page 2022

  21. [21]

    Contact tracing evaluation for COVID-19 transmis- sion in the different movement levels of a rural college town in the USA

    Sifat A Moon and Caterina M Scoglio. “Contact tracing evaluation for COVID-19 transmis- sion in the different movement levels of a rural college town in the USA”. In:Scientific reports 11.1 (2021), p. 4891

    work page 2021

  22. [22]

    A spatio-temporal individual-based network framework for West Nile virus in the USA: Spreading pattern of West Nile virus

    Sifat A Moon et al. “A spatio-temporal individual-based network framework for West Nile virus in the USA: Spreading pattern of West Nile virus”. In:PLoS computational biology15.3 (2019), e1006875

    work page 2019

  23. [23]

    Sifat Afroj Moon.HALE data and simulation. Apr. 2026.doi:10.5281/zenodo.19829913. url:https://doi.org/10.5281/zenodo.19829913

    work page doi:10.5281/zenodo.19829913 2026

  24. [24]

    Long-distance dispersal and accelerating waves of disease: em- pirical relationships

    Christopher C Mundt et al. “Long-distance dispersal and accelerating waves of disease: em- pirical relationships”. In:The American Naturalist173.4 (2009), pp. 456–466

    work page 2009

  25. [25]

    A population data-driven workflow for COVID-19 modeling and learn- ing

    Jonathan Ozik et al. “A population data-driven workflow for COVID-19 modeling and learn- ing”. In:The International Journal of High Performance Computing Applications35.5 (2021), pp. 483–499

    work page 2021

  26. [26]

    Large language models can infer psychological disposi- tions of social media users

    Heinrich Peters and Sandra C Matz. “Large language models can infer psychological disposi- tions of social media users”. In:PNAS nexus3.6 (2024), pgae231

    work page 2024

  27. [27]

    Performance and biases of large language models in public opinion simulation

    Yao Qu and Jue Wang. “Performance and biases of large language models in public opinion simulation”. In:Humanities and Social Sciences Communications11.1 (2024), pp. 1–13

    work page 2024

  28. [28]

    Data Assimilation for Robust UQ Within Agent-Based Simulation on HPC Systems

    Adam Spannaus et al. “Data Assimilation for Robust UQ Within Agent-Based Simulation on HPC Systems”. In:Proceedings of the Platform for Advanced Scientific Computing Con- ference. 2025, pp. 1–11

    work page 2025

  29. [29]

    Quantifying asymptomatic infection and transmission of COVID-19 in New York City using observed cases, serology, and testing capacity

    Rahul Subramanian, Qixin He, and Mercedes Pascual. “Quantifying asymptomatic infection and transmission of COVID-19 in New York City using observed cases, serology, and testing capacity”. In:Proceedings of the National Academy of Sciences118.9 (2021), e2019716118

    work page 2021

  30. [30]

    Joe Tuccillo.Downscaling Synthetic Populations to Realistic Residential Locations. Tech. rep. Oak Ridge National Laboratory (ORNL), Oak Ridge, TN (United States), 2024

    work page 2024

  31. [31]

    UrbanPop: A spatial microsimulation framework for exploring demo- graphic influences on human dynamics

    Joseph Tuccillo et al. “UrbanPop: A spatial microsimulation framework for exploring demo- graphic influences on human dynamics”. In:Applied Geography151 (2023), p. 102844

    work page 2023

  32. [32]

    COVID-19 Omicron variant infectious period and transmission from people with asymptomatic compared with symptomatic infection: A rapid review

    UKHSA. “COVID-19 Omicron variant infectious period and transmission from people with asymptomatic compared with symptomatic infection: A rapid review”. In:GOV-14430.{UK Health Security Agency}(2023)

    work page 2023

  33. [33]

    Generative agent-based modeling with actions grounded in physical, social, or digital space using Concordia

    Alexander Sasha Vezhnevets et al. “Generative agent-based modeling with actions grounded in physical, social, or digital space using Concordia”. In:arXiv preprint arXiv:2312.03664 (2023)

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  34. [34]

    Asymptomatic SARS-CoV-2 infection by age: a global systematic review and meta-analysis

    Bing Wang et al. “Asymptomatic SARS-CoV-2 infection by age: a global systematic review and meta-analysis”. In:The Pediatric Infectious Disease Journal42.3 (2023), pp. 232–239

    work page 2023

  35. [35]

    Efficient Guided Generation for Large Language Models

    Brandon T Willard and R´ emi Louf. “Efficient Guided Generation for Large Language Mod- els”. In:arXiv preprint arXiv:2307.09702(2023)

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  36. [36]

    Epidemic Modeling with Generative Agents

    Ross Williams et al. “Epidemic modeling with generative agents”. In:arXiv preprint arXiv:2307.04986 (2023). 16

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  37. [37]

    Evaluating the ability of large Language models to predict human social decisions

    Feng Xiao and XT XiaoTian Wang. “Evaluating the ability of large Language models to predict human social decisions”. In:Scientific Reports15.1 (2025), p. 32290

    work page 2025

  38. [38]

    A baseline structure inventory with critical attribution for the US and its territories

    Hsiuhan Lexie Yang et al. “A baseline structure inventory with critical attribution for the US and its territories”. In:Scientific Data11.1 (2024), p. 502

    work page 2024

  39. [39]

    LLM-AIDSim: LLM-Enhanced Agent-Based Influence Diffusion Simulation in Social Networks

    Lan Zhang et al. “LLM-AIDSim: LLM-Enhanced Agent-Based Influence Diffusion Simulation in Social Networks”. In:Systems13.1 (2025), p. 29

    work page 2025

  40. [40]

    Exploring the role of large language models in the scientific method: from hypothesis to discovery

    Yanbo Zhang et al. “Exploring the role of large language models in the scientific method: from hypothesis to discovery”. In:npj Artificial Intelligence1.1 (2025), p. 14. 17

    work page 2025