pith. machine review for the scientific record. sign in

arxiv: 2605.01333 · v2 · submitted 2026-05-02 · 💻 cs.CL

Recognition: no theorem link

OralMLLM-Bench: Evaluating Cognitive Capabilities of Multimodal Large Language Models in Dental Practice

Rongyang Wang , Shuang Zhou , Jiashuo Wang , Wenya Xie , Xiaoxia Che
Authors on Pith no claims yet

Pith reviewed 2026-05-11 01:00 UTC · model grok-4.3

classification 💻 cs.CL
keywords multimodal large language modelsdental radiographycognitive capabilitiesbenchmark evaluationclinical AIimage analysisdentistry
0
0 comments X

The pith

A new benchmark reveals multimodal AI models fall short of dentists in cognitive tasks for analyzing dental radiographs.

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

This paper presents OralMLLM-Bench to assess how multimodal large language models handle the cognitive demands of dental radiographic analysis. The benchmark covers three X-ray modalities and organizes evaluation into perception, comprehension, prediction, and decision-making. Through 27 tasks and thousands of clinician comparisons, it maps out where models succeed and where they fail relative to human experts. This matters for developing AI that can safely support dental diagnostics by matching the layered thinking clinicians use in practice.

Core claim

The benchmark evaluates frontier MLLMs on dental image analysis tasks and finds they underperform clinicians, while pinpointing specific strengths in basic perception, weaknesses in higher-level prediction and decision-making, along with common failure modes and suggestions for better alignment with clinical needs.

What carries the argument

OralMLLM-Bench, consisting of 27 tasks across four cognitive categories and three radiograph types, evaluated against 3,820 clinician assessments.

Load-bearing premise

The selected 27 tasks and clinician assessments represent the full multi-level cognitive processes in real-world dental radiographic analysis without major selection or annotation biases.

What would settle it

Demonstrating that MLLMs achieve clinician-level performance on an independent set of clinical dental radiographs using the same cognitive categories would challenge the reported performance gap.

read the original abstract

Multimodal large language models (MLLMs) have emerged as a promising paradigm for dental image analysis. However, their ability to capture the multi-level cognitive processes required for radiographic analysis remains unclear. Here, we present a comprehensive benchmark to evaluate the cognitive capabilities of MLLMs in dental radiographic analysis. It spans three critical imaging modalities, i.e., periapical, panoramic, and lateral cephalometric radiographs, and defines four cognitive categories: perception, comprehension, prediction, and decision-making. The benchmark comprises 27 clinically grounded tasks derived from public datasets, with manually curated annotations and 3,820 clinician assessments for evaluation. Six frontier MLLMs, including GPT-5.2 and GLM-4.6, are evaluated. We demonstrate the performance gap between MLLMs and clinicians in dental practice, delineate model strengths and limitations, characterize failure patterns, and provide recommendations for improvement. This data resource will facilitate the development of next-generation artificial intelligence systems aligned with clinical cognition, safety requirements, and workflow complexity in dental practice.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper introduces OralMLLM-Bench, a benchmark to evaluate the cognitive capabilities of multimodal large language models (MLLMs) for dental radiographic analysis. It spans three imaging modalities (periapical, panoramic, lateral cephalometric) and four cognitive categories (perception, comprehension, prediction, decision-making), comprising 27 tasks derived from public datasets with manually curated annotations and 3,820 clinician assessments. Six frontier MLLMs are evaluated against clinician baselines to demonstrate performance gaps, delineate strengths and limitations, characterize failure patterns, and offer improvement recommendations.

Significance. If the benchmark construction and clinician evaluations prove robust, this work could meaningfully advance AI alignment with clinical cognition in dentistry by supplying a structured, multi-level evaluation resource. It would help developers target specific cognitive failure modes in MLLMs and support safer deployment in radiographic workflows.

major comments (2)
  1. [§3] §3 (Benchmark Construction): The description of the 27 tasks and their derivation from public datasets provides no details on validation procedures, clinical expert review process, or checks for selection bias, which is load-bearing for the central claim that the tasks accurately capture multi-level cognitive processes required for real-world radiographic analysis.
  2. [Evaluation section] Evaluation section: No inter-rater agreement statistics, prompt engineering details, or statistical tests for performance gap significance are reported for the 3,820 clinician assessments, making it impossible to determine whether the observed MLLM-clinician differences are robust or could be artifacts of annotation variability.
minor comments (2)
  1. [Abstract] Abstract: Model names GPT-5.2 and GLM-4.6 are non-standard; clarify exact versions or checkpoints used to allow reproducibility.
  2. [Task descriptions] Task descriptions: Provide at least one concrete example per cognitive category in the main text to illustrate how perception differs from comprehension in the radiographic context.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We address each major comment below and will revise the manuscript to incorporate additional details where appropriate.

read point-by-point responses

  1. Referee: [§3] §3 (Benchmark Construction): The description of the 27 tasks and their derivation from public datasets provides no details on validation procedures, clinical expert review process, or checks for selection bias, which is load-bearing for the central claim that the tasks accurately capture multi-level cognitive processes required for real-world radiographic analysis.

    Authors: We agree that Section 3 would benefit from expanded methodological transparency. In the revised manuscript we will add a dedicated subsection detailing: (1) the exact selection criteria and sampling procedure used to derive the 27 tasks from the cited public datasets, (2) the multi-stage clinical validation workflow (including the number of board-certified dentists who independently reviewed each annotation for clinical accuracy and cognitive-category alignment), and (3) explicit checks for selection bias (e.g., stratification by modality, anatomical region, and difficulty level). These additions will directly support the claim that the tasks reflect real-world radiographic cognition. revision: yes

  2. Referee: [Evaluation section] Evaluation section: No inter-rater agreement statistics, prompt engineering details, or statistical tests for performance gap significance are reported for the 3,820 clinician assessments, making it impossible to determine whether the observed MLLM-clinician differences are robust or could be artifacts of annotation variability.

    Authors: We acknowledge the omission of these quantitative safeguards. In the revised Evaluation section we will report: (1) inter-rater agreement metrics (Fleiss’ kappa and percentage agreement) computed across the multiple clinician assessments per task, (2) the full prompt templates and engineering choices used for both MLLM inference and clinician elicitation, and (3) statistical significance tests (paired Wilcoxon signed-rank tests with Bonferroni correction and effect sizes) for all reported MLLM–clinician performance gaps. These additions will allow readers to assess the robustness of the observed differences. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper introduces a benchmark (OralMLLM-Bench) consisting of 27 manually curated tasks across three imaging modalities and four cognitive categories, evaluated via 3,820 independent clinician assessments. No equations, derivations, parameter fitting, or predictions derived from the benchmark itself are present. All load-bearing elements (task definitions, annotations, and clinician baselines) are externally sourced or human-curated rather than self-referential. No self-citation chains, uniqueness theorems, or ansatzes reduce the central claims to the paper's own inputs by construction. The evaluation is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

No mathematical derivations, fitted parameters, or new theoretical entities are described in the abstract; the contribution is an empirical benchmark built on public datasets and human annotations.

pith-pipeline@v0.9.0 · 5493 in / 1120 out tokens · 38798 ms · 2026-05-11T01:00:10.536351+00:00 · methodology

Review history (2 revisions) →

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

73 extracted references · 31 canonical work pages · 5 internal anchors

  1. [1]

    & Mazevet, M

    Vergnes, J.-N. & Mazevet, M. Oral diseases: a global public health challenge. Lancet 395, 186 (2020)

    2020

  2. [2]

    Tattar, R., da Costa, B. D. C. & Neves, V. C. M. The interrelationship between periodontal disease and systemic health. Br. Dent. J. 239, 103–108 (2025)

    2025

  3. [3]

    Ali, M., Irfan, M., Ali, T., Wei, C. R. & Akilimali, A. Artificial intelligence in dental radiology: a narrative review. Ann. Med. Surg. (Lond.) 87, 2212–2217 (2025)

    2025

  4. [4]

    & Cox, S

    Hegde, S., Gao, J., Vasa, R. & Cox, S. Factors affecting interpretation of dental radiographs. Dentomaxillofac. Radiol. 52, 20220279 (2023)

    2023

  5. [5]

    B., Rodrigues, J

    Diniz, M. B., Rodrigues, J. A., Neuhaus, K. W., Cordeiro, R. C. L. & Lussi, A. Influence of examiner’s clinical experience on the reproducibility and accuracy of radiographic examination in detecting occlusal caries. Clin. Oral Investig. 14, 515–523 (2010)

    2010

  6. [6]

    & Krois, J

    Schwendicke, F., Golla, T., Dreher, M. & Krois, J. Convolutional neural networks for dental image diagnostics: A scoping review. J. Dent. 91, 103226 (2019)

    2019

  7. [7]

    Zhou, Z. et al. Deep learning in dental image analysis: A systematic review of datasets, methodologies, and emerging challenges. arXiv [cs.CV] (2025) doi:10.48550/arXiv.2510.20634

    work page doi:10.48550/arxiv.2510.20634 2025

  8. [8]

    & Gruhn, V

    Kü hnisch, J., Meyer, O., Hesenius, M., Hickel, R. & Gruhn, V. Caries detection on intraoral images using artificial intelligence. J. Dent. Res. 101, 158–165 (2022)

    2022

  9. [9]

    Lee, S. W. et al. Evaluation by dental professionals of an artificial intelligence -based application to measure alveolar bone loss. BMC Oral Health 25, 329 (2025)

    2025

  10. [10]

    Alharbi, S. S. & Alhasson, H. F. Exploring the applications of artificial intelligence in dental image detection: A systematic review. Diagnostics (Basel) 14, 2442 (2024)

    2024

  11. [11]

    Hegde, S. et al. Machine learning algorithms enhance the accuracy of radiographic diagnosis of dental caries: a comparative study. Dentomaxillofac. Radiol. 54, 632–641 (2025)

    2025

  12. [12]

    Singh, N. K. & Raza, K. Progress in deep learning -based dental and maxillofacial image analysis: A systematic review. Expert Syst. Appl. 199, 116968 (2022)

    2022

  13. [13]

    Li, X. et al. Deep learning for classifying the stages of periodontitis on dental images: a systematic review and meta-analysis. BMC Oral Health 23, 1017 (2023)

    2023

  14. [14]

    & Armoundas, A

    Christof, M. & Armoundas, A. A. Implications of integrating large language models into clinical decision making. Commun. Med. (Lond.) 5, 490 (2025)

    2025

  15. [15]

    & Vogt, J

    Sokol, K., Fackler, J. & Vogt, J. E. Artificial intelligence should genuinely support clinical reasoning and decision making to bridge the translational gap. NPJ Digit. Med. 8, 345 (2025)

    2025

  16. [16]

    Yu, K. et al. Multimodal artificial intelligence agents in healthcare: A scoping review. Authorea Inc. (2025) doi:10.22541/au.176055853.39564234/v1

    work page doi:10.22541/au.176055853.39564234/v1 2025

  17. [17]

    Zhou, S. et al. Large language models for disease diagnosis: a scoping review. NPJ Artif. Intell. 1, 9 (2025)

    2025

  18. [18]

    Tu, T. et al. Towards conversational diagnostic artificial intelligence. Nature (2025) doi:10.1038/s41586-025-08866-7

    work page doi:10.1038/s41586-025-08866-7 2025

  19. [19]

    Zhou, S. et al. HeartAgent: An autonomous agent system for explainable differential diagnosis in cardiology. arXiv [cs.CL] (2026) doi:10.48550/arXiv.2603.10764

    work page doi:10.48550/arxiv.2603.10764 2026

  20. [20]

    Liu, X. et al. Developing and evaluating multimodal large language model for orthopantomography analysis to support clinical dentistry. Cell Rep. Med. 7, 102652 (2026)

    2026

  21. [21]

    Hao, J. et al. OralGPT-Omni: A Versatile Dental Multimodal Large Language Model. arXiv [cs.CV] (2025) doi:10.48550/arXiv.2511.22055

    work page doi:10.48550/arxiv.2511.22055 2025

  22. [22]

    Mallick, A. N. D., Ahmed, S., Mohammed, N., Dudley, J. & Farook, T. H. Assessing the feasibility of a multimodal annotation and segmentation system to detect carious lesions from intraoral photographs. Int. J. Dent. 2026, (2026)

    2026

  23. [23]

    Gao, X. et al. Multimodal language model for jaw osteonecrosis diagnosis and treatment. J. Dent. Res. 104, 1324–1332 (2025)

    2025

  24. [24]

    Cai, Z. et al. DentalGPT: Incentivizing Multimodal Complex Reasoning in Dentistry. arXiv [cs.CV] (2025) doi:10.48550/arXiv.2512.11558

    work page doi:10.48550/arxiv.2512.11558 2025

  25. [25]

    Wafaie, K. et al. Diagnostic accuracy of generative large language artificial intelligence models for the assessment of dental crowding. BMC Oral Health 25, 1558 (2025)

    2025

  26. [26]

    Hao, J. et al. Characteristics, licensing, and ethical considerations of openly accessible oral- maxillofacial imaging datasets: a systematic review. NPJ Digit. Med. 8, 412 (2025)

    2025

  27. [27]

    Park, W., Huh, J. -K. & Lee, J. -H. Automated deep learning for classification of dental implant radiographs using a large multi-center dataset. Sci. Rep. 13, 4862 (2023)

    2023

  28. [28]

    A., Yilmaz, R

    Karcioglu, A. A., Yilmaz, R. M., Yaganoglu, M., Almohammad, M. & Laloglu, A. Advancing forensic dentistry: a comprehensive review of machine learning and deep learning applications in dental image analysis. Neural Comput. Appl. 37, 24997–25032 (2025)

    2025

  29. [29]

    Hao, J. et al. Towards better dental AI: A multimodal benchmark and instruction dataset for panoramic X-ray analysis. arXiv [cs.CV] (2025) doi:10.48550/arXiv.2509.09254

    work page doi:10.48550/arxiv.2509.09254 2025

  30. [30]

    Yu, Z. et al. OPGAgent: An agent for auditable dental panoramic X -ray interpretation. arXiv [cs.CV] (2026) doi:10.48550/arXiv.2603.00462

    work page doi:10.48550/arxiv.2603.00462 2026

  31. [31]

    Liu, Z. et al. DentVLM: A multimodal vision -language model for comprehensive dental diagnosis and enhanced clinical practice. Research Square (2025) doi: 10.21203/rs.3.rs- 7403627/v1

    work page doi:10.21203/rs.3.rs- 2025

  32. [32]

    & Liu, Z

    Zhu, H., Xu, Y., Li, Y., Meng, Z. & Liu, Z. DentalBench: Benchmarking and advancing LLMs capability for bilingual dentistry understanding. arXiv [cs.CL] (2025) doi:10.48550/arXiv.2508.20416

    work page doi:10.48550/arxiv.2508.20416 2025

  33. [33]

    Lv, H. et al. A benchmark multimodal oro-dental dataset for large vision-language models. arXiv [cs.CV] (2025) doi:10.48550/arXiv.2511.04948

    work page doi:10.48550/arxiv.2511.04948 2025

  34. [34]

    A., Vuong, T

    Nguyen, V. A., Vuong, T. Q. T. & Nguyen, V. H. Benchmarking large -language-model vision capabilities in oral and maxillofacial anatomy: A cross -sectional study. PLoS One 20, e0335775 (2025)

    2025

  35. [35]

    Xiong, Y. -T. et al. Evaluating the performance of large language models (LLMs) in answering and analysing the Chinese dental licensing examination. Eur. J. Dent. Educ. 29, 332– 340 (2025)

    2025

  36. [36]

    Xiong, H. et al. IOSVLM: A 3D vision-language model for unified dental diagnosis from intraoral scans. arXiv [cs.CV] (2026) doi:10.48550/arXiv.2603.16781

    work page doi:10.48550/arxiv.2603.16781 2026

  37. [37]

    Zhou, S. et al. Automating expert-level medical reasoning evaluation of large language models. NPJ Digit. Med. 9, 34 (2025)

    2025

  38. [38]

    Jin, Q. et al. Hidden flaws behind expert -level accuracy of multimodal GPT -4 vision in medicine. NPJ Digit. Med. 7, 190 (2024)

    2024

  39. [39]

    Zhou, S. et al. Mitigating ethical issues for large language models in oncology: A systematic review. JCO Clin. Cancer Inform. 9, e2500076 (2025)

    2025

  40. [40]

    Singh, A. et al. OpenAI GPT -5 System Card. arXiv [cs.CL] (2025) doi:10.48550/arXiv.2601.03267

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2601.03267 2025

  41. [41]

    Glm, T. et al. ChatGLM: A family of large language models from GLM -130B to GLM-4 All Tools. arXiv [cs.CL] (2024) doi:10.48550/arXiv.2406.12793

    work page internal anchor Pith review doi:10.48550/arxiv.2406.12793 2024

  42. [42]

    Liu, W. et al. Spatial reasoning in Multimodal large language models: A survey of tasks, benchmarks and methods. arXiv [cs.AI] (2025) doi:10.48550/arXiv.2511.15722

    work page doi:10.48550/arxiv.2511.15722 2025

  43. [43]

    & Yuksel, D

    Griot, M., Hemptinne, C., Vanderdonckt, J. & Yuksel, D. Large Language Models lack essential metacognition for reliable medical reasoning. Nat. Commun. 16, 642 (2025)

    2025

  44. [44]

    Huang, C. et al. Mimicking or reasoning: Rethinking multi -modal in-context learning in vision-language models. arXiv [cs.CV] (2025) doi:10.48550/arXiv.2506.07936

    work page doi:10.48550/arxiv.2506.07936 2025

  45. [45]

    Hamamci, I. E. et al. DENTEX: Dental Enumeration and tooth pathosis detection benchmark for Panoramic X-ray. arXiv [cs.CV] (2025) doi:10.48550/arXiv.2305.19112

    work page doi:10.48550/arxiv.2305.19112 2025

  46. [46]

    Minimax-01: Scaling foundation models with lightning attention.arXiv preprint arXiv:2501.08313, 2025

    MiniMax et al. MiniMax-01: Scaling foundation models with lightning attention. arXiv [cs.CL] (2025) doi:10.48550/arXiv.2501.08313

    work page doi:10.48550/arxiv.2501.08313 2025

  47. [47]

    Kimi K2: Open Agentic Intelligence

    Kimi Team et al. Kimi K2: Open Agentic Intelligence. arXiv [cs.LG] (2026) doi:10.48550/arXiv.2507.20534

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2507.20534 2026

  48. [48]

    Bai, J. et al. Qwen-VL: A versatile vision-language model for understanding, localization, text reading, and beyond. arXiv [cs.CV] (2023) doi:10.48550/arXiv.2308.12966

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2308.12966 2023

  49. [49]

    & Lee, Y

    Liu, H., Li, C., Wu, Q. & Lee, Y. J. Visual instruction tuning. Adv. Neural Inf. Process. Syst. 36, 34892–34916 (2023)

    2023

  50. [50]

    Rasekh, A., Ranjbar, S. K. & Gottschalk, S. Multi-rationale explainable object recognition via contrastive conditional inference. arXiv [cs.CV] (2025) doi:10.48550/arXiv.2508.14280

    work page doi:10.48550/arxiv.2508.14280 2025

  51. [51]

    Chen, H. et al. REV: Information -theoretic evaluation of free -text rationales. in Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) (Association for Computational Linguistics, Stroudsburg, PA, USA, 2023). doi:10.18653/v1/2023.acl-long.112

    work page doi:10.18653/v1/2023.acl-long.112 2023

  52. [52]

    Zhu, J., Miao, S., Ying, R. & Li, P. Towards understanding sensitive and decisive patterns in explainable AI: A case study of model interpretation in geometric deep learning. arXiv [cs.LG] (2024) doi:10.48550/arXiv.2407.00849

    work page doi:10.48550/arxiv.2407.00849 2024

  53. [53]

    Klaib, A. F. et al. Automated assessment and detection of third molar and inferior alveolar nerve relations using UNet and transfer learning models. Sci. Rep. 15, 34529 (2025)

    2025

  54. [54]

    C., Toprak, T., Selver, M

    Ulusoy, A. C., Toprak, T., Selver, M. A., Güneri, P. & İlhan, B. Panoramic radiographic features for machine learning based detection of mandibular third molar root and inferior alveolar canal contact. Sci. Rep. 15, 4178 (2025)

    2025

  55. [55]

    Kruse, M. et al. Large language models with temporal reasoning for longitudinal clinical summarization and prediction. Find. ACL EMNLP 2025, 20715–20735 (2025)

    2025

  56. [56]

    Zhang, Y. et al. Period-LLM: Extending the periodic capability of multimodal large language model. arXiv [cs.CV] (2025) doi:10.48550/arXiv.2505.24476

    work page doi:10.48550/arxiv.2505.24476 2025

  57. [57]

    F., Lyu, C

    Imam, M. F., Lyu, C. & Aji, A. F. Can Multimodal LLMs do Visual Temporal Understanding and Reasoning? The answer is No! arXiv [cs.CV] (2025) doi:10.48550/arXiv.2501.10674

    work page doi:10.48550/arxiv.2501.10674 2025

  58. [58]

    Zhou, S. et al. Uncertainty-aware large language models for explainable disease diagnosis. NPJ Digit. Med. 8, 690 (2025)

    2025

  59. [59]

    Kim, J. et al. Limitations of large Language Models in clinical problem -solving arising from inflexible reasoning. arXiv [cs.CL] (2025) doi:10.48550/arXiv.2502.04381

    work page doi:10.48550/arxiv.2502.04381 2025

  60. [60]

    Lee, D. et al. Breaking the visual shortcuts in Multimodal Knowledge -Based Visual Question Answering. arXiv [cs.CV] (2026) doi:10.48550/arXiv.2511.22843

    work page doi:10.48550/arxiv.2511.22843 2026

  61. [61]

    Croxford, E. et al. Evaluating clinical AI summaries with large language models as judges. NPJ Digit. Med. 8, 640 (2025)

    2025

  62. [62]

    Periapical X-rays

    Sajad, M. Periapical X-rays. Kaggle (2021)

    2021

  63. [63]

    Dental Periapical X-rays

    Aglan, N. Dental Periapical X-rays. Kaggle (2024)

    2024

  64. [64]

    Silva, B. et al. OdontoAI: A human-in-the-loop labeled data set and an online platform to boost research on dental panoramic radiographs. arXiv [cs.CV] (2022) doi:10.48550/arXiv.2203.15856

    work page doi:10.48550/arxiv.2203.15856 2022

  65. [65]

    Khalid, M. A. et al. A benchmark dataset for automatic cephalometric landmark detection and CVM stage classification. Sci. Data 12, 1336 (2025)

    2025

  66. [66]

    Zhang, H. et al. Deep learning techniques for automatic lateral X -ray Cephalometric Landmark Detection: Is the problem solved? arXiv [cs.CV] (2024) doi:10.48550/arXiv.2409.15834

    work page doi:10.48550/arxiv.2409.15834 2024

  67. [67]

    Kazimierczak, N. et al. Detection accuracy of an AI platform for dental treatment features on panoramic radiographs - tooth- and patient-level analyses. Sci. Rep. 16, 2436 (2025)

    2025

  68. [68]

    Zhang, J.-W. et al. Diagnostic accuracy of artificial intelligence-assisted caries detection: a clinical evaluation. BMC Oral Health 24, 1095 (2024)

    2024

  69. [69]

    Van Calster, B. et al. Evaluation of performance measures in predictive artificial intelligence models to support medical decisions: overview and guidance. Lancet Digit. Health 7, 100916 (2025)

    2025

  70. [70]

    Wang, S. et al. A novel evaluation benchmark for medical LLMs illuminating safety and effectiveness in clinical domains. NPJ Digit. Med. 9, 91 (2025)

    2025

  71. [71]

    Tang, X., Ambrose, G. A. & Cheng, Y. Designing reliable LLM-assisted rubric scoring for constructed responses: Evidence from physics exams. arXiv [cs.AI] (2026) doi:10.48550/arXiv.2604.12227

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2604.12227 2026

  72. [72]

    arXiv preprint arXiv:2601.08843 (2025)

    Deng, H., Farber, C., Lee, J. & Tang, D. Rubric -conditioned LLM grading: Alignment, uncertainty, and robustness. arXiv [cs.CL] (2025) doi:10.48550/arXiv.2601.08843

    work page doi:10.48550/arxiv.2601.08843 2025

  73. [73]

    & Setlur, V

    Palani, S. & Setlur, V. Lexara: A user-centered toolkit for evaluating large language models for conversational visual analytics. in Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems 1–25 (ACM, New York, NY, USA, 2026)

    2026