Design Principles and Observable Indicators for AI-Enabled Pedagogical Accompaniment: Evidence from the Amico Dual-Mode Prototype in Italy and China

Pier Paolo Benedetti

arxiv: 2605.20665 · v1 · pith:JVWCOWZ6new · submitted 2026-05-20 · 💻 cs.HC · cs.CY

Design Principles and Observable Indicators for AI-Enabled Pedagogical Accompaniment: Evidence from the Amico Dual-Mode Prototype in Italy and China

Pier Paolo Benedetti This is my paper

Pith reviewed 2026-05-21 04:09 UTC · model grok-4.3

classification 💻 cs.HC cs.CY

keywords AI in educationpedagogical designhuman-AI collaborationrelational bridgedesign principleseducational technologyvocational trainingethical AI

0 comments

The pith

AI-enabled pedagogical systems work best when they bridge students toward human educators rather than acting as substitutes.

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

This paper develops a framework for using AI in education that keeps human responsibility at the center. It introduces the idea of a relational bridge made up of small mediation steps that help students move from AI support to real interactions with teachers and peers. The authors list specific design principles such as making the AI's limits clear, guiding users to human contact, using questioning to encourage thinking, avoiding over-reliance, and keeping data use minimal. These principles are tied to indicators that educators can observe in practice. Pilot tests with a prototype called Amico in Italy and China showed that this approach feels useful in vocational training, suggesting it can be applied more widely.

Core claim

The central claim is that AI can serve as effective pedagogical accompaniment when designed according to a human-in-command model. In this model, the AI functions as a relational bridge consisting of micro-mediations that lower barriers to educational relationships and promote transitions to human teachers, peers, and practice communities. Key design principles include transparency about the system's identity and capabilities, scaffolding that leads toward human contact, maieutic questioning to stimulate reflection, measures to prevent dependency, and minimization of data collection. These are mapped to observable indicators for use in real settings, and early evidence from the Amico dual-m

What carries the argument

The relational bridge as a sequence of micro-mediations that lower the threshold of access to educational relationships and facilitate transitions toward meaningful human interaction.

Load-bearing premise

The assumption that pilot observations from the Amico prototype in Italy and China can be generalized to validate the design principles and indicators in wider educational settings and with different groups of students.

What would settle it

A controlled study where one group uses the AI system and another does not, showing no difference in the amount of human teacher interactions or an increase in dependency on the AI, would challenge the claim that the relational bridge effectively scaffolds toward human contact.

read the original abstract

AI-enabled systems are increasingly introduced into educational contexts, yet their effectiveness depends less on technological sophistication than on the quality of pedagogical mediation, ethical constraints, and context-sensitive design. This paper proposes a replicable framework for AI-enabled pedagogical accompaniment, grounded in a human-in-command approach in which adult responsibility remains central and AI functions as an enabling, non-substitutive infrastructure. Building on the Amico project, we operationalize the concept of a relational bridge as a sequence of micro-mediations that lower the threshold of access to educational relationships and facilitate transitions toward meaningful human interaction with teachers, peers, and communities of practice. The contribution synthesizes a set of design principles, including transparency of system identity and limits, scaffolding toward human contact, maieutic questioning, prevention of dependency dynamics, and data minimization, and maps them to observable indicators suitable for real educational settings. The paper also outlines an initial cross-context exploration of the prototype in Italy and China and discusses how the two interaction modes, AmicoMio (structured, task-oriented) and AmicoTuo (reflective, supportive), can be used as complementary pedagogical mediations. Pilot observations and participant feedback suggested feasibility and perceived usefulness in vocational contexts, motivating the present framework, informing the subsequent doctoral research program, and supporting the proposed collaborative research agenda.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note private letter to a colleague

The paper gives a usable checklist of design principles for keeping AI as a bridge to human teaching rather than a replacement, but the pilot evidence is too light on details to back the general claims.

read the letter

This paper puts forward a practical framework for AI tools meant to support teaching without taking over. It centers a human-in-command model and turns that into five principles—transparency, scaffolding to human contact, maieutic questioning, avoiding dependency, and data minimization—then maps each to things an observer could actually notice in a classroom or workshop. The Amico prototype supplies the two modes, AmicoMio for structured tasks and AmicoTuo for more open reflection, tested in vocational settings in Italy and China. That cross-context piece and the focus on lowering the barrier to real human relationships are the clearest additions to the usual human-centered AI education literature. The relational-bridge idea is a reasonable way to organize the micro-steps between AI prompt and teacher or peer interaction. Credit is due for trying to make the principles observable rather than leaving them at the level of good intentions. The pilots are said to have shown feasibility and usefulness, which at least gives the framework a starting point from real use. The main weakness is that the supporting observations stay at the level of summary impressions. No numbers on participants, no description of how feedback was collected or coded, and no comparison to other approaches appear in the text. That leaves the mapping from experience to principles more suggestive than tested, so claims about replicability across broader contexts rest on limited ground. The work is aimed at designers and researchers who build or evaluate AI companions for vocational or adult education and want a short list of guardrails to check. It could also interest people running small pilots who need language for what to watch. It deserves peer review because the principles are concrete enough to discuss and improve, and referees can ask for the missing method details or tighter ties to prior studies without starting from zero.

Referee Report

2 major / 2 minor

Summary. The paper proposes a replicable framework for AI-enabled pedagogical accompaniment grounded in a human-in-command approach, in which AI functions as enabling infrastructure rather than a substitute for human educators. It operationalizes the concept of a 'relational bridge' as a sequence of micro-mediations that facilitate access to educational relationships and transitions toward human contact with teachers, peers, and communities. The framework synthesizes five design principles (transparency of system identity and limits, scaffolding toward human contact, maieutic questioning, prevention of dependency dynamics, and data minimization) and maps them to observable indicators. These are motivated and illustrated by pilot observations from the Amico dual-mode prototype (AmicoMio: structured/task-oriented; AmicoTuo: reflective/supportive) deployed in vocational contexts in Italy and China, where the observations suggested feasibility and perceived usefulness.

Significance. If the design principles and their mapping to observable indicators can be validated through more rigorous empirical work, the framework would offer a concrete, ethically oriented contribution to HCI and educational technology by foregrounding relational mediation and human oversight. The dual-mode prototype and cross-context exploration provide a practical starting point for implementation studies. The work's emphasis on context-sensitive, non-substitutive AI use aligns with ongoing debates in the field and could inform both design guidelines and future collaborative research agendas.

major comments (2)

[Abstract and pilot observations section] Abstract and pilot observations section: The central claim that the framework is 'grounded in' and 'motivated by' the Amico prototype rests on the statement that 'pilot observations and participant feedback suggested feasibility and perceived usefulness,' yet the manuscript supplies no participant counts, recruitment details, data collection protocols, analysis methods, exclusion criteria, or specific qualitative/quantitative findings. Without these elements, the derivation of the five design principles and their observable indicators from the pilots cannot be evaluated for reliability or replicability.
[Framework presentation (relational bridge and design principles)] Framework presentation (relational bridge and design principles): The mapping from the relational-bridge concept to the five principles and indicators is presented as synthesized from prototype experience, but the manuscript does not describe the process by which participant feedback was translated into the specific principles (e.g., how 'maieutic questioning' or 'data minimization' were identified or refined). This leaves the framework's internal logic and falsifiability under-specified for a replicable contribution.

minor comments (2)

[Prototype description] The terms AmicoMio and AmicoTuo are introduced without explanation of their linguistic or cultural origins in the Italian and Chinese contexts, which would aid reader understanding of the dual-mode design.
[Observable indicators] The manuscript would benefit from explicit discussion of how the proposed observable indicators could be measured in practice (e.g., inter-rater reliability, coding schemes) to support the claim that they are 'suitable for real educational settings.'

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their thoughtful and constructive feedback on our manuscript. We appreciate the opportunity to clarify the nature of our pilot observations and the development process of the proposed framework. Below, we address each major comment in detail.

read point-by-point responses

Referee: [Abstract and pilot observations section] Abstract and pilot observations section: The central claim that the framework is 'grounded in' and 'motivated by' the Amico prototype rests on the statement that 'pilot observations and participant feedback suggested feasibility and perceived usefulness,' yet the manuscript supplies no participant counts, recruitment details, data collection protocols, analysis methods, exclusion criteria, or specific qualitative/quantitative findings. Without these elements, the derivation of the five design principles and their observable indicators from the pilots cannot be evaluated for reliability or replicability.

Authors: We agree that the current manuscript lacks sufficient detail on the pilot observations to allow full evaluation of how they informed the framework. The Amico prototype was tested in small-scale, informal pilots in vocational education settings in Italy and China, primarily to assess initial feasibility and gather anecdotal feedback rather than to conduct a formal empirical study. Specific participant numbers were not systematically recorded as the focus was on prototype iteration. To strengthen the manuscript, we will revise the relevant sections to explicitly describe the exploratory nature of these pilots, provide available contextual details (such as approximate number of sessions and participant roles), and clarify that the design principles represent a synthesis of design experience, pedagogical theory, and initial observations rather than a direct empirical derivation. This will better position the work as a conceptual contribution with illustrative examples. revision: partial
Referee: [Framework presentation (relational bridge and design principles)] Framework presentation (relational bridge and design principles): The mapping from the relational-bridge concept to the five principles and indicators is presented as synthesized from prototype experience, but the manuscript does not describe the process by which participant feedback was translated into the specific principles (e.g., how 'maieutic questioning' or 'data minimization' were identified or refined). This leaves the framework's internal logic and falsifiability under-specified for a replicable contribution.

Authors: The referee correctly identifies that the manuscript does not detail the synthesis process. The principles emerged from iterative design reflections by the team, informed by observations of how the dual-mode prototype facilitated or hindered relational aspects in practice, combined with relevant literature on dialogic pedagogy and ethical AI design. For instance, maieutic questioning was highlighted based on instances where the AI's questioning style encouraged deeper student reflection leading to subsequent human discussions. We will add a new subsection or paragraph in the framework section that outlines this development process, including examples of how specific principles were refined from prototype interactions. This will enhance the replicability and transparency of the contribution. revision: yes

Circularity Check

0 steps flagged

Framework derived from pilot observations without circular derivations or self-referential reductions

full rationale

The paper synthesizes design principles (transparency, scaffolding toward human contact, maieutic questioning, dependency prevention, data minimization) and maps them to observable indicators from pilot observations and participant feedback in the Amico dual-mode prototype deployments. No equations, fitted parameters, or mathematical derivations are present. The central framework is motivated by and grounded in external pilot data from Italy and China rather than being defined in terms of itself or reducing to self-citations by construction. The derivation chain relies on inductive synthesis from empirical impressions, which function as independent inputs, rendering the presentation self-contained with no load-bearing circular steps.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 3 invented entities

The central claim depends on domain assumptions about the superiority of human-in-command models and the value of relational bridges, with new entities introduced to describe the prototype without external falsifiable evidence beyond initial pilot feedback.

axioms (2)

domain assumption Adult responsibility remains central and AI functions as an enabling, non-substitutive infrastructure
Explicitly stated as the grounding for the human-in-command approach in the abstract.
domain assumption Micro-mediations can lower the threshold of access to educational relationships and facilitate transitions to human interaction
Core operationalization of the relational bridge concept presented as foundational to the framework.

invented entities (3)

relational bridge no independent evidence
purpose: Sequence of micro-mediations that lower access thresholds to educational relationships and facilitate human interactions
Introduced as the key operational concept for the Amico system.
AmicoMio no independent evidence
purpose: Structured, task-oriented interaction mode
One of the two complementary modes in the dual-mode prototype.
AmicoTuo no independent evidence
purpose: Reflective, supportive interaction mode
Complementary mode to AmicoMio for different pedagogical needs.

pith-pipeline@v0.9.0 · 5774 in / 1667 out tokens · 48693 ms · 2026-05-21T04:09:07.748900+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

25 extracted references · 25 canonical work pages

[1]

Systematic review of research on artificial intelligence applications in higher education —where are the educators?

O. Zawacki -Richter, V. I. Marín, M. Bond, and F. Gouverneur, “Systematic review of research on artificial intelligence applications in higher education —where are the educators?” International Journal of Educational Technology in Higher Education, vol. 16, Art. no. 39, 2019, doi: 10.1186/s41239-019-0171-0

work page doi:10.1186/s41239-019-0171-0 2019
[2]

Llm agents for education: Advances and applications,

Z. Chu, S. Wang, J. Xie, T. Zhu, Y. Yan, J. Ye, A. Zhong, X. Hu, J. Liang, P. S. Yu, and Q. Wen, “LLM agents for education: advances and applications,” arXiv preprint, arXiv:2503.11733v2, Feb. 2026. doi: 10.48550/arXiv.2503.11733

work page doi:10.48550/arxiv.2503.11733 2026
[3]

Guidance for generative AI in education and research,

UNESCO, “Guidance for generative AI in education and research,” 2023. [Online]. Available: https://unesdoc.unesco.org/ark:/48223/pf0000386693

work page 2023
[4]

ChatGPT for Good? On Opportunities and Chal- lenges of Large Language Models for Education

E. Kasneci, K. Sessler, S. Küchemann, M. Bannert, D. Dementieva, F. Fischer, and G. Kasneci, “ChatGPT for good? On opportunities and challenges of large language models for education,” Learning and Individual Differences, vol. 103, p. 102274, 2023, doi: 10.1016/j.lindif.2023.102274

work page doi:10.1016/j.lindif.2023.102274 2023
[5]

Dewey, Democracy and Education: An Introduction to the Philosophy of Education

J. Dewey, Democracy and Education: An Introduction to the Philosophy of Education. New York, NY, USA: Macmillan, 1916

work page 1916
[6]

Mortari, La pratica dell’aver cura

L. Mortari, La pratica dell’aver cura . Milano, Italy: Bruno Mondadori, 2006

work page 2006
[7]

Meirieu, Le choix d’éduquer

P. Meirieu, Le choix d’éduquer. Paris, France: ESF, 1991

work page 1991
[9]

Available: https://arxiv.org/abs/2507.02180

[Online]. Available: https://arxiv.org/abs/2507.02180

work page arXiv
[10]

Ethical guidelines on the use of artificial intelligence (AI) and data in teaching and learning for educators

European Commission, DG EAC , “Ethical guidelines on the use of artificial intelligence (AI) and data in teaching and learning for educators.” Publications Office of the European Union, 2022, doi: 10.2766/153756. [Online]. Available: https://data.europa.eu/doi/10.2766/153756

work page doi:10.2766/153756 2022
[11]

[Online]

European Parliament and Council of the European Union, Regulation (EU) 2024/1689 (Artificial Intelligence Act), Official Journal of the European Union, 2024. [Online]. Available: https://eur - lex.europa.eu/eli/reg/2024/1689/oj

work page 2024
[12]

Governing artificial intelligence in China and the European Union: comparing aims and promoting ethical outcomes,

H. Roberts, J. Cowls, E. Hine, and J. Morley, “Governing artificial intelligence in China and the European Union: comparing aims and promoting ethical outcomes,” The Information Society, vol. 39, no. 2, pp. 79–97, 2023, doi: 10.1080/01972243.2022.2124565

work page doi:10.1080/01972243.2022.2124565 2023
[13]

Large language models for education: A survey and outlook

S. Wang, T. Xu, H. Li, C. Zhang, J. Liang, J. Tang, P. Yu, and Q. Wen, “Large language models for education: a survey and outlook,” arXiv : 2403.18105, 2024. [Online]. Available: https://arxiv.org/abs/2403.18105

work page arXiv 2024
[14]

Emotion and sociable humanoid robots,

C. Breazeal, “Emotion and sociable humanoid robots,” International Journal of Human -Computer Studies, vol. 59, nos. 1 –2, pp. 119 –155, 2003

work page 2003
[15]

Socially intelligent robots: dimensions of human –robot interaction,

K. Dautenhahn, “Socially intelligent robots: dimensions of human –robot interaction,” Philosophical Transactions of the Royal Society B: Biological Sciences, vol. 362, no. 1480, pp. 679 –704, 2007, doi: 10.1098/rstb.2006.2004

work page doi:10.1098/rstb.2006.2004 2007
[16]

Freire, Pedagogy of the Oppressed

P. Freire, Pedagogy of the Oppressed. New York, NY, USA: Continuum, 1970

work page 1970
[17]

M. C. Nussbaum, Creating Capabilities: The Human Development Approach. Cambridge, MA, USA: Harvard University Press, 2011

work page 2011
[18]

Sen, Development as Freedom

A. Sen, Development as Freedom. New York, NY, USA: Alfred A. Knopf, 1999

work page 1999
[19]

The CMS experiment at the CERN LHC

A. Vedovato, El humano algoritmo. Quito, Ecuador: Abya -Yala, 2024. [Online]. Available: https://dspace.ups.edu.ec/handle/123456789/26994

work page arXiv 2024
[20]

Curriculum reform in China: challenges and reflections,

Q. Zhong, “Curriculum reform in China: challenges and reflections,” Frontiers of Education in China, vol. 1, no. 3, pp. 370 –382, 2006, doi: 10.1007/s11516-006-0014-3

work page doi:10.1007/s11516-006-0014-3 2006
[21]

New basic education and me,

L. Ye, “New basic education and me,” Frontiers of Education in China, vol. 4, no. 4, pp. 558–609, 2009, doi: 10.1007/s11516 -009-0031-0

work page doi:10.1007/s11516 2009
[22]

[Online]

European Parliament and Council of the European Union, Regulation (EU) 2016/679 (General Data Protection Regulation), Official Journal of the European Union, 2016. [Online]. Available: https://eur - lex.europa.eu/eli/reg/2016/679/oj

work page 2016
[23]

Logical equivalences, homomorphism indistinguishability, and forbidden minors

R. Deng, M. Jiang, X. Yu, Y. Lu, and S. Liu, “Does ChatGPT enhance student learning? A systematic review and meta -analysis of experimental studies,” Computers & Education, vol. 227, p. 105224, 2025, doi: 10.1016/j.compedu.2024.105224

work page doi:10.1016/j.compedu.2024.105224 2025
[24]

Personal Information Protection Law of the People’s Republic of China,

National People’s Congress of the People’s Republic of China, “Personal Information Protection Law of the People’s Republic of China,” 2021. [Online]. Available: http://en.npc.gov.cn.cdurl.cn/2021 - 12/29/c_694559.htm

work page 2021
[25]

Text generation: temperature and nucleus sampling (top_p),

OpenAI, “Text generation: temperature and nucleus sampling (top_p),” OpenAI API documentation. [Online]. Available: https://developers.openai.com/api/reference/resources/completions/meth ods/create

work page
[26]

Interim Measures for the Administration of Generative Artificial Intelligence

China Law Translate , “Interim Measures for the Administration of Generative Artificial Intelligence ” Jul. 13, 2023. [Online]. Available: https://www.chinalawtranslate.com/en/generative -ai-interim/

work page 2023

[1] [1]

Systematic review of research on artificial intelligence applications in higher education —where are the educators?

O. Zawacki -Richter, V. I. Marín, M. Bond, and F. Gouverneur, “Systematic review of research on artificial intelligence applications in higher education —where are the educators?” International Journal of Educational Technology in Higher Education, vol. 16, Art. no. 39, 2019, doi: 10.1186/s41239-019-0171-0

work page doi:10.1186/s41239-019-0171-0 2019

[2] [2]

Llm agents for education: Advances and applications,

Z. Chu, S. Wang, J. Xie, T. Zhu, Y. Yan, J. Ye, A. Zhong, X. Hu, J. Liang, P. S. Yu, and Q. Wen, “LLM agents for education: advances and applications,” arXiv preprint, arXiv:2503.11733v2, Feb. 2026. doi: 10.48550/arXiv.2503.11733

work page doi:10.48550/arxiv.2503.11733 2026

[3] [3]

Guidance for generative AI in education and research,

UNESCO, “Guidance for generative AI in education and research,” 2023. [Online]. Available: https://unesdoc.unesco.org/ark:/48223/pf0000386693

work page 2023

[4] [4]

ChatGPT for Good? On Opportunities and Chal- lenges of Large Language Models for Education

E. Kasneci, K. Sessler, S. Küchemann, M. Bannert, D. Dementieva, F. Fischer, and G. Kasneci, “ChatGPT for good? On opportunities and challenges of large language models for education,” Learning and Individual Differences, vol. 103, p. 102274, 2023, doi: 10.1016/j.lindif.2023.102274

work page doi:10.1016/j.lindif.2023.102274 2023

[5] [5]

Dewey, Democracy and Education: An Introduction to the Philosophy of Education

J. Dewey, Democracy and Education: An Introduction to the Philosophy of Education. New York, NY, USA: Macmillan, 1916

work page 1916

[6] [6]

Mortari, La pratica dell’aver cura

L. Mortari, La pratica dell’aver cura . Milano, Italy: Bruno Mondadori, 2006

work page 2006

[7] [7]

Meirieu, Le choix d’éduquer

P. Meirieu, Le choix d’éduquer. Paris, France: ESF, 1991

work page 1991

[8] [9]

Available: https://arxiv.org/abs/2507.02180

[Online]. Available: https://arxiv.org/abs/2507.02180

work page arXiv

[9] [10]

Ethical guidelines on the use of artificial intelligence (AI) and data in teaching and learning for educators

European Commission, DG EAC , “Ethical guidelines on the use of artificial intelligence (AI) and data in teaching and learning for educators.” Publications Office of the European Union, 2022, doi: 10.2766/153756. [Online]. Available: https://data.europa.eu/doi/10.2766/153756

work page doi:10.2766/153756 2022

[10] [11]

[Online]

European Parliament and Council of the European Union, Regulation (EU) 2024/1689 (Artificial Intelligence Act), Official Journal of the European Union, 2024. [Online]. Available: https://eur - lex.europa.eu/eli/reg/2024/1689/oj

work page 2024

[11] [12]

Governing artificial intelligence in China and the European Union: comparing aims and promoting ethical outcomes,

H. Roberts, J. Cowls, E. Hine, and J. Morley, “Governing artificial intelligence in China and the European Union: comparing aims and promoting ethical outcomes,” The Information Society, vol. 39, no. 2, pp. 79–97, 2023, doi: 10.1080/01972243.2022.2124565

work page doi:10.1080/01972243.2022.2124565 2023

[12] [13]

Large language models for education: A survey and outlook

S. Wang, T. Xu, H. Li, C. Zhang, J. Liang, J. Tang, P. Yu, and Q. Wen, “Large language models for education: a survey and outlook,” arXiv : 2403.18105, 2024. [Online]. Available: https://arxiv.org/abs/2403.18105

work page arXiv 2024

[13] [14]

Emotion and sociable humanoid robots,

C. Breazeal, “Emotion and sociable humanoid robots,” International Journal of Human -Computer Studies, vol. 59, nos. 1 –2, pp. 119 –155, 2003

work page 2003

[14] [15]

Socially intelligent robots: dimensions of human –robot interaction,

K. Dautenhahn, “Socially intelligent robots: dimensions of human –robot interaction,” Philosophical Transactions of the Royal Society B: Biological Sciences, vol. 362, no. 1480, pp. 679 –704, 2007, doi: 10.1098/rstb.2006.2004

work page doi:10.1098/rstb.2006.2004 2007

[15] [16]

Freire, Pedagogy of the Oppressed

P. Freire, Pedagogy of the Oppressed. New York, NY, USA: Continuum, 1970

work page 1970

[16] [17]

M. C. Nussbaum, Creating Capabilities: The Human Development Approach. Cambridge, MA, USA: Harvard University Press, 2011

work page 2011

[17] [18]

Sen, Development as Freedom

A. Sen, Development as Freedom. New York, NY, USA: Alfred A. Knopf, 1999

work page 1999

[18] [19]

The CMS experiment at the CERN LHC

A. Vedovato, El humano algoritmo. Quito, Ecuador: Abya -Yala, 2024. [Online]. Available: https://dspace.ups.edu.ec/handle/123456789/26994

work page arXiv 2024

[19] [20]

Curriculum reform in China: challenges and reflections,

Q. Zhong, “Curriculum reform in China: challenges and reflections,” Frontiers of Education in China, vol. 1, no. 3, pp. 370 –382, 2006, doi: 10.1007/s11516-006-0014-3

work page doi:10.1007/s11516-006-0014-3 2006

[20] [21]

New basic education and me,

L. Ye, “New basic education and me,” Frontiers of Education in China, vol. 4, no. 4, pp. 558–609, 2009, doi: 10.1007/s11516 -009-0031-0

work page doi:10.1007/s11516 2009

[21] [22]

[Online]

European Parliament and Council of the European Union, Regulation (EU) 2016/679 (General Data Protection Regulation), Official Journal of the European Union, 2016. [Online]. Available: https://eur - lex.europa.eu/eli/reg/2016/679/oj

work page 2016

[22] [23]

Logical equivalences, homomorphism indistinguishability, and forbidden minors

R. Deng, M. Jiang, X. Yu, Y. Lu, and S. Liu, “Does ChatGPT enhance student learning? A systematic review and meta -analysis of experimental studies,” Computers & Education, vol. 227, p. 105224, 2025, doi: 10.1016/j.compedu.2024.105224

work page doi:10.1016/j.compedu.2024.105224 2025

[23] [24]

Personal Information Protection Law of the People’s Republic of China,

National People’s Congress of the People’s Republic of China, “Personal Information Protection Law of the People’s Republic of China,” 2021. [Online]. Available: http://en.npc.gov.cn.cdurl.cn/2021 - 12/29/c_694559.htm

work page 2021

[24] [25]

Text generation: temperature and nucleus sampling (top_p),

OpenAI, “Text generation: temperature and nucleus sampling (top_p),” OpenAI API documentation. [Online]. Available: https://developers.openai.com/api/reference/resources/completions/meth ods/create

work page

[25] [26]

Interim Measures for the Administration of Generative Artificial Intelligence

China Law Translate , “Interim Measures for the Administration of Generative Artificial Intelligence ” Jul. 13, 2023. [Online]. Available: https://www.chinalawtranslate.com/en/generative -ai-interim/

work page 2023