Pith. sign in

REVIEW 4 major objections 8 minor 3 cited by

Generative AI in Education: From Foundational Insights to the Socratic Playground for Learning

T0 review · 4 major / 8 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read The paper claims that the Socratic Playground for Learning, a GPT-4-powered tutor, overcomes AutoTutor's limits, with a pilot showing significant gains in tutoring interactions and dialogue-based tutoring functionality.

desk verdict A useful design paper on an LLM-based Socratic tutor, but the central efficacy claim rests entirely on self-cited pilots not included here. read the letter →

arxiv 2501.06682 v1 pith:KYU3RZFL submitted 2025-01-12 cs.AI

Xiangen Hu , Sheng Xu , Richard Tong , Art Graesser This is my paper · ORCID
classification cs.AI
keywords intelligenttutoringsystemslargelanguagemodelsSocraticmethodAutoTutorexpectation–misconceptiontailoredfeedbacklearnercharacteristicscurvepersonalizedlearninggenerativeAIineducation
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

Drawing on the history of AutoTutor, one of the earliest dialogue-based tutoring systems, the paper argues that the bottleneck for AI tutoring is the pedagogical structure wrapped around the language model, not the model itself. AutoTutor's expectation–misconception tailored (EMT) approach worked in principle but was held back by shallow natural-language processing, scripted content, and costly scaling. The paper's constructive claim is that the Socratic Playground for Learning (SPL), built on GPT-4, removes those limits by combining transformer-based language understanding with the Socratic method. A pilot implementation is reported to have shown significant improvements in tutoring interactions and dialogue-based intelligent tutoring system functionality. The concrete artifact is a JSON tutoring prompt that scores each learner turn against weighted expectations and misconceptions and decides when the dialogue is complete.

What carries the argument

The load-bearing mechanism is an explicit expectation–misconception tailored loop re-implemented as a structured prompt. The JSON template contains weighted lists of expectations and misconceptions, pairing rules, and turn-by-turn scoring instructions using the Learner Characteristics Curve categories (Relevant-New, Irrelevant-New, Relevant-Old, Irrelevant-Old). The accumulated correct minus wrong contribution produces an Overall_Score, and the completion condition triggers a DONE status when that score exceeds 0.8. This machinery gives an LLM-driven tutor a transparent, auditable procedure for deciding what to say next and when to stop.

What would settle it

Give a fixed set of learner turns to the JSON-prompted model and to human raters using the same expectation and misconception lists; if LCC category agreement is low, or if learners whose sessions reach the 0.8 completion threshold show no greater learning gain than a control group, the central claim loses support.

Watch

Extended reading notes

Core claim

The paper asserts that modern LLM-based tutoring systems can realize the personalized, adaptive dialogue that AutoTutor promised but never fully delivered. SPL keeps AutoTutor's EMT loop—compare the learner's answer to expectations and misconceptions, then tailor hints, prompts, and feedback—while replacing older statistical language analysis with transformers and adding dynamic scenario generation. The system's five modes (assessment, tutoring, vicarious observation, gaming, and teachable agent) move learners from diagnosis to mastery. The included JSON prompt instructs the model to classify each contribution into Relevant-New, Irrelevant-New, Relevant-Old, and Irrelevant-Old categories, update accumulated correct and wrong contributions, and mark the dialogue DONE when the Overall_Score exceeds 0.8. The paper's evidence is a GPT-4-powered pilot described as demonstrating significant improvements in tutoring interactions and dialogue-based ITS functionalities.

Load-bearing premise

The load-bearing premise is that a large language model can reliably follow the JSON scoring instructions—categorizing learner statements into the LCC categories, respecting the weights, and judging semantic similarity—so that the scores and feedback it produces genuinely improve learning; this is asserted rather than demonstrated in the pilot.

Editorial extensions

If this is right

  • Because SPL generates scenarios and feedback on the fly, dialogue-based tutoring can scale to new subjects without hand-authoring every misconception and hint.
  • Educators can audit tutor decisions: the JSON output makes each turn's expectation and misconception matches and the resulting scores visible.
  • The five-mode progression gives learners a coherent route from initial assessment to deep mastery, with the teachable-agent mode as the capstone.
  • A tutoring dialogue can have an explicit stopping rule—Overall_Score above 0.8—rather than continuing indefinitely or ending arbitrarily.
  • The EMT and LCC machinery, once executed by an LLM, can be packaged as a modular prompt and reused across courses instead of being rebuilt per domain.

Reading between the lines

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

  • A natural next experiment, not run in the paper, would measure inter-rater agreement between the LLM's LCC classifications and human coders on the same learner turns; low agreement would mean the scoring loop is unreliable.
  • The JSON scaffold is model-agnostic in principle, so porting it to smaller or open-weight models would test whether the pedagogy or the model size is doing the work.
  • The 0.8 completion threshold is an arbitrary-looking design choice; calibrating it against external learning-gain measures would turn the stopping rule into an evidence-based mastery criterion.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 8 minor

Summary. This manuscript argues for a pedagogy-first integration of generative AI into education and introduces the Socratic Playground for Learning (SPL), a conversation-based intelligent tutoring system (ITS) powered by GPT-4 that is presented as overcoming AutoTutor's scalability and adaptivity constraints. The paper's first three sections develop an argument that parallels between LLMs and human cognition motivate conversation-based tutoring, while cautioning that technological capabilities only produce learning gains when embedded in sound pedagogical frameworks. Section 4 reviews AutoTutor's expectation-misconception tailored (EMT) dialogue mechanism, the Learner's Characteristics Curve (LCC) of Hu, Morrison, and Cai, and the ONR STEM Grand Challenge, documenting the project's unfulfilled aspirations. Section 5 describes SPL's five interactive modes (Assessment, Tutoring, Vicarious, Gaming, Teachable Agent) and asserts that a GPT-4-powered pilot implementation 'demonstrated significant improvements in tutoring interactions and dialogue-based ITS functionalities,' citing Zhang et al. (2024) and Liu et al. (2024) as the evidence.

Significance. If the pilot claims were substantiated, SPL would be a valuable proof-of-concept that modern LLMs can realize AutoTutor's EMT vision at scale, and the paper's synthesis of AutoTutor's legacy and the LCC framework would be a useful reference for ITS researchers. The manuscript has genuine strengths: it ships a concrete, auditable prompt artifact (Appendix B), explicitly ties design choices to established learning theory (Chi, Vygotsky, Paul and Elder), documents AutoTutor's scalability limits honestly, and articulates a defensible pedagogy-first stance. The five-mode progression is thoughtfully motivated and the discussion of equity and metacognitive dashboards is sensible. However, the scientific contribution currently rests on an unverified empirical claim whose evidence is entirely external and self-cited, and the scoring protocol that is supposed to produce the adaptive behavior is under-specified. The significance of the work cannot be assessed until that evidence is provided and the protocol is shown to be executable by the model.

major comments (4)
  1. [Section 5, paragraph 2.] The claim that a GPT-4-powered SPL pilot 'demonstrated significant improvements in tutoring interactions and dialogue-based ITS functionalities' is the paper's central empirical assertion, but no data from the pilot appear in the manuscript: there are no sample sizes, effect sizes, outcome measures, comparison conditions, or statistical tests. The only support is two citations, one of which (Zhang et al., 2024) is an unpublished manuscript with no venue, and the other (Liu et al., 2024) is from the same research group. Either the evaluation must be reported in this paper in sufficient detail for a reader to verify the claim, or the claim must be removed and the paper reframed as a design proposal; as written, Section 5 asserts an empirical result that the manuscript does not contain.
  2. [Appendix B (Listing 1).] The Appendix B scoring protocol is load-bearing because it is the mechanism by which SPL is supposed to drive adaptive feedback, but as specified in Listing 1 it is not internally consistent or reproducible. The apparent contradiction between 'Total scores for all categories should sum to 1' (Score_Computation) and 'Overall_Score... can be negative' (Scoring_Criteria) is actually reconcilable if category scores are nonnegative and sum to 1 per turn, since Overall_Score = Correct - Wrong can then be negative; the deeper problems are elsewhere. Expectations and Misconceptions each have weight pools that are required to 'sum to 1' (Initial_Interaction), and since Relevant_and_New scores are keyed to expectation weights while Irrelevant_and_New scores are keyed to misconception weights, a single learner turn draws on two independent pools that each sum to 1, so the category-sum constraint cannot hold in general. In addition, Accumulated_Correct_Contribution and Accumulated_Wrong_Contribution grow across turns with no stated normalization, so the completion threshold 'greater than 0.8' (Completion_Condition) has no well-defined meaning independent of the number of turns, and the prompt never specifies how partial credit based on 'degree of semantic similarity' is computed or how a turn that matches multiple expectations and misconceptions simultaneously is decomposed into the four category scores. These are definitional gaps, not stylistic ones, and they make the protocol non-reproducible as written.
  3. [Section 5; Section 4.3.] The evidentiary loop for SPL is self-referential: the scoring engine of the prompt is the authors' own LCC framework (Hu et al., 2013), the pilot studies cited as evidence are by the same group (Zhang et al., 2024; Liu et al., 2024), and the comparison baseline is AutoTutor's limitations as narrated by the authors themselves in Section 4. No external benchmark, independent evaluation, or neutral control condition is offered, so the reader cannot determine whether the claimed improvements come from the EMT/LCC pedagogy, from GPT-4's general conversational competence, or from interface novelty. A minimal remedy is to report inter-rater agreement between the LLM's LCC classifications and human coders, and to include a baseline condition that isolates the contribution of the JSON scoring protocol (e.g., the same tutor without the LCC scoring instructions).
  4. [Section 5.3 and Appendix B.] The entire SPL contribution depends on the untested premise that GPT-4 can reliably execute the Appendix B protocol: labeling each learner turn as Relevant/Irrelevant and New/Old, detecting contradictions and redundancies across turns, respecting the weight constraints, and emitting valid JSON with correct status transitions. The manuscript provides no evidence for this premise, including no confusion matrices, no agreement statistics against human LCC coders, no calibration of the 0.8 completion threshold, and no actual system transcripts (the Appendix A walkthroughs are hand-authored illustrations of intended behavior). If the model's classifications are noisy or systematically biased, the adaptive feedback loop central to the claimed learning benefits does not exist. This premise is testable with existing data from the cited pilots and should be tested before the central claim can be evaluated.
minor comments (8)
  1. [Section 1.2.] The specific claims about OpenAI's o3 model (96.7% on the 2024 AIME, doctoral-level performance in physics, chemistry, and biology) are given without a citation; a source should be added.
  2. [References.] Zhang et al. (2024) is listed without a publication venue; a DOI or repository link should be provided so the pilot can be located and verified.
  3. [Section 5.] The abstract and Section 5 call SPL a 'next-generation ITS,' but the manuscript reports no implementation details or deployment status; one sentence should clarify that the pilot evidence is reported in the cited papers and only summarized here.
  4. [Section 4.3.] The text says LCC was 'a key mechanism for AutoTutor's adaptivity in the ONR STEM Grand Challenge,' but the same section describes the initiative's aspirations as unfulfilled; please clarify whether LCC was actually implemented and run in that program or only designed for it.
  5. [Appendix A and figures.] Figures 3-8 are described in the text and captions but were not accessible in the version under review; please ensure all figures are present in the published version and that their content matches the captions (for example, the weights '0.5, 0.3, 0.2' discussed in Section A.2 should appear in Figure 5).
  6. [Sections 1-2.] The manuscript contains numerous missing-space typos (e.g., 'enrichthat experience,' 'Giventhat,' 'suchas,' 'forexample,' 'outcomesfor' in Sections 1-2) and inconsistent use of 'chapter' versus 'section' (e.g., Section 2.1 refers to 'the final subsection of the previous chapter'); a full copyedit is needed.
  7. [Appendix B.2.] Listing 1 contains placeholders '${Consider_Context()}' and '${theLang}' that are never explained; Section B.2 should state how these placeholders are instantiated in practice.
  8. [Section 3.3.] The subsection 'Research and Development Framework' is empty: its content appears in Sections 3.4 and 3.5, so the heading should either be removed or the promised framework content should be added under it.

Circularity Check

2 steps flagged · score 6.0 of 10

SPL's central efficacy claim is supported only by same-author pilot citations, and its scoring rubric is the authors' own LCC framework, forming a self-referential evidence chain.

  1. self citation load bearing [Section 5, 'The Socratic Playground for Learning', second paragraph]
    "A pilot implementation of the Socratic Playground for Learning (SPL), powered by GPT-4, demonstrated significant improvements in tutoring interactions and dialogue-based ITS functionalities. (Zhang et al., 2024; Liu et al., 2024)"

    This is the paper's only empirical support for SPL's central efficacy claim. The two cited pilot papers share authors with this manuscript: Zhang et al. (2024) lists Xu and Hu; Liu et al. (2024) lists Hu. No data, effect sizes, sample sizes, outcome measures, or independent evaluation appear in this preprint, so the claim that the pilot 'demonstrated significant improvements' is carried entirely by a self-citation chain rather than by evidence presented or independently reproduced in the paper.

  2. self citation load bearing [Section 4.3 (LCC definition) and Appendix B, 'Scoring_Criteria' / 'Score_Computation']
    "A key mechanism for AutoTutor's adaptivity in the ONR STEM Grand Challenge was the Learner's Characteristics Curve (LCC), originally developed by Hu, Morrison, and Cai (Hu et al., 2013)."

    The SPL prompt presented as the working heart of the system (Appendix B) operationalizes exactly these LCC categories as its scoring rubric. The claimed improvements of SPL are therefore evaluated through a framework developed by the same authors, with no inter-rater reliability, confusion matrix, or external validation of the LCC scoring provided in this manuscript. The rubric is imported as established and then used as the measure of system success, making the evaluation instrument self-referential.

full rationale

Most of the manuscript is a position/design narrative (cognitive parallels, AutoTutor history, Socratic modes, JSON prompt) and does not derive quantitative predictions, so no fitted-input or definitional circularity arises in those sections. The circularity is concentrated in the empirical claim: Section 5 asserts that an SPL pilot 'demonstrated significant improvements' and cites only Zhang et al. 2024 and Liu et al. 2024, both with present-author overlap; no supporting statistics or independent benchmark are included. The evaluation mechanism behind that claim is the authors' LCC framework from Hu et al. 2013, embedded in the Appendix B prompt. Thus the central efficacy claim rests on a self-citation chain and a self-authored scoring rubric lacking in-manuscript validation. The absence of any external benchmark or reproducibility evidence in the manuscript is the key reason this is not a 0-2 case. The score is 6 rather than higher because the paper remains a design/pedagogy proposal with substantial non-circular content (AutoTutor review, Socratic mode taxonomy, prompt engineering illustration), and no equation-level reduction of a prediction to an input was found.

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

The paper introduces no new theoretical entities. Its design artifact, the JSON prompt, relies on the authors' existing LCC and EMT frameworks, which are carried over from previous publications. The free parameters are the hand-chosen weights and thresholds in the prompt.

free parameters (2)
  • LCC weights for expectations and misconceptions = e.g., 0.5, 0.3, 0.2 (as shown in the seatbelt example)
    The JSON prompt assigns weights to expectations and misconceptions by hand, without any empirical calibration. These weights affect the scoring and completion decision.
  • Completion threshold for Overall_Score = 0.8
    In the JSON prompt's Completion_Condition, a status of DONE is set when Overall_Score is greater than 0.8. This threshold is chosen arbitrarily and is not derived from learning data.
assumptions (4)
  • domain assumption Socratic questioning improves learning outcomes.
    The paper relies on the Socratic method as an effective pedagogy (Section 3.1) without providing new empirical support.
  • domain assumption LLMs can emulate human tutoring dialogue sufficiently to foster learning.
    The premise of the Socratic Playground is that transformer models can interpret learner inputs and generate effective tutoring moves (Section 5).
  • domain assumption The LCC categories (Relevant-New, Irrelevant-New, Relevant-Old, Irrelevant-Old) are a valid model of learner progress.
    The JSON prompt's scoring criteria are based on LCC, which is taken from the authors' own prior work (Hu et al. 2013) without independent validation.
  • ad hoc to paper The cited pilot studies (Zhang et al. 2024; Liu et al. 2024) are methodologically sound.
    The paper relies on these self-cited pilot studies for the 'significant improvements' claim, but does not include their data or detailed methodology.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Generative AI in Education: From Foundational Insights to the Socratic Playground for Learning." pith.science (2026). https://pith.science/paper/KYU3RZFL

@misc{pith2026250106682,
  author       = {Pith},
  title        = {Pith review of: Generative AI in Education: From Foundational Insights to the Socratic Playground for Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KYU3RZFL}},
  note         = {Machine review of arXiv:2501.06682}
}
read the original abstract

This paper explores the synergy between human cognition and Large Language Models (LLMs), highlighting how generative AI can drive personalized learning at scale. We discuss parallels between LLMs and human cognition, emphasizing both the promise and new perspectives on integrating AI systems into education. After examining challenges in aligning technology with pedagogy, we review AutoTutor-one of the earliest Intelligent Tutoring Systems (ITS)-and detail its successes, limitations, and unfulfilled aspirations. We then introduce the Socratic Playground, a next-generation ITS that uses advanced transformer-based models to overcome AutoTutor's constraints and provide personalized, adaptive tutoring. To illustrate its evolving capabilities, we present a JSON-based tutoring prompt that systematically guides learner reflection while tracking misconceptions. Throughout, we underscore the importance of placing pedagogy at the forefront, ensuring that technology's power is harnessed to enhance teaching and learning rather than overshadow it.

Figures

Figures reproduced from arXiv: 2501.06682 by the authors.

Figure 1
Figure 1. The EMT framework in AutoTutor, illustrating how the system compares a stu [PITH_FULL_IMAGE:figures/full_fig_p012_1.png] view at source ↗
Figure 2
Figure 2. High-level design of the ONR STEM Grand Challenge application, integrating [PITH_FULL_IMAGE:figures/full_fig_p015_2.png] view at source ↗
Figure 3
Figure 3. An adaptive self-assessment interaction showing both the learner’s question/answer [PITH_FULL_IMAGE:figures/full_fig_p024_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: A seatbelt-themed tutoring scenario. The learner initially expresses misconceptions [PITH_FULL_IMAGE:figures/full_fig_p026_4.png]
Figure 5
Figure 5. Figure 5: LCC analysis of each learner contribution, showing scores for Relevant/Irrelevant [PITH_FULL_IMAGE:figures/full_fig_p028_5.png]
Figure 6
Figure 6. Figure 6: A Vicarious Mode interface with virtual discussants (Alice, Bob, Charlie, Diana, [PITH_FULL_IMAGE:figures/full_fig_p030_6.png]
Figure 7
Figure 7. Figure 7: A Jeopardy-Style interface for practicing acceleration concepts. Learners choose [PITH_FULL_IMAGE:figures/full_fig_p032_7.png]
Figure 8
Figure 8. Figure 8: A teachable-agent interface: The system, acting as Casey, expresses a misunder [PITH_FULL_IMAGE:figures/full_fig_p035_8.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Understanding How University Guidelines Address Privacy and Security Issues of Generative AI in Academic Settings

    cs.HC 2025-06 conditional novelty 5.0 of 10

    Qualitative analysis of 46 university GenAI policy documents shows privacy and security concerns are acknowledged but inconsistently addressed, with vague terminology, reliance on existing frameworks, and limited conc...

  2. Rethinking Higher Education: From Fixed Curricula to Learnity Graphs

    cs.CY 2026-08 conditional novelty 3.0 of 10

    Learning can be represented as an evolving graph of knowledge, skills, experience, and artifacts, called a learnity graph, proposed as a basis for lifelong personalized education.

  3. Robust pid sliding mode control for dc servo motor speed control

    eess.SY 2025-08 unverdicted novelty 2.0 of 10

    An abstract-only claim that SMC-PID outperforms PID for DC servo motor speed on the CE110 trainer; the submitted body text is an unrelated paper, so the result is unverifiable.

Reference graph

Works this paper leans on

40 extracted references · 38 canonical work pages · cited by 3 Pith papers

  1. [1]

    Athanassopoulos, S., Manoli, P., Gouvi, M., Lavidas, K., and Komis, V. (2023). The use of chatgpt as a learning tool to improve foreign language writing in a multilingual and multicultural classroom. Advances in Mobile Learning Educational Research , 3(2):818--824

    work page 2023

  2. [2]

    and Ansah, L

    Baidoo-Anu, D. and Ansah, L. O. (2023). Education in the era of generative artificial intelligence (ai): Understanding the potential benefits of chatgpt in promoting teaching and learning. Journal of AI , 7(1):52--62

    work page 2023

  3. [3]

    J., Pashler, H., Vul, E., Wixted, J

    Cepeda, N. J., Pashler, H., Vul, E., Wixted, J. T., and Rohrer, D. (2006). Distributed practice in verbal recall tasks: A review and quantitative synthesis. Psychological Bulletin , 132(3):354--380

    work page 2006

  4. [4]

    Chi, M. T. H. (2009). Active-constructive-interactive: A conceptual framework for differentiating learning activities. Topics in Cognitive Science , 1(1):73--105

    work page 2009

  5. [5]

    Clark, R. C. and Mayer, R. E. (2016). E-learning and the science of instruction: Proven guidelines for consumers and designers of multimedia learning . John Wiley & Sons

    work page 2016

  6. [6]

    and Halverson, R

    Collins, A. and Halverson, R. (2018). Rethinking Education in the Age of Technology: The Digital Revolution and Schooling in America . Teachers College Press

    work page 2018

  7. [7]

    Cooper, G. (2023). Examining science education in chatgpt: An exploratory study of generative artificial intelligence. Journal of Science Education and Technology , 32(3):444--452

    work page 2023

  8. [8]

    Council, N. R. (2000). How People Learn: Brain, Mind, Experience, and School . National Academy Press, Washington, DC

    work page 2000

Show all 40 references
  1. [9]

    Facione, P. A. (1990). Critical thinking: A statement of expert consensus for purposes of educational assessment and instruction. Delphi Report: Committee on Pre-College Philosophy of the American Philosophical Association

    1990

  2. [10]

    Gligorea, I., Cioca, M., Oancea, R., Gorski, A.-T., Gorski, H., and Tudorache, P. (2023). Adaptive learning using artificial intelligence in e-learning: A literature review. Education Sciences , 13(12)

    2023

  3. [11]

    C., Person, N

    Graesser, A. C., Person, N. K., et al. (2004). Autotutor: A cognitive system that simulates a tutor that facilitates learning through mixed-initiative dialogue. Cognitive Systems: Human Cognitive Models in Systems Design , 2:19--31

    2004

  4. [12]

    Heffernan, N. T. et al. (2020). Extending the knowledge tracing model to multilingual content. International Journal of Artificial Intelligence in Education , 30:581--606

    2020

  5. [13]

    Hu, X. (2018). Introduction to team tutoring. In Sottilare, R., Graesser, A., Hu, X., and Sinatra, A., editors, Design Recommendations for Intelligent Tutoring Systems: Volume 6 - Team Tutoring , volume 6, chapter 1, pages 19--21

    2018

  6. [14]

    M., and Cai, Z

    Hu, X., Morrison, D. M., and Cai, Z. (2013). On the use of learner micromodels as partial solutions to complex problems in a multiagent, conversation-based intelligent tutoring system. In Sottilare, R., Graesser, A., Hu, X., and Holden, H., editors, Design Recommendations for ...

    2013

  7. [15]

    L., and Kim, J

    Hu, X., Tong, R., Cai, Z., Cockroft, J. L., and Kim, J. W. (2019). Self-improvable adaptive instructional systems ( SIAIS s) – a proposed model. In Sinatra, A., Graesser, A., Hu, X., Brawner, K., and Rus, V., editors, Design Recommendations for Intelligent Tutoring Systems: Vo...

    2019

  8. [16]

    Kasneci, E., Sessler, K., Küchemann, S., Bannert, M., Dementieva, D., Fischer, F., Gasser, U., Groh, G., Günnemann, S., Hüllermeier, E., Krusche, S., Kutyniok, G., Michaeli, T., Nerdel, C., Pfeffer, J., Poquet, O., Sailer, M., Schmidt, A., Seidel, T., Stadler, M., Weller, J., ...

    2023

  9. [17]

    Koehler, M. J. and Mishra, P. (2009). What is technological pedagogical content knowledge? Contemporary Issues in Technology and Teacher Education , 9(1):60--70

    2009

  10. [18]

    Li, X., Li, B., and Cho, S.-J. (2023). Empowering chinese language learners from low-income families to improve their chinese writing with chatgpt’s assistance afterschool. Languages , 8(4):238

    2023

  11. [19]

    Liu, S., Guo, X., Hu, X., and Zhao, X. (2024). Advancing generative intelligent tutoring systems with gpt-4: Design, evaluation, and a modular framework for future learning platforms. Electronics , 13(24)

    2024

  12. [20]

    Luckin, R., Holmes, W., Griffiths, M., and Forcier, L. B. (2016). Intelligence unleashed: An argument for ai in education. In Pearson Report

    2016

  13. [21]

    and Deroy, A

    Maity, S. and Deroy, A. (2024). Generative ai and its impact on personalized intelligent tutoring systems. arXiv preprint arXiv:2410.10650

    2024 arXiv

  14. [22]

    Mayer, R. E. (2002). Multimedia learning. Psychology of Learning and Motivation , 41:85--139

    2002

  15. [23]

    and Doleck, T

    Memarian, B. and Doleck, T. (2023). Chatgpt in education: Methods, potentials and limitations. Computers in Human Behavior: Artificial Humans , page 100022

    2023

  16. [24]

    C., and Hu, X

    Nye, B., Graesser, A. C., and Hu, X. (2014). Deep learning by explaining why correct answers are correct and incorrect answers are incorrect. Frontiers in Psychology , 5:715

    2014

  17. [25]

    Gpt-4 technical report

    OpenAI (2023). Gpt-4 technical report. https://openai.com

    2023

  18. [26]

    and Elder, L

    Paul, R. and Elder, L. (2008). The miniature guide to critical thinking concepts and tools. Foundation for Critical Thinking Press

    2008

  19. [27]

    Perrotta, C., Featherstone, G., Aston, H., and Houghton, E. (2013). Promoting technology-enhanced learning: a critical study of the current approaches. British Journal of Educational Technology , 44(2):E21--E23

    2013

  20. [28]

    N., Leinonen, J., MacNeil, S., Randrianasolo, A

    Prather, J., Reeves, B. N., Leinonen, J., MacNeil, S., Randrianasolo, A. S., Becker, B. A., Kimmel, B., Wright, J., and Briggs, B. (2024). The widening gap: The benefits and harms of generative ai for novice programmers. In Proceedings of the 2024 ACM Conference on Internation...

    2024

  21. [29]

    and Huang, Y.-M

    Shadiev, R. and Huang, Y.-M. (2020). Investigating student attention, meditation, cognitive load, and satisfaction during lectures in a foreign language supported by speech-enabled language translation. Computer Assisted Language Learning , 33(3):301--326

    2020

  22. [30]

    Shetye, S. (2024). An evaluation of khanmigo, a generative ai tool, as a computer-assisted language learning app. Studies in Applied Linguistics and TESOL , 24(1)

    2024

  23. [31]

    Tong, R., Rowe, J., and Goldberg, B. (2019). Architecture implications for building macro and micro level self-improving aiss. In Sinatra, A., Graesser, A., Hu, X., Brawner, K., and Rus, V., editors, Design Recommendations for Intelligent Tutoring Systems: Volume 7 - Self-Impr...

    2019

  24. [32]

    J., Cao, C

    Tong, R. J., Cao, C. C., Lee, T. X., Zhao, G., Wan, R., Wang, F., Hu, X., Schmucker, R., Pan, J., Quevedo, J., et al. (2023). Neolaf, an llm-powered neural-symbolic cognitive architecture. arXiv preprint arXiv:2308.03990

    2023 arXiv

  25. [33]

    Tong, R. J. and Hu, X. (2024). Future of education with neuro-symbolic ai agents in self-improving adaptive instructional systems. Frontiers of Digital Education , 1(2):198

    2024

  26. [34]

    Tong, R. J. and Lee, T. X. (2023). Trustworthy AI that engages humans as partners in teaching and learning. Computer , 56(5):62--73

    2023

  27. [35]

    N., Łukasz Kaiser, and Polosukhin, I

    Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Łukasz Kaiser, and Polosukhin, I. (2017). Attention is all you need. Advances in Neural Information Processing Systems , 30

    2017

  28. [36]

    Vygotsky, L. S. (1978). Mind in Society: The Development of Higher Psychological Processes . Harvard University Press, Cambridge, MA

    1978

  29. [37]

    Woolf, B. P. (2021). Building Intelligent Interactive Tutors: Student-centered strategies for revolutionizing e-learning . Elsevier (Morgan Kaufmann)

    2021

  30. [38]

    Zhai, C., Wibowo, S., and Li, L. D. (2024). The effects of over-reliance on ai dialogue systems on students' cognitive abilities: a systematic review. Smart Learning Environments , 11(1):28

    2024

  31. [39]

    Zhang, L., Lin, J., Kuang, Z., Xu, S., and Hu, X. (2024). Spl: A socratic playground for learning powered by large language model

    2024

  32. [40]

    X., Zhou, K., Li, J., Tang, T., Wang, X., Hou, Y., Min, Y., Zhang, B., Zhang, J., Dong, Z., et al

    Zhao, W. X., Zhou, K., Li, J., Tang, T., Wang, X., Hou, Y., Min, Y., Zhang, B., Zhang, J., Dong, Z., et al. (2023). A survey of large language models. arXiv preprint arXiv:2303.18223

    2023 arXiv

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.