Pith. sign in

REVIEW 3 major objections 6 minor 57 references

Fine-Tuning Large Language Models for Codebook-Guided Coding of Students' Mathematics Metaphor Responses

T0 review · 3 major / 6 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read This paper shows that LoRA fine-tuning on human-coded examples makes compact open-weight language models competitive with, and often better than, prompt-only proprietary models for codebook-guided coding of students' mathematics metaphors.

desk verdict Solid, honest applied work on fine-tuning small open-weight models for student metaphor coding; the one load-bearing ambiguity is whether the train/test split was at the student level. read the letter →

arxiv 2608.10276 v1 pith:ICGFYYDW submitted 2026-08-10 cs.HC cs.CY

Liang Zhang , Stephen Hwang , Yue Ma , Jinfa Cai This is my paper · ORCID
classification cs.HCcs.CY
keywords largelanguagemodelsLoRAfine-tuningcodebook-guidedcodingstudentmathematicsmetaphorsvalence-intensitythematicopen-weighteducationalmeasurement
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper tries to establish that compact open-weight language models, fine-tuned with LoRA on a small corpus of human-coded student responses, can perform codebook-guided coding of students' mathematics metaphors as well as or better than much larger proprietary models used with a zero-shot prompt. It tests two coding tasks on 2,265 Grade 6-8 food- and animal-metaphor responses: valence-intensity coding, which assigns a single 1-5 affective code, and thematic coding, which assigns one or more semantic theme codes. Across a held-out test set, fine-tuning raised the open-weight models' accuracy and run-to-run reliability substantially, and the fine-tuned 7B model outperformed both proprietary prompt-only baselines on nearly every reported metric. The practical stake is that education researchers could run private, locally hosted, task-adapted AI coding of open-ended student text without sending sensitive responses to third-party APIs.

What carries the argument

The load-bearing mechanism is LoRA-based supervised fine-tuning: training small low-rank adapter matrices (rank 16, scaling factor 32) on the frozen base weights of DeepSeek-R1 1.5B and Mistral 7B, with each training instance pairing a student's metaphor object and explanation with the human-assigned valence and thematic codes. The codebook itself is the second mechanism, constraining outputs to a 1-5 valence scale with flags and a 22-root-code thematic scheme across four families, embedded in the prompt for every model. What the LoRA adapters learn is the mapping from response language to these expert codes, which is why fine-tuned models produce codes that the base versions essentially never emitted for rare thematic families.

What would settle it

Train the same LoRA pipeline on a version of the training labels in which a random 20% of codes are shuffled; if held-out accuracy does not fall far below the reported levels, the model is relying on artifacts rather than the human-code mapping. Alternatively, have an independent expert team re-code a random 100-response sample and compare their agreement with the original labels to the model's agreement with those labels; if independent experts agree with the original labels no more often than the model does, the gold standard is too unstable to support the claims.

Watch

Extended reading notes

Core claim

The central discovery is that parameter-efficient supervised fine-tuning transfers human expert coding judgments into compact language models well enough to make them the best coders in the comparison. Using LoRA adapters with rank 16 on DeepSeek-R1 1.5B and Mistral 7B, trained on 1,568 human-coded responses plus a validation set, the paper reports valence-intensity accuracy rising from 0.369 to 0.787 on food responses and from 0.325 to 0.685 on animal responses for DeepSeek, and from 0.207 to 0.778 and 0.267 to 0.766 for Mistral; thematic subset accuracy at the exact code-set level rose from 0.063 to 0.502 and 0.052 to 0.555 for DeepSeek. Fine-tuned Mistral beat both proprietary prompt-only models on every reported metric except one tie in thematic Macro-F1 on food responses. The authors interpret this as task-specific alignment with the human-coded codebook rather than general language superiority.

Load-bearing premise

The paper's result depends on the human-assigned codes being correct on nearly all of the 2,265 responses; if the human labels are systematically wrong in places, the fine-tuned models learn those errors and the reported accuracies overstate how well they code student metaphors.

Editorial extensions

If this is right

  • A 7B-parameter model fine-tuned on roughly 1,500 examples can serve as a locally deployable coder for educational open-response data, matching or exceeding proprietary API models on the tested tasks.
  • Run-to-run reliability improves with fine-tuning, with DeepSeek's thematic exact-code agreement across runs rising from 57.6% to 96.0% and Mistral's from 98.1% to 99.4% (the base Mistral agreement is noted by the authors to reflect near-constant output).
  • Rare thematic families remain fragile: after fine-tuning, DeepSeek still did not recover family D codes, and both models remained below the proprietary models on thematic Macro-F1 in places.
  • The gains are task-specific: the authors attribute them to alignment with the human-coded codebook, not to general language ability, and restrict their claims to the food- and animal-metaphor corpus.
  • The method offers a privacy-conscious path for AI-assisted measurement of students' mathematical attitudes, beliefs, and experiences from open-ended responses.

Reading between the lines

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

  • Because temperature was fixed at zero, the reported stability may understate variability under stochastic decoding; a natural extension is to sample multiple outputs per response at higher temperature and measure code-level consistency.
  • The same LoRA fine-tuning recipe should transfer to other codebook-guided coding tasks with similar label structures, such as sentiment or stance coding of short open responses, but that transfer is untested outside food- and animal-metaphor prompts.
  • The comparison is between fine-tuned open-weight models and prompt-only proprietary models, so it does not establish that open models beat closed models when both are fine-tuned; a stronger test would fine-tune the API models on the same human-coded labels.
  • Since human labels are both the training signal and the evaluation ground truth, a practical safeguard is to train on an adjudicated subset and evaluate against independently re-coded responses, separating model error from label noise.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. The manuscript reports a controlled comparison of codebook-guided LLM coding for student mathematics metaphor responses. Using 2,265 human-coded Grade 6–8 responses to food- and animal-based prompts, the authors evaluate zero-shot prompt-only proprietary models (GPT-4o mini, GPT-5 mini) against two open-weight models (DeepSeek-R1 1.5B, Mistral 7B) in both base and LoRA-fine-tuned conditions. Two coding tasks are considered: valence–intensity coding (single-code, ordinal) and thematic coding (multi-code). Evaluation uses accuracy, macro-F1, and QWK for valence, and subset accuracy, macro-F1, and micro-F1 for themes, averaged over three runs on a held-out test set of 347 responses. The central claim is that LoRA fine-tuning substantially improves the coding performance and run-to-run reliability of the open-weight models, making them competitive with or better than the proprietary prompt-only baselines.

Significance. The study addresses a timely practical question: whether compact, locally deployable open-weight LLMs can be adapted to specialized educational coding tasks without sending student data to third-party APIs. The evaluation has genuine strengths: a held-out test set, multiple complementary metrics, three repeated runs, and transparent table notes—for example, the near-constant A22 output of base Mistral 7B is explicitly flagged in Table 3. The code-level analysis in Table 4, including the uneven gains across rare families B/C/D, is informative and helps bound the claims. If the methodological concerns below are resolved, the paper would be a useful contribution to AI-assisted educational measurement and to the broader evidence on fine-tuned small models versus proprietary prompt-only models.

major comments (3)
  1. [§3 Evaluations (dataset split)] The train/validation/test split is described only in response counts: training (783 food, 785 animal), validation (176 food, 174 animal), and test (180 food, 167 animal). Because the corpus consists of paired food and animal responses from the same students, a random response-level split would place one of a test student's paired responses in the training set. After three epochs of LoRA fine-tuning, the model could exploit student-specific lexical or affective cues from the paired training response, inflating the reported held-out accuracy, the run-to-run reliability gains, and the comparison against prompt-only proprietary models. The authors must state whether the split was at the student level; if it was not, they should re-run the evaluation with a student-level split or otherwise demonstrate that no test student appears in the training set.
  2. [§4 Results] The main text reports point estimates averaged over three runs and states that "Detailed significance test results for the coding performance gains are provided in Appendix C," but Appendix C is not present in the manuscript. As a result, the reader cannot assess whether headline gains such as DeepSeek-R1 1.5B valence accuracy improving from 0.369 to 0.787 on the food subset are distinguishable from run-to-run variability. The authors should include the significance tests or, at minimum, add confidence intervals or error bars to Figure 1 and Table 3 so that the precision of the estimates is visible in the main text.
  3. [§3 Dataset (gold-standard coding)] Inter-rater reliability is reported only for subsets: Krippendorff's α = 0.84–0.91 for valence–intensity coding and 92% exact-match agreement on 100 responses for thematic coding. The full corpus of 2,265 responses was coded by experts with discussion, but the manuscript does not describe how disagreements were resolved or whether any additional reliability check was performed on the full corpus. If systematic coding biases exist in the full corpus, the fine-tuned models will learn them, and the reported accuracies against the same gold standard could overstate agreement with an independent expert panel. Please clarify the full-corpus coding procedure and, if feasible, report a reliability check on a random sample of the test set.
minor comments (6)
  1. [Figure 1] Since all results are averages over three runs, please show per-run variation (e.g., error bars or jittered points) so the reader can see the stability behind each point estimate.
  2. [Table 4 caption] The caption states n = 347 for the held-out test set, but the valence–intensity rows exclude two responses labeled X; please state n = 345 for those rows to avoid confusion.
  3. [§3 Evaluations] The text says the temperature of the LLM models was set to 0, then notes that GPT-5 mini does not support explicit temperature control; please state clearly that the temperature setting applies to GPT-4o mini, DeepSeek-R1 1.5B, and Mistral 7B only.
  4. [§2 Related Work] There is a subject-verb agreement error in "supervised fine-tuning improve the ability"; it should be "improves." Please proofread for similar issues.
  5. [§3 Dataset] The thematic-coding agreement of 92% exact match on 100 responses would be easier to interpret if a chance-corrected index such as Cohen's kappa or Krippendorff's alpha were also reported.
  6. [References] Several reference entries contain "and 1 others" and the two Alizadeh et al. entries appear to describe the same work in different venues; please complete and harmonize the author lists.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: fine-tuned performance is an out-of-sample empirical comparison, not a derivation from the labels.

full rationale

After walking the claimed derivation chain, I find no circular step. The study's central claim is empirical: LoRA fine-tuning on human-coded examples improves held-out coding accuracy and run-to-run reliability of open-weight models relative to their base versions, and makes them competitive with prompt-only proprietary baselines. The human-assigned codes were produced independently of the models, the training/validation/test split is described in response counts, and the reported metrics (accuracy, macro-F1, QWK, subset accuracy, micro-F1) are computed on a held-out test set that the adapters never saw during training. The fine-tuning loss is computed only on training cases, so the evaluation is a genuine out-of-sample measurement. The codebook is the authors' own research artifact, but that is a generalizability/measurement concern, not a circularity, because the target result (model coding performance) is not an input to the codebook construction. One methodological worry, that the response-level split may place paired food and animal responses from the same student in different splits and thus inflate generalization estimates, is a leakage/external-validity risk rather than a self-definitional or fitted-input circularity; it does not make the reported numbers equivalent to the training labels by construction. Since no step reduces to its own input, the circularity score is 0.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central claim rests on the quality of the human coding, the codebook's coverage, and the specific fine-tuning choices. No invented entities are introduced. The reported gains are conditional on a single set of LoRA hyperparameters and a single decoding temperature.

free parameters (2)
  • LoRA fine-tuning hyperparameter set = rank=16, scale=32, dropout=0.05, max_seq_len=2048, epochs=3, batch_size=8, grad_accum=2, lr=1e-4, wd=0.01…
    All chosen by hand with no reported sensitivity analysis. The reported gains are conditional on these values; different choices, especially epochs and learning rate, could change the results for rare codes.
  • Decoding temperature = 0 (GPT-5 mini not controllable)
    A single decoding setting. The paper itself lists this as a limitation, and run-to-run reliability for GPT-5 mini is lower, possibly due to uncontrolled temperature.
assumptions (4)
  • domain assumption Human-assigned codes are a valid gold standard for the affective and thematic content of student metaphor responses.
    Used as training labels and evaluation reference in Section 3. If the human codes are systematically wrong, the fine-tuned models learn these errors and the reported performance is biased.
  • domain assumption The codebook developed from 250 responses generalizes to the full 2,265-response corpus.
    The thematic codebook was built from a small subset and refined with double coding; the paper assumes these categories capture the full range of student metaphors. Section 3.
  • domain assumption LLM outputs can be reliably parsed into the required JSON schema and mapped to valid codes under the stated valid-output checking rules.
    The evaluation depends on parsing and validation; the exact rules are not specified, so any leniency could inflate accuracy. Section 3 Evaluations.
  • ad hoc to paper The fixed LoRA training schedule (3 epochs) yields an adapter with stable performance; no early stopping or hyperparameter search is reported.
    The choice of 3 epochs is specific to this paper and is not derived from first principles. The result is contingent on this schedule. Section 3 Fine-tuning Setup.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Fine-Tuning Large Language Models for Codebook-Guided Coding of Students' Mathematics Metaphor Responses." pith.science (2026). https://pith.science/paper/ICGFYYDW

@misc{pith2026260810276,
  author       = {Pith},
  title        = {Pith review of: Fine-Tuning Large Language Models for Codebook-Guided Coding of Students' Mathematics Metaphor Responses},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ICGFYYDW}},
  note         = {Machine review of arXiv:2608.10276}
}
read the original abstract

Student-generated metaphors about mathematics can reveal students' attitudes, beliefs, identities, and experiences, but human expert coding of these thematically and semantically complex open-ended responses is time-intensive and difficult to scale. This study examines whether LoRA-based supervised fine-tuning of large language models (LLMs) can improve their performance on codebook-guided coding tasks for student mathematics metaphors. We used a human-coded corpus of 2,265 Grade 6-8 responses to food- and animal-based metaphor prompts and instructed the LLMs to perform two coding tasks: valence-intensity coding to capture the direction and strength of students' affective orientations toward mathematics, and thematic coding to capture students' framings of mathematics as expressed through their metaphors. We compared two proprietary models, GPT-4o mini and GPT-5 mini, under prompt-only conditions with two open-weight models, DeepSeek-R1 1.5B and Mistral 7B, evaluated before and after fine-tuning. Results show that fine-tuning substantially improved the performance and run-to-run reliability of the open-weight models across both tasks relative to their base versions. The fine-tuned compact open-weight models became competitive with, and often outperformed, the proprietary prompt-only models. These findings suggest that compact open-weight LLMs can support scalable, locally controllable, and privacy-conscious AI-assisted measurement of students' metaphor responses in mathematics education.

Figures

Figures reproduced from arXiv: 2608.10276 by the authors.

Figure 1
Figure 1. Coding performance across LLMs for food- and animal-based metaphor responses. [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

57 extracted references · 42 canonical work pages

  1. [1]

    1980 , publisher=

    Metaphors We Live By , author=. 1980 , publisher=

    work page 1980

  2. [2]

    2000 , publisher=

    Where mathematics comes from: How the embodied mind brings mathematics into being , author=. 2000 , publisher=

    work page 2000

  3. [3]

    Beliefs about teaching: Persistent or malleable? A longitudinal study of prospective student teachers' beliefs , journal =

    Erika Löfström and Katrin Poom-Valickis , keywords =. Beliefs about teaching: Persistent or malleable? A longitudinal study of prospective student teachers' beliefs , journal =. 2013 , issn =. doi:https://doi.org/10.1016/j.tate.2013.06.004 , url =

    work page doi:10.1016/j.tate.2013.06.004 2013

  4. [4]

    Teaching and Teacher Education , volume =

    Using metaphor in mathematics teacher preparation , author=. Teaching and Teacher Education , volume =

    work page

  5. [5]

    The Journal of Mathematical Behavior , volume=

    Metaphor as a possible pathway to more formal understanding of the definition of sequence convergence , author=. The Journal of Mathematical Behavior , volume=

    work page

  6. [6]

    Educational Studies in Mathematics , volume=

    Metaphors for learning and doing mathematics in advanced mathematics lectures , author=. Educational Studies in Mathematics , volume=

    work page

  7. [7]

    Educational Studies in Mathematics , volume=

    On metaphors in thinking about preparing mathematics for teaching , author=. Educational Studies in Mathematics , volume=

    work page

  8. [8]

    European Journal of Science and Mathematics Education , volume=

    Math is like a lion hunting a sleeping gazelle: preservice elementary teachers’ metaphors of mathematics , author=. European Journal of Science and Mathematics Education , volume=

    work page

Show all 57 references
  1. [9]

    Journal of Humanistic Mathematics , volume=

    Metaphors and Mathematical Identity: Math is Like a Tornado in Kansas , author=. Journal of Humanistic Mathematics , volume=

  2. [10]

    Journal of Mathematics Teacher Education , volume=

    Investigating teachers' images of mathematics , author=. Journal of Mathematics Teacher Education , volume=

  3. [11]

    Joseph Merlino , title =

    Jinfa Cai and F. Joseph Merlino , title =. 2011 , chapter =

    2011

  4. [12]

    Australian Mathematics Teachers , volume=

    Sandra Frid , title =. Australian Mathematics Teachers , volume=

  5. [13]

    Yee , title =

    Sean P. Yee , title =. School Science and Mathematics , year =. 2017 , publisher=

    2017

  6. [14]

    School Science and Mathematics , volume=

    Students’ and teachers’ conceptual metaphors for mathematical problem solving , author=. School Science and Mathematics , volume=. 2017 , publisher=

    2017

  7. [15]

    School science and mathematics , volume=

    Using metaphors to unpack student beliefs about mathematics , author=. School science and mathematics , volume=. 2008 , publisher=

    2008

  8. [16]

    Learning and Individual Differences , volume=

    What metaphors of learning can (and cannot) tell us about students' learning , author=. Learning and Individual Differences , volume=. 2020 , publisher=

    2020

  9. [17]

    ACM Transactions on Interactive Intelligent Systems (TiiS) , volume=

    Using machine learning to support qualitative coding in social science: Shifting the focus to ambiguity , author=. ACM Transactions on Interactive Intelligent Systems (TiiS) , volume=. 2018 , publisher=

    2018

  10. [18]

    Companion proceedings of the 28th international conference on intelligent user interfaces , pages=

    Supporting qualitative analysis with large language models: Combining codebook with GPT-3 for deductive coding , author=. Companion proceedings of the 28th international conference on intelligent user interfaces , pages=

  11. [19]

    Proceedings of the Association for Information Science and Technology , volume=

    Assessing the Reliability of Large Language Models for Deductive Qualitative Coding: A Comparative Intervention Study with ChatGPT , author=. Proceedings of the Association for Information Science and Technology , volume=. 2025 , publisher=

    2025

  12. [20]

    Social Science Computer Review , volume=

    Large language models outperform expert coders and supervised classifiers at annotating political social media messages , author=. Social Science Computer Review , volume=. 2025 , publisher=

    2025

  13. [21]

    International Journal of Qualitative Methods , volume=

    An examination of the use of large language models to aid analysis of textual data , author=. International Journal of Qualitative Methods , volume=. 2024 , publisher=

    2024

  14. [22]

    , author=

    Qualitative Coding with GPT-4: Where It Works Better. , author=. Journal of Learning Analytics , volume=. 2025 , publisher=

    2025

  15. [23]

    arXiv preprint arXiv:2403.14541 , year=

    Edt: Improving large language models' generation by entropy-based dynamic temperature sampling , author=. arXiv preprint arXiv:2403.14541 , year=

    arXiv

  16. [24]

    Proceedings of the 55th ACM Technical Symposium on Computer Science Education V

    Understanding the role of temperature in diverse question generation by GPT-4 , author=. Proceedings of the 55th ACM Technical Symposium on Computer Science Education V. 2 , pages=

  17. [25]

    OpenAI API Reference: Responses , howpublished =

  18. [26]

    Using GPT-5.2: GPT-5.2 Parameter Compatibility , howpublished =

  19. [27]

    Information processing & management , volume=

    A systematic analysis of performance measures for classification tasks , author=. Information processing & management , volume=. 2009 , publisher=

    2009

  20. [28]

    , author=

    Weighted kappa: Nominal scale agreement provision for scaled disagreement or partial credit. , author=. Psychological bulletin , volume=. 1968 , publisher=

    1968

  21. [29]

    Applied Intelligence , volume=

    Confidence interval for micro-averaged F 1 and macro-averaged F 1 scores , author=. Applied Intelligence , volume=. 2022 , publisher=

    2022

  22. [30]

    , author=

    Lora: Low-rank adaptation of large language models. , author=. Iclr , volume=

  23. [31]

    Advances in neural information processing systems , volume=

    Training language models to follow instructions with human feedback , author=. Advances in neural information processing systems , volume=

  24. [32]

    Chat Templates , year =

  25. [33]

    DeepSeek-R1-Distill-Qwen-1.5B Model Card , year =

  26. [34]

    Mistral-7B-Instruct-v0.1 Model Card , year =

  27. [35]

    Psychometrika , volume=

    Note on the sampling error of the difference between correlated proportions or percentages , author=. Psychometrika , volume=. 1947 , publisher=

    1947

  28. [36]

    Neural computation , volume=

    Approximate statistical tests for comparing supervised classification learning algorithms , author=. Neural computation , volume=. 1998 , publisher=

    1998

  29. [37]

    Journal of Machine Learning Research , volume=

    Scaling instruction-finetuned language models , author=. Journal of Machine Learning Research , volume=

  30. [38]

    arXiv preprint arXiv:2308.10092 , year=

    Open, closed, or small language models for text classification? , author=. arXiv preprint arXiv:2308.10092 , year=

    arXiv

  31. [39]

    Journal of Computational Social Science , volume=

    Open-source LLMs for text annotation: a practical guide for model setting and fine-tuning , author=. Journal of Computational Social Science , volume=. 2025 , publisher=

    2025

  32. [40]

    arXiv preprint arXiv:2406.08660 , year=

    Fine-tuned'small'LLMs (still) significantly outperform zero-shot generative AI models in text classification , author=. arXiv preprint arXiv:2406.08660 , year=

    arXiv

  33. [41]

    arXiv preprint arXiv:2501.06658 , year=

    Comparing few-shot prompting of GPT-4 LLMs with BERT classifiers for open-response assessment in tutor equity training , author=. arXiv preprint arXiv:2501.06658 , year=

    arXiv

  34. [42]

    EPJ Data Science , volume=

    Scaling hermeneutics: a guide to qualitative coding with LLMs for reflexive content analysis , author=. EPJ Data Science , volume=. 2025 , publisher=

    2025

  35. [43]

    2010 , school=

    Scalable multi-label classification , author=. 2010 , school=

    2010

  36. [44]

    IEEE transactions on knowledge and data engineering , volume=

    A review on multi-label learning algorithms , author=. IEEE transactions on knowledge and data engineering , volume=. 2013 , publisher=

    2013

  37. [45]

    Information retrieval , volume=

    An evaluation of statistical approaches to text categorization , author=. Information retrieval , volume=. 1999 , publisher=

    1999

  38. [46]

    Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing , pages=

    Large language models do multi-label classification differently , author=. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing , pages=

    2025

  39. [47]

    GPT-4o mini: Advancing cost-efficient intelligence , year =

  40. [48]

    arXiv preprint arXiv:2307.02179 , year=

    Open-source llms for text annotation: A practical guide for model setting and fine-tuning.” arxiv , author=. arXiv preprint arXiv:2307.02179 , year=

    arXiv

  41. [49]

    Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP) , pages=

    An empirical study on large-scale multi-label text classification including few and zero-shot labels , author=. Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP) , pages=

    2020

  42. [50]

    British journal of educational technology , volume=

    Ethical and privacy principles for learning analytics , author=. British journal of educational technology , volume=. 2014 , publisher=

    2014

  43. [51]

    British Journal of Educational Technology , volume=

    Practical and ethical challenges of large language models in education: A systematic scoping review , author=. British Journal of Educational Technology , volume=. 2024 , publisher=

    2024

  44. [52]

    Nature , volume=

    Why open-source generative AI models are an ethical way forward for science , author=. Nature , volume=. 2023 , publisher=

    2023

  45. [53]

    Motivation and disposition: Pathways to learning mathematics , pages=

    Metaphor: A powerful means for assessing students’ mathematical disposition , author=. Motivation and disposition: Pathways to learning mathematics , pages=. 2011 , publisher=

    2011

  46. [54]

    2012 , publisher=

    Mathematical dispositions and student learning: A metaphorical analysis , author=. 2012 , publisher=

    2012

  47. [55]

    Nature Machine Intelligence , volume=

    The dangers of using proprietary LLMs for research , author=. Nature Machine Intelligence , volume=. 2024 , publisher=

    2024

  48. [56]

    Investigations in Mathematics Learning , volume=

    Accessing and assessing components of elementary and middle school students’ mathematical disposition through metaphors , author=. Investigations in Mathematics Learning , volume=. 2025 , publisher=

    2025

  49. [57]

    Computing Krippendorff's alpha-reliability , author=

Pith tools

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