---
title: "Training LLM-Based Tutors to Improve Student Learning"
description: "The authors propose training a small open-source model, Llama 3.1 8B… 0.76 AUC on the MathDial. 53 references, cited by 17 papers here."
authors: ["Alexander Scarlatos", "Naiming Liu", "Jaewook Lee", "Richard Baraniuk", "Andrew Lan"]
year: 2025
venue: "Lecture Notes in Computer Science"
source: "https://arxiv.org/abs/2503.06424"
doi: "10.1007/978-3-031-98414-3_18"
arxiv: "2503.06424"
licence: "CC BY 4.0 (arXiv)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/scarlatos-2025-training-llm-based-tutors-improve/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/scarlatos-2025-training-llm-based-tutors-improve/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2025](https://md.research.elamur.ai/free/en/papers/2025/page/3/) / Training LLM-Based Tutors to Improve Student Learning Outcomes in Dialogues

# Training LLM-Based Tutors to Improve Student Learning Outcomes in Dialogues

**Authors:** [Alexander Scarlatos](https://md.research.elamur.ai/free/en/authors/alexander-scarlatos/), [Naiming Liu](https://md.research.elamur.ai/free/en/authors/naiming-liu/), Jaewook Lee, [Richard Baraniuk](https://md.research.elamur.ai/free/en/authors/richard-baraniuk/) and [Andrew Lan](https://md.research.elamur.ai/free/en/authors/andrew-lan/) · **Year:** 2025 · **Venue:** [Lecture Notes in Computer Science](https://md.research.elamur.ai/free/en/venues/lecture-notes-in-computer-science/) ([Lecture notes in computer science 2025](https://md.research.elamur.ai/free/en/venues/lecture-notes-in-computer-science/2025/)) · **Publisher:** [Springer Science+Business Media](https://md.research.elamur.ai/free/en/publishers/springer-science-business-media/) · **Licence:** [CC BY 4.0 (arXiv)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://doi.org/10.1007/978-3-031-98414-3_18) · [PDF](https://arxiv.org/pdf/2503.06424) · [DOI 10.1007/978-3-031-98414-3\_18](https://doi.org/10.1007/978-3-031-98414-3_18) · [arXiv 2503.06424](https://arxiv.org/abs/2503.06424) · [OpenAlex](https://openalex.org/W4413069692)

**Citations:** 21 citations (OpenAlex, 2026-10-14) · [17](https://md.research.elamur.ai/free/en/papers/scarlatos-2025-training-llm-based-tutors-improve/cited-by/) in our library, of the 149,439 papers whose bibliographies we read

## Summary
LLM tutors are usually built by prompting or fine-tuning models to imitate good teaching practice. They are rarely optimized for whether students actually learn. The authors propose training a small open-source model, Llama 3.1 8B, to produce tutor turns that make a correct student reply more likely while still respecting pedagogical norms.

Using math dialogues from MathDial, they collect several candidate next tutor turns per dialogue step. The candidates come from the original human tutors and from various LLMs. Each candidate receives two scores:
- **Predicted correctness:** the LLMKT student model estimates the chance of a correct student reply.
- **Pedagogical quality:** GPT-4o grades the turn against a hand-designed rubric.

High- and low-scoring candidates are paired, and the model is trained with Direct Preference Optimization (DPO). The resulting tutor substantially raises predicted student correctness and reaches pedagogical quality comparable to GPT-4o. A small human study also favours it.

The work matters because it shows that compact, controllable, privacy-friendly models can be steered toward learning outcomes rather than style alone. The main caveat is that all "students" are simulated.

## Key points
- **Student model:** LLMKT, a knowledge-tracing model for dialogues, serves as the simulator. It reaches 0.76 AUC on the MathDial test set.
- **Rubric:** six binary items grouped under Accuracy, Progress and Guidance. The Guidance items cover error identification, strategic hinting, withholding the answer and encouragement.
- **Human evaluation:** two annotators rated 50 tutor turns. DPO beat human tutors and GPT-4o on both measures, and all differences were significant (p \< 0.05, paired t-test).
  - Mean correctness rank (lower is better): DPO 1.75, GPT-4o 2.13, human tutors 2.12.
  - Rubric score (1–10): DPO 8.55, GPT-4o 8.07, human tutors 7.36.
- **Agreement:** judging likely correctness proved very subjective (Kendall's τ of 0.06 between annotators). Rubric agreement was higher, with Pearson's ρ of 0.27 between GPT-4o and the annotators.
- **Limitations and release:** there are no experiments with real students, and the work covers the math domain only. The code is released at github.com/umass-ml4ed/tutorbot-dpo.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

## Institutions

- [University of Massachusetts Amherst](https://md.research.elamur.ai/free/en/universities/university-of-massachusetts-amherst/) — University, Amherst Center, United States · 122 works · 69,574 citations
- [Rice University](https://md.research.elamur.ai/free/en/universities/rice-university/) — University, Houston, United States · 15 works · 2,036 citations

📚 Shelves (5): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 19,503 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 18,805 works · [🗄️ Information Systems](https://md.research.elamur.ai/free/en/shelves/computer-science/information-systems/) · 1,888 works · [🩺 Medicine](https://md.research.elamur.ai/free/en/shelves/medicine/) · 288 works · [🩺 Health Informatics](https://md.research.elamur.ai/free/en/shelves/medicine/health-informatics/) · 227 works

🧭 Topics (1): [🗨️ Dialogue systems](https://md.research.elamur.ai/free/en/topics/dialogue-systems/) · 664 works

## Cite

Cite as: Alexander Scarlatos, Naiming Liu, Jaewook Lee, Richard Baraniuk and Andrew Lan. 2025. Training LLM-Based Tutors to Improve Student Learning Outcomes in Dialogues. Lecture Notes in Computer Science.

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{scarlatos2025training,
  title = {{Training LLM-Based Tutors to Improve Student Learning Outcomes in Dialogues}},
  author = {Alexander Scarlatos, Naiming Liu, Jaewook Lee, Richard Baraniuk and Andrew Lan},
  booktitle = {Lecture Notes in Computer Science},
  year = {2025},
  doi = {10.1007/978-3-031-98414-3\_18},
  eprint = {2503.06424},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.1007/978-3-031-98414-3_18}
}
```

- [← previous Revisiting the Test-Time Scaling of o1-like Models: Do they Truly Possess Test-Time Scaling Capabilities?](https://md.research.elamur.ai/free/en/papers/zeng-2025-revisiting-test-time-scaling-o1/)
- [next → UFO: A UI-Focused Agent for Windows OS Interaction](https://md.research.elamur.ai/free/en/papers/zhang-2025-ufo-ui-focused-agent-windows/)
- [↑ back to papers of 2025](https://md.research.elamur.ai/free/en/papers/2025/page/3/)

## References

53 references, 17 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. John R Anderson, Albert T Corbett, Kenneth R Koedinger, and Ray Pelletier. Cognitive tutors: Lessons learned. The journal of the learning sciences, 4(2):167–207, 1995. [find](https://scholar.google.com/scholar?q=John+R+Anderson%2C+Albert+T+Corbett%2C+Kenneth+R+Koedinger%2C+and+Ray+Pelletier%2E+Cognitive+tutors%3A+Lessons+learned%2E+The+journal+of+the+learning+sciences%2C+4%282%29%3A167%E2%80%93207%2C+1995%2E)
2. Nischal Ashok Kumar and Andrew Lan. [Improving socratic question generation using data augmentation and preference optimization](https://md.research.elamur.ai/free/en/papers/kumar-2024-improving-socratic-question-generation-data/). In Proceedings of the 19th Workshop on Innovative Use of NLP for Building Educational Applications (BEA 2024), pages 108–118, Mexico City, Mexico, June 2024. Association fo…
3. Dejun Cai, Yuan Zhang, and Bintao Dai. Learning path recommendation based on knowledge tracing model and reinforcement learning. In 2019 IEEE 5th international conference on computer and communications (ICCC), pages 1881–1885. IEEE, 2019. [find](https://scholar.google.com/scholar?q=Dejun+Cai%2C+Yuan+Zhang%2C+and+Bintao+Dai%2E+Learning+path+recommendation+based+on+knowledge+tracing+model+and+reinforcement+learning%2E+In+2019+IEEE+5th+international+conference+on+computer+and+communication)
4. Carnegie Learning. Livehint overview. Online: [https://support.carnegielearning.com/help-center/](https://support.carnegielearning.com/help-center/) math/livehint/article/livehint-overview/, 2024. [↗](https://support.carnegielearning.com/help-center/)
5. Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, Christopher Hesse, and John Schulman. [Training verifiers to solve math word problems](https://md.research.elamur.ai/free/en/papers/cobbe-2021-training-verifiers-solve-math-word/). arXiv preprint arXiv:2110.14168, 2021.
6. Common Core State Standards Initiative. Mathematics standards. Online: [https://www.thecorestandards.org/Math/](https://www.thecorestandards.org/Math/), 2024. [↗](https://www.thecorestandards.org/Math/)
7. Albert Corbett and John Anderson. Knowledge tracing: Modeling the acquisition of procedural knowledge. User Model. User-adapted Interact., 4(4):253–278, Dec. 1994. [find](https://scholar.google.com/scholar?q=Albert+Corbett+and+John+Anderson%2E+Knowledge+tracing%3A+Modeling+the+acquisition+of+procedural+knowledge%2E+User+Model%2E+User%2Dadapted+Interact%2E%2C+4%284%29%3A253%E2%80%93278%2C+Dec%2E+1994%2E)
8. Peng Cui and Mrinmaya Sachan. Adaptive and personalized exercise generation for online language learning. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 10184–10198, 2023. [find](https://scholar.google.com/scholar?q=Peng+Cui+and+Mrinmaya+Sachan%2E+Adaptive+and+personalized+exercise+generation+for+online+language+learning%2E+In+Proceedings+of+the+61st+Annual+Meeting+of+the+Association+for+Computational+Linguistics+%28Vo)
9. Dorottya Demszky, Jing Liu, Zid Mancenido, Julie Cohen, Heather Hill, Dan Jurafsky, and Tatsunori B Hashimoto. [Measuring conversational uptake: A case study on student-teacher interactions](https://md.research.elamur.ai/free/en/papers/demszky-2021-measuring-conversational-uptake-case-study/). In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th Intern…
10. Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, et al. [The llama 3 herd of models](https://md.research.elamur.ai/free/en/papers/grattafiori-2024-llama-3-herd-models/). arXiv preprint arXiv:2407.21783, 2024.
11. Hugging Face. Hugging face - the ai community building the future. [https://huggingface.co/](https://huggingface.co/), 2025. Accessed: 2025-02-18. [↗](https://huggingface.co/)
12. Nigel Fernandez, Alexander Scarlatos, Wanyong Feng, Simon Woodhead, and Andrew Lan. DiVERT: Distractor generation with variational errors represented as text for math multiple-choice questions. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 9063–9081… [find](https://scholar.google.com/scholar?q=Nigel+Fernandez%2C+Alexander+Scarlatos%2C+Wanyong+Feng%2C+Simon+Woodhead%2C+and+Andrew+Lan%2E+DiVERT%3A+Distractor+generation+with+variational+errors+represented+as+text+for+math+multiple%2Dchoice+questions%2E+In+Pro)
13. Aritra Ghosh, Neil Heffernan, and Andrew S Lan. [Context-aware attentive knowledge tracing](https://md.research.elamur.ai/free/en/papers/ghosh-2020-context-aware-attentive-knowledge-tracing/). In Proc. ACM SIGKDD, pages 2330–2339, 2020.
14. Arthur C Graesser, Kurt VanLehn, Carolyn P Rosé, Pamela W Jordan, and Derek Harter. Intelligent tutoring systems with conversational dialogue. AI magazine, 22(4):39–39, 2001. [find](https://scholar.google.com/scholar?q=Arthur+C+Graesser%2C+Kurt+VanLehn%2C+Carolyn+P+Ros%C3%A9%2C+Pamela+W+Jordan%2C+and+Derek+Harter%2E+Intelligent+tutoring+systems+with+conversational+dialogue%2E+AI+magazine%2C+22%284%29%3A39%E2%80%9339%2C+2001%2E)
15. Jieun Han, Haneul Yoo, Junho Myung, Minsun Kim, Hyunseung Lim, Yoonsu Kim, Tak Yeon Lee, Hwajung Hong, Juho Kim, So-Yeon Ahn, et al. Llm-as-a-tutor in efl writing education: Focusing on evaluation of student-llm interaction. In Proceedings of the 1st Workshop on Customizable NLP: Progress and Challe… [find](https://scholar.google.com/scholar?q=Jieun+Han%2C+Haneul+Yoo%2C+Junho+Myung%2C+Minsun+Kim%2C+Hyunseung+Lim%2C+Yoonsu+Kim%2C+Tak+Yeon+Lee%2C+Hwajung+Hong%2C+Juho+Kim%2C+So%2DYeon+Ahn%2C+et+al%2E+Llm%2Das%2Da%2Dtutor+in+efl+writing+education%3A+Focusing+on+evaluation+of+)
16. Joy He-Yueya, Noah D. Goodman, and Emma Brunskill. [Evaluating and optimizing educational content with large language model judgments](https://md.research.elamur.ai/free/en/papers/he-yueya-2024-evaluating-optimizing-educational-content-large/). In Proceedings of the 17th International Conference on Educational Data Mining, page 68–82, 2024.
17. Joy He-Yueya and Adish Singla. Quizzing policy using reinforcement learning for inferring the student knowledge state. International Educational Data Mining Society, 2021. [find](https://scholar.google.com/scholar?q=Joy+He%2DYueya+and+Adish+Singla%2E+Quizzing+policy+using+reinforcement+learning+for+inferring+the+student+knowledge+state%2E+International+Educational+Data+Mining+Society%2C+2021%2E)
18. Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. [Lora: Low-rank adaptation of large language models](https://md.research.elamur.ai/free/en/papers/hu-2021-lora-low-rank-adaptation-large/). arXiv preprint arXiv:2106.09685, 2021.
19. Qinjin Jia, Mitchell Young, Yunkai Xiao, Jialin Cui, Chengyuan Liu, Parvez Rashid, and Edward Gehringer. Insta-reviewer: A data-driven approach for generating instant feedback on students’ project reports. International Educational Data Mining Society, 2022. [find](https://scholar.google.com/scholar?q=Qinjin+Jia%2C+Mitchell+Young%2C+Yunkai+Xiao%2C+Jialin+Cui%2C+Chengyuan+Liu%2C+Parvez+Rashid%2C+and+Edward+Gehringer%2E+Insta%2Dreviewer%3A+A+data%2Ddriven+approach+for+generating+instant+feedback+on+students%E2%80%99+project+rep)
20. Sanjit Kakarla, Danielle Thomas, Jionghao Lin, Shivang Gupta, and Ken Koedinger. Using large language models to assess tutors’ performance in reacting to students making math errors. In AI for Education: Bridging Innovation and Responsibility at the 38th AAAI Annual Conference on AI, 2024. [find](https://scholar.google.com/scholar?q=Sanjit+Kakarla%2C+Danielle+Thomas%2C+Jionghao+Lin%2C+Shivang+Gupta%2C+and+Ken+Koedinger%2E+Using+large+language+models+to+assess+tutors%E2%80%99+performance+in+reacting+to+students+making+math+errors%2E+In+AI+for+Educati)
21. Maurice George Kendall. Rank correlation methods. 1948. [find](https://scholar.google.com/scholar?q=Maurice+George+Kendall%2E+Rank+correlation+methods%2E+1948%2E)
22. Khan Academy. Supercharge your teaching experience with khanmigo. Online: [https://www.khanmigo.ai/](https://www.khanmigo.ai/), 2023. [↗](https://www.khanmigo.ai/)
23. Kyuyoung Kim, Ah Jeong Seo, Hao Liu, Jinwoo Shin, and Kimin Lee. Margin matching preference optimization: Enhanced model alignment with granular feedback. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 13554–13570, Miami, Florida, USA, November 2024. Association for… [find](https://scholar.google.com/scholar?q=Kyuyoung+Kim%2C+Ah+Jeong+Seo%2C+Hao+Liu%2C+Jinwoo+Shin%2C+and+Kimin+Lee%2E+Margin+matching+preference+optimization%3A+Enhanced+model+alignment+with+granular+feedback%2E+In+Findings+of+the+Association+for+Computatio)
24. Lars Krupp, Steffen Steinert, Maximilian Kiefer-Emmanouilidis, Karina E Avila, Paul Lukowicz, Jochen Kuhn, Stefan Küchemann, and Jakob Karolus. Challenges and opportunities of moderating usage of large language models in education. In Proceedings of the International Conference on Mobile and Ubiquit… [find](https://scholar.google.com/scholar?q=Lars+Krupp%2C+Steffen+Steinert%2C+Maximilian+Kiefer%2DEmmanouilidis%2C+Karina+E+Avila%2C+Paul+Lukowicz%2C+Jochen+Kuhn%2C+Stefan+K%C3%BCchemann%2C+and+Jakob+Karolus%2E+Challenges+and+opportunities+of+moderating+usage+of+larg)
25. Joseph Lee Rodgers and W Alan Nicewander. Thirteen ways to look at the correlation coefficient. The American Statistician, 42(1):59–66, 1988. [find](https://scholar.google.com/scholar?q=Joseph+Lee+Rodgers+and+W+Alan+Nicewander%2E+Thirteen+ways+to+look+at+the+correlation+coefficient%2E+The+American+Statistician%2C+42%281%29%3A59%E2%80%9366%2C+1988%2E)
26. Anna Lieb and Toshali Goel. Student interaction with newtbot: An llm-as-tutor chatbot for secondary physics education. In Extended Abstracts of the CHI Conference on Human Factors in Computing Systems, pages 1–8, 2024. [find](https://scholar.google.com/scholar?q=Anna+Lieb+and+Toshali+Goel%2E+Student+interaction+with+newtbot%3A+An+llm%2Das%2Dtutor+chatbot+for+secondary+physics+education%2E+In+Extended+Abstracts+of+the+CHI+Conference+on+Human+Factors+in+Computing+Systems)
27. Naiming Liu, Shashank Sonkar, Zichao Wang, Simon Woodhead, and Richard G Baraniuk. [Novice learner and expert tutor: Evaluating math reasoning abilities of large language models with misconceptions](https://md.research.elamur.ai/free/en/papers/liu-2023-novice-learner-expert-tutor-evaluating/). arXiv preprint arXiv:2310.02439, 2023.
28. Naiming Liu, Zichao Wang, Richard Baraniuk, and Andrew Lan. Open-ended knowledge tracing for computer science education. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, 2022. [find](https://scholar.google.com/scholar?q=Naiming+Liu%2C+Zichao+Wang%2C+Richard+Baraniuk%2C+and+Andrew+Lan%2E+Open%2Dended+knowledge+tracing+for+computer+science+education%2E+In+Proceedings+of+the+2022+Conference+on+Empirical+Methods+in+Natural+Language+)
29. Jakub Macina, Nico Daheim, Sankalan Chowdhury, Tanmay Sinha, Manu Kapur, Iryna Gurevych, and Mrinmaya Sachan. [MathDial: A dialogue tutoring dataset with rich pedagogical properties grounded in math reasoning problems](https://md.research.elamur.ai/free/en/papers/macina-2023-mathdial-dialogue-tutoring-dataset-rich/). In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 56…
30. Andre Nickow, Philip Oreopoulos, and Vincent Quan. The impressive effects of tutoring on prek-12 learning: A systematic review and meta-analysis of the experimental evidence. Working Paper 27476, National Bureau of Economic Research, July 2020. [find](https://scholar.google.com/scholar?q=Andre+Nickow%2C+Philip+Oreopoulos%2C+and+Vincent+Quan%2E+The+impressive+effects+of+tutoring+on+prek%2D12+learning%3A+A+systematic+review+and+meta%2Danalysis+of+the+experimental+evidence%2E+Working+Paper+27476%2C+Nati)
31. Benjamin D Nye, Dillon Mee, and Mark G Core. Generative large language models for dialog-based tutoring: An early consideration of opportunities and concerns. In LLM@ AIED, pages 78–88, 2023. [find](https://scholar.google.com/scholar?q=Benjamin+D+Nye%2C+Dillon+Mee%2C+and+Mark+G+Core%2E+Generative+large+language+models+for+dialog%2Dbased+tutoring%3A+An+early+consideration+of+opportunities+and+concerns%2E+In+LLM%40+AIED%2C+pages+78%E2%80%9388%2C+2023%2E)
32. OpenAI. Hello gpt-4o, May 2024. Accessed: 2025-02-19. [find](https://scholar.google.com/scholar?q=OpenAI%2E+Hello+gpt%2D4o%2C+May+2024%2E+Accessed%3A+2025%2D02%2D19%2E)
33. Sankalan Pal Chowdhury, Vilém Zouhar, and Mrinmaya Sachan. [Autotutor meets large language models: A language model tutor with rich pedagogy and guardrails](https://md.research.elamur.ai/free/en/papers/chowdhury-2024-autotutor-meets-large-language-models/). In Proceedings of the Eleventh ACM Conference on Learning@ Scale, pages 5–15, 2024.
34. Minju Park, Sojung Kim, Seunghyun Lee, Soonwoo Kwon, and Kyuseok Kim. Empowering personalized learning through a conversation-based tutoring system with student modeling. In Extended Abstracts of the CHI Conference on Human Factors in Computing Systems, pages 1–10, 2024. [find](https://scholar.google.com/scholar?q=Minju+Park%2C+Sojung+Kim%2C+Seunghyun+Lee%2C+Soonwoo+Kwon%2C+and+Kyuseok+Kim%2E+Empowering+personalized+learning+through+a+conversation%2Dbased+tutoring+system+with+student+modeling%2E+In+Extended+Abstracts+of+the+)
35. Chris Piech, Jonathan Bassen, Jonathan Huang, Surya Ganguli, Mehran Sahami, Leonidas J Guibas, and Jascha Sohl-Dickstein. [Deep knowledge tracing](https://md.research.elamur.ai/free/en/papers/piech-2015-deep-knowledge-tracing/). In Proc. NeurIPS, pages 505–513, 2015.
36. Rafael Rafailov, Archit Sharma, Eric Mitchell, Christopher D Manning, Stefano Ermon, and Chelsea Finn. [Direct preference optimization: Your language model is secretly a reward model](https://md.research.elamur.ai/free/en/papers/rafailov-2023-direct-preference-optimization-language-model/). Advances in Neural Information Processing Systems, 36, 2024.
37. Anna N Rafferty, Emma Brunskill, Thomas L Griffiths, and Patrick Shafto. Faster teaching via pomdp planning. Cognitive science, 40(6):1290–1332, 2016. [find](https://scholar.google.com/scholar?q=Anna+N+Rafferty%2C+Emma+Brunskill%2C+Thomas+L+Griffiths%2C+and+Patrick+Shafto%2E+Faster+teaching+via+pomdp+planning%2E+Cognitive+science%2C+40%286%29%3A1290%E2%80%931332%2C+2016%2E)
38. Alexander Scarlatos, Ryan S. Baker, and Andrew Lan. Exploring knowledge tracing in tutor-student dialogues using llms. In Proceedings of the 15th Learning Analytics and Knowledge Conference, LAK 2025, Dublin, Ireland, March 3-7, 2025. ACM, 2025. [find](https://scholar.google.com/scholar?q=Alexander+Scarlatos%2C+Ryan+S%2E+Baker%2C+and+Andrew+Lan%2E+Exploring+knowledge+tracing+in+tutor%2Dstudent+dialogues+using+llms%2E+In+Proceedings+of+the+15th+Learning+Analytics+and+Knowledge+Conference%2C+LAK+2025%2C)
39. Alexander Scarlatos, Digory Smith, Simon Woodhead, and Andrew Lan. Improving the validity of automatically generated feedback via reinforcement learning. In International Conference on Artificial Intelligence in Education, pages 280–294. Springer, 2024. — [Improving the Validity of Automatically Generated Feedback via Reinforcement Learning](https://md.research.elamur.ai/free/en/papers/scarlatos-2024-improving-validity-automatically-generated-feedback/)
40. Robin Schmucker, Meng Xia, Amos Azaria, and Tom Mitchell. Ruffle&riley: Insights from designing and evaluating a large language model-based conversational tutoring system. In International Conference on Artificial Intelligence in Education, pages 75–90. Springer, 2024. [find](https://scholar.google.com/scholar?q=Robin+Schmucker%2C+Meng+Xia%2C+Amos+Azaria%2C+and+Tom+Mitchell%2E+Ruffle%26riley%3A+Insights+from+designing+and+evaluating+a+large+language+model%2Dbased+conversational+tutoring+system%2E+In+International+Conference+)
41. Shashank Sonkar, Naiming Liu, Debshila Mallick, and Richard Baraniuk. [CLASS: A design framework for building intelligent tutoring systems based on learning science principles](https://md.research.elamur.ai/free/en/papers/sonkar-2023-class-design-framework-building-intelligent/). In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 1941–1961, Singapore, December 2023. Associa…
42. Shashank Sonkar, Kangqi Ni, Sapana Chaudhary, and Richard Baraniuk. [Pedagogical alignment of large language models](https://md.research.elamur.ai/free/en/papers/sonkar-2024-pedagogical-alignment-large-language-models/). In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 13641–13650, Miami, Florida, USA, November 2024. Association for Computational Linguistics.
43. John Stamper, Ruiwei Xiao, and Xinying Hou. Enhancing llm-based feedback: Insights from intelligent tutoring systems and the learning sciences. In International Conference on Artificial Intelligence in Education, pages 32–43. Springer, 2024. [find](https://scholar.google.com/scholar?q=John+Stamper%2C+Ruiwei+Xiao%2C+and+Xinying+Hou%2E+Enhancing+llm%2Dbased+feedback%3A+Insights+from+intelligent+tutoring+systems+and+the+learning+sciences%2E+In+International+Conference+on+Artificial+Intelligence+i)
44. Jacob Steiss, Tamara Tate, Steve Graham, Jazmin Cruz, Michael Hebert, Jiali Wang, Youngsun Moon, Waverly Tseng, Mark Warschauer, and Carol Booth Olson. Comparing the quality of human and chatgpt feedback of students’ writing. Learning and Instruction, 91:101894, 2024. [find](https://scholar.google.com/scholar?q=Jacob+Steiss%2C+Tamara+Tate%2C+Steve+Graham%2C+Jazmin+Cruz%2C+Michael+Hebert%2C+Jiali+Wang%2C+Youngsun+Moon%2C+Waverly+Tseng%2C+Mark+Warschauer%2C+and+Carol+Booth+Olson%2E+Comparing+the+quality+of+human+and+chatgpt+feedb)
45. LearnLM Team. [Learnlm: Improving gemini for learning](https://md.research.elamur.ai/free/en/papers/team-2024-learnlm-improving-gemini-learning/), 2024.
46. Kurt VanLehn, Pamela W Jordan, Carolyn P Rosé, Dumisizwe Bhembe, Michael Böttner, Andy Gaydos, Maxim Makatchev, Umarani Pappuswamy, Michael Ringenberg, Antonio Roque, et al. The architecture of why2-atlas: A coach for qualitative physics essay writing. In Intelligent Tutoring Systems: 6th Internatio… [find](https://scholar.google.com/scholar?q=Kurt+VanLehn%2C+Pamela+W+Jordan%2C+Carolyn+P+Ros%C3%A9%2C+Dumisizwe+Bhembe%2C+Michael+B%C3%B6ttner%2C+Andy+Gaydos%2C+Maxim+Makatchev%2C+Umarani+Pappuswamy%2C+Michael+Ringenberg%2C+Antonio+Roque%2C+et+al%2E+The+architecture+of+why2%2Da)
47. Rose Wang, Qingyang Zhang, Carly Robinson, Susanna Loeb, and Dorottya Demszky. Bridging the novice-expert gap via models of decision-making: A case study on remediating math mistakes. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics… [find](https://scholar.google.com/scholar?q=Rose+Wang%2C+Qingyang+Zhang%2C+Carly+Robinson%2C+Susanna+Loeb%2C+and+Dorottya+Demszky%2E+Bridging+the+novice%2Dexpert+gap+via+models+of+decision%2Dmaking%3A+A+case+study+on+remediating+math+mistakes%2E+In+Proceedings+o)
48. Rose E Wang, Ana T Ribeiro, Carly D Robinson, Susanna Loeb, and Dora Demszky. Tutor copilot: A human-ai approach for scaling real-time expertise. arXiv preprint arXiv:2410.03017, 2024. [find](https://scholar.google.com/scholar?q=Rose+E+Wang%2C+Ana+T+Ribeiro%2C+Carly+D+Robinson%2C+Susanna+Loeb%2C+and+Dora+Demszky%2E+Tutor+copilot%3A+A+human%2Dai+approach+for+scaling+real%2Dtime+expertise%2E+arXiv+preprint+arXiv%3A2410%2E03017%2C+2024%2E)
49. Canwen Xu, Corby Rosset, Ethan Chau, Luciano Corro, Shweti Mahajan, Julian McAuley, Jennifer Neville, Ahmed Awadallah, and Nikhil Rao. Automatic pair construction for contrastive post-training. In Findings of the Association for Computational Linguistics: NAACL 2024, pages 149–162, Mexico City, Mexi… [find](https://scholar.google.com/scholar?q=Canwen+Xu%2C+Corby+Rosset%2C+Ethan+Chau%2C+Luciano+Corro%2C+Shweti+Mahajan%2C+Julian+McAuley%2C+Jennifer+Neville%2C+Ahmed+Awadallah%2C+and+Nikhil+Rao%2E+Automatic+pair+construction+for+contrastive+post%2Dtraining%2E+In+Fin)
50. Wenda Xu, Guanglei Zhu, Xuandong Zhao, Liangming Pan, Lei Li, and William Wang. [Pride and prejudice: LLM amplifies self-bias in self-refinement](https://md.research.elamur.ai/free/en/papers/xu-2024-pride-prejudice-llm-amplifies-self/). In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 15474–15492, Bangkok, Thailand, A…
51. Boyang Yang, Haoye Tian, Weiguo Pian, Haoran Yu, Haitao Wang, Jacques Klein, Tegawendé F Bissyandé, and Shunfu Jin. Cref: An llm-based conversational software repair framework for programming tutors. In Proceedings of the 33rd ACM SIGSOFT International Symposium on Software Testing and Analysis, pag… [find](https://scholar.google.com/scholar?q=Boyang+Yang%2C+Haoye+Tian%2C+Weiguo+Pian%2C+Haoran+Yu%2C+Haitao+Wang%2C+Jacques+Klein%2C+Tegawend%C3%A9+F+Bissyand%C3%A9%2C+and+Shunfu+Jin%2E+Cref%3A+An+llm%2Dbased+conversational+software+repair+framework+for+programming+tutors%2E+)
52. Chuanpeng Yang, Yao Zhu, Wang Lu, Yidong Wang, Qian Chen, Chenlong Gao, Bingjie Yan, and Yiqiang Chen. Survey on knowledge distillation for large language models: Methods, evaluation, and application. ACM Trans. Intell. Syst. Technol., October 2024. [find](https://scholar.google.com/scholar?q=Chuanpeng+Yang%2C+Yao+Zhu%2C+Wang+Lu%2C+Yidong+Wang%2C+Qian+Chen%2C+Chenlong+Gao%2C+Bingjie+Yan%2C+and+Yiqiang+Chen%2E+Survey+on+knowledge+distillation+for+large+language+models%3A+Methods%2C+evaluation%2C+and+application%2E)
53. He Zhang, Jingyi Xie, Chuhao Wu, Jie Cai, ChanMin Kim, and John M Carroll. The future of learning: Large language models through the lens of students. In Proceedings of the 25th Annual Conference on Information Technology Education, pages 12–18, 2024. [find](https://scholar.google.com/scholar?q=He+Zhang%2C+Jingyi+Xie%2C+Chuhao+Wu%2C+Jie+Cai%2C+ChanMin+Kim%2C+and+John+M+Carroll%2E+The+future+of+learning%3A+Large+language+models+through+the+lens+of+students%2E+In+Proceedings+of+the+25th+Annual+Conference+on+I)

*Updated 3 November 2026*