#StudentSim
LightNews LightNews
glonce.com
Microsoft's StudentSim trains AI tutors on simulated students
Microsoft and the University of Illinois built StudentSim, which creates a digital replica of each student from just a handful of past records so AI tutors can get fast, cheap feedback; the researchers report it beats GPT-5.4 at…
Microsoft's StudentSim trains AI tutors on simulated students
Microsoft and the University of Illinois built StudentSim, which creates a digital replica of each student from just a handful of past records so AI tutors can get fast, cheap feedback; the researchers report it beats GPT-5.4 at mimicking real students across chess, English, and math.
glonce.com
September 21, 2026 at 7:34 AM
aidailypost.com
Imagine a chess coach that trains itself on virtual pupils—StudentSim lets an AI tutor practice on simulated students, speeding up GPT‑5.4’s learning curve. Microsoft’s adaptive learning breakthrough could change AI tutoring. #AITutoring #StudentSim #AdaptiveLearning

🔗
September 20, 2026 at 10:11 AM
sipirtu.com
Microsoft and the University of Illinois built StudentSim to replicate individual students from limited data and give AI tutors fast, low-cost feedback.

Source: The Decoder
Simulated students that make realistic mistakes help AI tutors learn faster
Microsoft and the University of Illinois built StudentSim to replicate individual students from limited data and give AI tutors fast, low-cost feedback. In tests covering 60 students across chess, English, and math, it outperformed GPT-5.4. A chess tutor trained with StudentSim also earned the highest expert ratings among three versions tested.
the-decoder.com
September 20, 2026 at 10:18 AM
ai-sight.bsky.social
System StudentSim buduje spersonalizowane symulacje konkretnych uczniów na podstawie zaledwie kilku próbek ich pracy, odtwarzając ich unikalne błędy i reakcje na wskazówki. Pozwala to na skuteczne trenowanie korepetytorów AI bez udziału prawdziwych ludzi i zaporowych kosztów zbierania danych.
Microsoft Research: Cyfrowe repliki uczniów nauczą AI, jak pomagać ludziom
System StudentSim buduje spersonalizowane symulacje konkretnych uczniów na podstawie zaledwie kilku próbek ich pracy, odtwarzając ich unikalne błędy i reakcje na wskazówki. Pozwala to na skuteczne trenowanie korepetytorów AI bez udziału prawdziwych ludzi i zaporowych kosztów zbierania danych.
aisight.pl
September 21, 2026 at 3:02 PM
ainieuwtjes.bsky.social
Simulated students that make realistic mistakes help AI tutors learn faster

Microsoft and the University of Illinois built StudentSim to replicate individual students from limited data and give AI tutors fast, low-cost feedback. In tests covering 60 students across chess, Eng

Via The Decoder
Simulated students that make realistic mistakes help AI tutors learn faster
the-decoder.com
September 20, 2026 at 7:36 PM
aspihuk.bsky.social
🚀 ASPiH and iRIS are thrilled to bring you Student-Sim Stars 2025! 🎉🚀

🔗 irissimulationauthoring.com/studentsim/

Sign up for a launch webinar to find out more:

Webinar 11th April: events.teams.microsoft.com/event/e7c290...

Webinar: 30th April: events.teams.microsoft.com/event/3ab95f...
April 7, 2025 at 10:17 AM
aspihuk.bsky.social
ASPiH and iRIS are thrilled to bring you Student-Sim Stars 2025! 🎉

If you would like to find out more join us for a launch webinar. Sign up below

Webinar: 30th April: events.teams.microsoft.com/event/3ab95f...

🔗 irissimulationauthoring.com/studentsim/
April 16, 2025 at 10:36 AM
aspihuk.bsky.social
This is your final reminder to register your student team for Student-Sim Stars 2025! 🎉

9th May: Deadline to register your student-team

Don’t miss your chance to be a part of Student-Sim Stars 2025! ✨🏆

🔗 irissimulationauthoring.com/studentsim/

#ASPiH #iRIS
May 8, 2025 at 9:15 AM
aspihuk.bsky.social
ASPiH & iRIS are thrilled to bring you Student-Sim Stars 2025! 🎉

This is your reminder that you have 1 week left to register your student team!

9th May: Deadline!

Don’t miss your chance to be a part of Student-Sim Stars 2025! ✨🏆

🔗 irissimulationauthoring.com/studentsim/
May 2, 2025 at 9:33 AM
paper.bsky.social
(1/1) 483 Likes, 4 Comments, 02 Sep 2026, Hugging Face
Paper page - StudentSim: Training LLM-based Student Simulators
Join the discussion on this paper page
huggingface.co
September 8, 2026 at 12:09 AM
paper.bsky.social
Links: abs, pdf
Search: Bluesky, Twitter, Reddit, Hacker News, Hugging Face, alphaXiv
StudentSim: Training LLM-based Student Simulators
AI tutors are most useful when they adapt to each student's strengths, weaknesses, and preferred guidance, but evidence about which guidance works for which student is sparse, slow, and costly to coll...
arxiv.org
September 8, 2026 at 12:09 AM
paper.bsky.social
[13/30] 483 Likes, 4 Comments, 1 Posts
2609.01591, cs․CL, 01 Sep 2026

🆕StudentSim: Training LLM-based Student Simulators

Ke Yang, Chenglong Wang, Michel Galley, Chandan Singh, Jeevana Priya Inala, ChengXiang Zhai, Jianfeng Gao
AI tutors are most useful when they adapt to each student's strengths, weaknesses, and preferred guidance, but evidence about which guidance works for which student is sparse, slow, and costly to collect from real learners. Student simulators can provide this signal as a proxy, yet existing approaches are limited: state-tracking models fit student behavior but struggle to process explanations or corrections, while LLM role-play follows guidance fluently but does not reliably match the competence of the student being imitated. We present StudentSim, a training framework that turns sparse per-student data into individualized simulators through pooled training followed by per-student specialization. The resulting simulators both mirror a student's own responses and update them under tutor guidance. We also introduce StudentSimEval, a standardized protocol covering 60 students across chess, second-language English writing, and mathematics, using public learner datasets with de-identified records shared for research. StudentSimEval measures behavioral fidelity (F), or how well a simulator matches a student's responses, and guidance responsiveness (R), or how readily it updates under tutor guidance, with all methods fit and evaluated on the same records. Across all three domains, StudentSim outperforms GPT-5.4 on both metrics. In chess, StudentSim reaches F=0.51 and R=0.91, compared with 0.23 and 0.72 for GPT-5.4 and 0.45 and 0.27 for Maia2. As a proof of concept, using StudentSim as a reward model for tutor reinforcement learning produces a chess tutor that expert humans rate as more accurate, better-guided, and more personalized than a no-RL baseline and a tutor trained against a GPT-5.4 simulator reward. Code is available at https://github.com/microsoft/StudentSim.
September 8, 2026 at 12:09 AM
pneumetron.bsky.social
📌 StudentSim: Bridging the Gap in AI Tutor Training

https://pneumetron.com/news/ai_research/studentsim-llm-student-simulators-18973e

#AIEducation #LLM #ReinforcementLearning #StudentSim
September 5, 2026 at 6:43 AM
nanoatp.bsky.social
(1/1) 459 Upvotes, 2 Comments, 01 Sep 2026, Hugging Face
Paper page - StudentSim: Training LLM-based Student Simulators
Join the discussion on this paper page
huggingface.co
September 3, 2026 at 12:07 AM
oludai.bsky.social
287 upvotes for « StudentSim: Training LLM-based Student Simulators » — simulating learners might finally make adaptive tutoring testable at scale.

https://huggingface.co/papers/2609.01591

#AI #MachineLearning #Research
September 2, 2026 at 1:01 PM
nanoatp.bsky.social
Links: abs, pdf
Search: Bluesky, Twitter, Reddit, Hacker News, Hugging Face, alphaXiv
StudentSim: Training LLM-based Student Simulators
AI tutors are most useful when they adapt to each student's strengths, weaknesses, and preferred guidance, but evidence about which guidance works for which student is sparse, slow, and costly to coll...
arxiv.org
September 3, 2026 at 12:07 AM
cscl-bot.bsky.social
Ke Yang, Chenglong Wang, Michel Galley, Chandan Singh, Jeevana Priya Inala, ChengXiang Zhai, Jianfeng Gao: StudentSim: Training LLM-based Student Simulators https://arxiv.org/abs/2609.01591 https://arxiv.org/pdf/2609.01591 https://arxiv.org/html/2609.01591
September 2, 2026 at 6:41 AM
nanoatp.bsky.social
[7/30] 459 Upvotes, 2 Comments, 1 Posts, arXiv:2609.01591

🆕StudentSim: Training LLM-based Student Simulators

Ke Yang, Chenglong Wang, Michel Galley, Chandan Singh, Jeevana Priya Inala, ChengXiang Zhai
AI tutors are most useful when they adapt to each student's strengths, weaknesses, and preferred guidance, but evidence about which guidance works for which student is sparse, slow, and costly to collect from real learners. Student simulators can provide this signal as a proxy, yet existing approaches are limited: state-tracking models fit student behavior but struggle to process explanations or corrections, while LLM role-play follows guidance fluently but does not reliably match the competence of the student being imitated. We present StudentSim, a training framework that turns sparse per-student data into individualized simulators through pooled training followed by per-student specialization. The resulting simulators both mirror a student's own responses and update them under tutor guidance. We also introduce StudentSimEval, a standardized protocol covering 60 students across chess, second-language English writing, and mathematics, using public learner datasets with de-identified records shared for research. StudentSimEval measures behavioral fidelity (F), or how well a simulator matches a student's responses, and guidance responsiveness (R), or how readily it updates under tutor guidance, with all methods fit and evaluated on the same records. Across all three domains, StudentSim outperforms GPT-5.4 on both metrics. In chess, StudentSim reaches F=0.51 and R=0.91, compared with 0.23 and 0.72 for GPT-5.4 and 0.45 and 0.27 for Maia2. As a proof of concept, using StudentSim as a reward model for tutor reinforcement learning produces a chess tutor that expert humans rate as more accurate, better-guided, and more personalized than a no-RL baseline and a tutor trained against a GPT-5.4 simulator reward. Code is available at https://github.com/microsoft/StudentSim.
September 3, 2026 at 12:07 AM
nanoatp.bsky.social
AIチューターは、生徒一人ひとりの長所、短所、そして希望する指導方法に合わせて対応できる場合に最も有用ですが、どの指導方法がどの生徒に効果的かに関する証拠は乏しく、実際の学習者から収集するには時間がかかり、コストもかかります。

学習者シミュレーターはこの信号を代用として提供できるが、既存のアプローチには限界がある。状態追跡モデルは学習者の行動には適合するが、説明や修正の処理には苦戦する一方、LLMによるロールプレイは指導内容を流暢に実行できるものの、模倣対象の学習者の能力を確実に再現することはできない。

本稿では、学生ごとの疎なデータを、プール学習とそれに続く学生ごとの特化処理を通じ...
AIチューターは、生徒一人ひとりの長所、短所、そして希望する指導方法に合わせて対応できる場合に最も有用ですが、どの指導方法がどの生徒に効果的かに関する証拠は乏しく、実際の学習者から収集するには時間がかかり、コストもかかります。 学習者シミュレーターはこの信号を代用として提供できるが、既存のアプローチには限界がある。状態追跡モデルは学習者の行動には適合するが、説明や修正の処理には苦戦する一方、LLMによるロールプレイは指導内容を流暢に実行できるものの、模倣対象の学習者の能力を確実に再現することはできない。 本稿では、学生ごとの疎なデータを、プール学習とそれに続く学生ごとの特化処理を通じて、個別化されたシミュレータに変換するトレーニングフレームワーク「StudentSim」を提案する。 こうして作成されたシミュレーターは、生徒自身の回答を反映するとともに、指導者の指導のもとでその回答を更新していく。 また、研究目的で共有された匿名化済みの公開学習者データセットを用い、チェス、第二言語としての英語作文、数学の3分野にわたる60名の学生を対象とした標準化されたプロトコル「StudentSimEval」も紹介します。 StudentSimEvalは、行動の忠実度（F）（シミュレータが学習者の反応とどの程度一致しているか）と、指導への反応性（R）（チューターの指導の下でシミュレータがどの程度迅速に更新されるか）を測定するもので、すべての手法は同一の記録データに基づいて適合・評価される。 3つの領域すべてにおいて、StudentSimは両方の評価指標でGPT-5.4を上回る性能を示している。 チェスにおいて、StudentSimはF=0.51、R=0.91を達成したのに対し、GPT-5.4は0.23と0.72、Maia2は0.45と0.27であった。 概念実証として、StudentSimをチューターの強化学習における報酬モデルとして用いたところ、人間の専門家が評価したところ、強化学習（RL）を使用しないベースラインや、GPT-5.4シミュレータの報酬を用いて学習させたチューターと比較して、より正確で、より適切な指導を行い、より個人に合わせたチェスのチューターが生成された。 コードは https://github.com/microsoft/StudentSim で公開されています。
September 3, 2026 at 12:07 AM
aspihuk.bsky.social
ASPiH and iRIS are thrilled to bring you Student-Sim Stars 2025! 🎉

If you would like to find out more join us tomorrow for a launch webinar.

Webinar: 30th April: events.teams.microsoft.com/event/3ab95f...

🔗 irissimulationauthoring.com/studentsim/

9 May: Deadline to register your student-team
April 29, 2025 at 9:05 AM
aspihuk.bsky.social
ASPiH and iRIS are thrilled to bring you Student-Sim Stars 2025! 🎉

If you would like to find out more join us for a launch webinar. Sign up below to register your attendance.

🔗 irissimulationauthoring.com/studentsim/

Webinar 11th April: events.teams.microsoft.com/event/e7c290...
April 10, 2025 at 10:10 AM
healthysim.bsky.social
studentSim – iRIS
Today at Midnight is the final day to register your team for the #ASPiH & #iRIS inaugural Student-Sim Stars competition - 16 Sept Register your team - 27 Sept Submit scenario - 4 Oct Finalist Announcement - 4-5 Nov shortlisted teams invited to #ASPiH2024
buff.ly
February 7, 2025 at 11:55 PM
healthysim.bsky.social
studentSim – iRIS
You have 1 week to register your team for the #ASPiH & #iRIS inaugural Student-Sim Stars competition: - 16 Sept Register your team - 27 Sept Submit scenario - 4 Oct Finalist Announcement - 4-5 Nov shortlisted teams invited to #ASPiH2024 Don’t miss out!
buff.ly
January 13, 2025 at 7:15 PM
healthysim.bsky.social
studentSim – iRIS
2 weeks left to register your team for the #ASPiH & #iRIS inaugural Student-Sim Stars competition -16 Sept Register your team -27 Sept Submit scenario -4 Oct Finalist Announcement -4-5 Nov shortlisted teams invited to #ASPiH2024 Next webinars 4 Sept
buff.ly
December 26, 2024 at 3:59 PM
healthysim.bsky.social
studentSim – iRIS
Key dates for entering a team for the #ASPiH & #iRIS inaugural Student-Sim Stars competition: - 16 Sept Register your team - 27 Sept Submit scenario - 4 Oct Finalist Announcement - 4-5 Nov shortlisted teams invited to #ASPiH2024 Next webinars: 4 Sept
buff.ly
December 18, 2024 at 2:53 AM