The reviewed record of science sign in
Pith

arxiv: 2606.12416 · v1 · pith:63UQ6SKJ · submitted 2026-05-07 · cs.CY

Who Designs the Designer? Behavioural Architecture for GenAI in Education

Sepinoud Azimi This is my paper

Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel 2026-06-30 23:36 UTCgrok-4.3pith:63UQ6SKJrecord.jsonopen to challenge →

classification cs.CY
keywords behavioural architectureGenAI in educationdesigner rolestudent co-authoringlearning recordsEU governanceAI ethics
0
0 comments X

The pith

The designer role in AI-in-education is unoccupied and requires EU-level infrastructure that does not yet exist.

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

Current responses to AI in education either ban the technology or build content-only tutors, both of which overlook decades of research showing that personality, motivation, and emotional state shape learning outcomes as strongly as cognitive ability. The paper proposes behavioural architecture as an alternative in which the system adapts to how a student learns rather than only to what they learn next. The student co-authors the record the system keeps, with rights to read, revise, and revoke it. This shifts the designer role—what the system treats as true about the student—from the AI vendor alone to a distribution among educator, student, and system. Because the institution operating the system is the same one assessing the student, the paper concludes that only EU-level governance can supply the structural protections this configuration demands.

Core claim

The designer role in AI-in-education is currently unoccupied, and occupying it requires infrastructure that does not yet exist.

What carries the argument

behavioural architecture, in which the system adapts to how a student learns and the student co-authors the record the system keeps about them

Load-bearing premise

That the institution operating the AI system is the same one assessing the student, so individual institutions cannot supply the structural protections this configuration demands and EU-level governance is therefore required.

What would settle it

Empirical results from the five proposed tests showing that students using the architecture achieve no measurable gains in agency or learning outcomes compared with standard AI tutors.

Figures

Figures reproduced from arXiv: 2606.12416 by Sepinoud Azimi.

Figure 1
Figure 1. Figure 1: Two paradigms. Left: content adaptation. The system adjusts material based on behavioural traces (response time, click patterns, errors) that the student does not see. Personality, motivation, and emotion are not addressed. Right: behavioural architecture. The system adapts to personality, motivation, and emotion, surfaced through conversation rather than inferred from traces. The student co-authors the re… view at source ↗
read the original abstract

AI in education is stuck between two failed responses: banning AI and building content-only tutors. Both fail because they ignore what decades of research has established: that personality, motivation, and emotional state shape learning outcomes as strongly as cognitive ability. This paper proposes behavioural architecture as an alternative. In the proposed architecture, the system adapts to how a student learns, not only to what they learn next. The student co-authors the record the system keeps, can read it, revise it, and revoke it. The designer role, what the system treats as true about the student, shifts from the AI vendor alone to a distribution among educator, student, and system. The paper argues that this architecture requires governance at EU level: the institution operating the system is the same one assessing the student, and individual institutions cannot provide the structural protections this configuration demands. Five empirical questions are proposed to test whether the architecture delivers on its claims. The contribution is naming a vacancy: the designer role in AI-in-education is currently unoccupied, and occupying it requires infrastructure that does not yet exist.

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.

Referee Report

3 major / 2 minor

Summary. The paper claims that current responses to AI in education—banning it or building content-only tutors—fail to account for the role of personality, motivation, and emotional state in learning, as established by decades of research. It proposes a 'behavioural architecture' in which GenAI systems adapt to how a student learns rather than only what they learn next, with the student co-authoring, reading, revising, and revoking the system's record. This shifts the 'designer role' (what the system treats as true about the student) from the AI vendor to a distribution among educator, student, and system. The paper argues that realizing this architecture requires governance at the EU level because the institution operating the AI is the same as the one assessing the student, and individual institutions cannot supply the necessary structural protections. It concludes by proposing five empirical questions to test the architecture and positions its contribution as naming the unoccupied designer role and the need for non-existent infrastructure.

Significance. If the proposed behavioural architecture can be developed and empirically validated, it would offer a meaningful integration of affective learning research into GenAI systems for education, potentially improving outcomes while enhancing student agency over their data. The framing of the 'designer role' as currently vacant provides a clear conceptual contribution that could guide future system design and policy. However, the manuscript supplies no data, derivations, or implementation details, so its significance is that of a well-motivated proposal rather than a completed advance.

major comments (3)
  1. [Governance argument] The claim that 'the institution operating the system is the same one assessing the student, and individual institutions cannot provide the structural protections this configuration demands' (abstract) is load-bearing for the EU-level governance recommendation but is not supported by any analysis, case examples, or references to existing governance frameworks such as FERPA, GDPR implementations in education, or institutional review board practices.
  2. [Behavioural architecture proposal] The architecture is described conceptually but without any formal specification, pseudocode, or example of how adaptation to 'how a student learns' would be implemented or how the distributed designer role would function in a concrete system.
  3. [Empirical questions] The five empirical questions are proposed to test the claims, but the manuscript provides no details on study design, variables, or success criteria, so they do not yet constitute a testable program within the paper.
minor comments (2)
  1. The abstract and text introduce 'behavioural architecture' and 'designer role' as new terms without explicit definitions or differentiation from related concepts in the literature on AI ethics or educational technology.
  2. No references are provided in the supplied abstract for the 'decades of research' on personality, motivation, and emotional state in learning, which would strengthen the grounding of the proposal.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the detailed and constructive report. The comments correctly identify areas where the conceptual proposal can be strengthened with additional support and clarification while preserving its focus as a position paper. We address each major comment below.

read point-by-point responses

  1. Referee: [Governance argument] The claim that 'the institution operating the system is the same one assessing the student, and individual institutions cannot provide the structural protections this configuration demands' (abstract) is load-bearing for the EU-level governance recommendation but is not supported by any analysis, case examples, or references to existing governance frameworks such as FERPA, GDPR implementations in education, or institutional review board practices.

    Authors: We agree that the governance claim requires more explicit grounding. The argument derives from the structural conflict when one entity both maintains the behavioural record (including affective and motivational data) and performs high-stakes assessment, creating risks of purpose creep and reduced student agency. In revision we will add references to GDPR Recital 75 and Article 35 on data protection impact assessments in education, plus literature on IRB limitations with algorithmic profiling (e.g., Metcalf et al. on educational data governance). We will also include a brief case contrast with FERPA's focus on access rather than co-authorship. These additions will support rather than alter the EU-level recommendation. revision: yes

  2. Referee: [Behavioural architecture proposal] The architecture is described conceptually but without any formal specification, pseudocode, or example of how adaptation to 'how a student learns' would be implemented or how the distributed designer role would function in a concrete system.

    Authors: The paper's contribution is the identification of the currently vacant 'designer role' and the high-level principles needed to occupy it. A full formal specification or pseudocode belongs to a subsequent systems paper. However, we accept that a concrete illustration would improve clarity. In revision we will add a figure showing the information flow among student co-authorship, educator oversight, and system adaptation, together with a short worked example contrasting content-only adaptation with adaptation that also incorporates a student-revisable motivational state variable. revision: partial

  3. Referee: [Empirical questions] The five empirical questions are proposed to test the claims, but the manuscript provides no details on study design, variables, or success criteria, so they do not yet constitute a testable program within the paper.

    Authors: The questions are offered as an initial research agenda rather than a complete protocol. We will revise the section to supply, for two of the five questions, example dependent variables (e.g., validated agency and trust scales) and high-level design outlines (e.g., within-subject comparison of systems with and without student revision rights). This keeps the paper within its conceptual scope while making the questions more actionable for future work. revision: partial

Circularity Check

0 steps flagged

No circularity; policy argument rests on external citations and explicit premises rather than self-referential reduction

full rationale

The paper advances a behavioural-architecture proposal and an EU-governance conclusion by citing decades of learning research on personality, motivation and emotional state, then stating the institutional-conflict premise as a factual precondition. No equations, fitted parameters, self-citations, or uniqueness theorems appear; the five empirical questions are forward tests, not retrofitted validations. The derivation therefore does not reduce any claim to its own inputs by construction and remains self-contained against the cited external literature.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 2 invented entities

The paper relies on one domain assumption from learning sciences and introduces two new conceptual entities without independent evidence or falsifiable handles outside the proposal itself.

axioms (1)
  • domain assumption Personality, motivation, and emotional state shape learning outcomes as strongly as cognitive ability.
    Invoked in the opening sentence of the abstract as established by decades of research.
invented entities (2)
  • behavioural architecture no independent evidence
    purpose: An AI system that adapts to how a student learns and lets the student co-author, read, revise, and revoke the record the system keeps.
    Introduced as the alternative to banning or content-only approaches.
  • designer role no independent evidence
    purpose: The entity or process that determines what the system treats as true about the student.
    Named as currently unoccupied and requiring new infrastructure.

pith-pipeline@v0.9.1-grok · 5708 in / 1350 out tokens · 34235 ms · 2026-06-30T23:36:12.096038+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

29 extracted references · 29 canonical work pages

  1. [1]

    International Journal for Educational Integrity , volume=

    Testing of detection tools for AI-generated text , author=. International Journal for Educational Integrity , volume=. 2023 , publisher=

    work page 2023

  2. [2]

    Patterns , volume=

    GPT detectors are biased against non-native English writers , author=. Patterns , volume=. 2023 , publisher=

    work page 2023

  3. [3]

    British Journal of Educational Technology , volume=

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

    work page 2024

  4. [4]

    International Journal of Educational Technology in Higher Education , volume=

    A systematic mapping review at the intersection of artificial intelligence and self-regulated learning , author=. International Journal of Educational Technology in Higher Education , volume=. 2025 , publisher=

    work page 2025

  5. [5]

    Ethical Principles of Psychologists and Code of Conduct , year =

    work page

  6. [6]

    and Childress, James F

    Beauchamp, Tom L. and Childress, James F. , title =

    work page

  7. [7]

    Educational researcher , volume=

    The 2 sigma problem: The search for methods of group instruction as effective as one-to-one tutoring , author=. Educational researcher , volume=. 1984 , publisher=

    work page 1984

  8. [8]

    Higher education , volume=

    Co-creation in learning and teaching: The case for a whole-class approach in higher education , author=. Higher education , volume=. 2020 , publisher=

    work page 2020

  9. [9]

    Review of educational research , volume=

    Pathways to student motivation: A meta-analysis of antecedents of autonomous and controlled motivations , author=. Review of educational research , volume=. 2022 , publisher=

    work page 2022

  10. [10]

    Learning and Instruction , volume=

    Dynamics of affective states during complex learning , author=. Learning and Instruction , volume=. 2012 , publisher=

    work page 2012

  11. [11]

    Learning and Instruction , volume=

    Confusion can be beneficial for learning , author=. Learning and Instruction , volume=. 2014 , publisher=

    work page 2014

  12. [12]

    1944 , publisher=

    Democracy and education: An introduction to the philosophy of education , author=. 1944 , publisher=

    work page 1944

  13. [13]

    Journal of medical Internet research , volume=

    Your robot therapist will see you now: ethical implications of embodied artificial intelligence in psychiatry, psychology, and psychotherapy , author=. Journal of medical Internet research , volume=. 2019 , publisher=

    work page 2019

  14. [14]

    Higher Education Policy Institute: London, UK , year=

    Student generative AI survey 2025 , author=. Higher Education Policy Institute: London, UK , year=

    work page 2025

  15. [15]

    Frontiers in Education , volume=

    How does co-regulation with Adaptive Learning Technologies affect primary school students' goal-setting, regulation of practice behavior and learning outcomes? , author=. Frontiers in Education , volume=. 2024 , organization=

    work page 2024

  16. [16]

    Perspectives on Psychological Science , volume=

    Student motivation and associated outcomes: A meta-analysis from self-determination theory , author=. Perspectives on Psychological Science , volume=. 2021 , publisher=

    work page 2021

  17. [17]

    Psychology Research and Behavior Management , pages=

    AI technology panic—is AI dependence bad for mental health? A cross-lagged panel model and the mediating roles of motivations for AI use among adolescents , author=. Psychology Research and Behavior Management , pages=. 2024 , publisher=

    work page 2024

  18. [18]

    Personality and individual differences , volume=

    The Big Five personality traits, learning styles, and academic achievement , author=. Personality and individual differences , volume=. 2011 , publisher=

    work page 2011

  19. [19]

    College & research libraries , volume=

    Reframing information literacy as a metaliteracy , author=. College & research libraries , volume=

    work page

  20. [20]

    Journal of personality , volume=

    Big Five personality traits and academic performance: A meta-analysis , author=. Journal of personality , volume=. 2022 , publisher=

    work page 2022

  21. [21]

    Computers in Human Behavior Reports , volume=

    Design and evaluation of a GenAI-based personalized educational content system tailored to personality traits and emotional responses for adaptive learning , author=. Computers in Human Behavior Reports , volume=. 2025 , publisher=

    work page 2025

  22. [22]

    Computers and Education: Artificial Intelligence , volume=

    From procrastination to engagement? An experimental exploration of the effects of an adaptive virtual assistant on self-regulation in online learning , author=. Computers and Education: Artificial Intelligence , volume=. 2023 , publisher=

    work page 2023

  23. [23]

    , author=

    A meta-analysis of the five-factor model of personality and academic performance. , author=. Psychological bulletin , volume=. 2009 , publisher=

    work page 2009

  24. [24]

    JAMA Network Open , volume=

    Use of generative AI for mental health advice among US adolescents and young adults , author=. JAMA Network Open , volume=. 2025 , publisher=

    work page 2025

  25. [25]

    2023 , publisher=

    The Oxford handbook of self-determination theory , author=. 2023 , publisher=

    work page 2023

  26. [26]

    Contemporary educational psychology , volume=

    Intrinsic and extrinsic motivation from a self-determination theory perspective: Definitions, theory, practices, and future directions , author=. Contemporary educational psychology , volume=. 2020 , publisher=

    work page 2020

  27. [27]

    , author=

    Expressing stigma and inappropriate responses prevents LLMs from safely replacing mental health providers. , author=. Proceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency , pages=

    work page 2025

  28. [28]

    JMIR Mental Health , volume=

    Regulating AI in mental health: ethics of care perspective , author=. JMIR Mental Health , volume=. 2024 , publisher=

    work page 2024

  29. [29]

    1978 , publisher=

    Mind in society: Development of higher psychological processes , author=. 1978 , publisher=

    work page 1978