pith. sign in

arxiv: 2604.07813 · v2 · pith:GPB6WFASnew · submitted 2026-04-09 · 💻 cs.AI · cs.HC

Agentivism: a learning theory for the age of artificial intelligence

Pith reviewed 2026-05-25 06:42 UTC · model grok-4.3

classification 💻 cs.AI cs.HC
keywords agentivismlearning theoryhuman-AI interactionAI-assisted learningdelegationepistemic monitoringinternalizationtransfer
0
0 comments X

The pith

Agentivism defines learning as durable human capability growth through selective AI delegation, epistemic monitoring, reconstructive internalization, and transfer under reduced support.

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

The paper proposes Agentivism as a learning theory tailored to human-AI interaction, where generative AI allows learners to delegate cognitive work such as explanation and problem solving. Existing theories like behaviourism, cognitivism, constructivism, and connectivism do not directly address when AI-assisted performance produces lasting human capability rather than temporary task completion. Agentivism identifies four processes that together turn delegation into durable growth: selective delegation, epistemic monitoring and verification, reconstructive internalization of outputs, and demonstrated transfer when support is reduced. This framework matters because AI systems are becoming a persistent feature of learning, so performance can no longer serve as a reliable proxy for actual learning.

Core claim

Existing learning theories remain important but do not directly explain when AI-assisted performance becomes durable human capability. We propose Agentivism, a learning theory for human-AI interaction. Agentivism defines learning as durable growth in human capability through selective delegation to AI, epistemic monitoring and verification of AI contributions, reconstructive internalization of AI-assisted outputs, and transfer under reduced support. The importance of Agentivism lies in explaining how learning remains possible when intelligent delegation is easy and human-AI interaction is becoming a persistent and expanding part of human learning.

What carries the argument

The four-component mechanism of Agentivism: selective delegation to AI combined with epistemic monitoring and verification, reconstructive internalization of AI-assisted outputs, and transfer under reduced support.

If this is right

  • Successful task completion with AI no longer indicates that learning has occurred.
  • Selective delegation requires learners to decide which cognitive work to offload rather than delegating everything.
  • Epistemic monitoring and verification of AI outputs must occur for contributions to support durable capability.
  • Reconstructive internalization converts AI-generated material into personally usable knowledge and judgment.
  • Transfer under reduced support serves as the test that durable capability has been achieved.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • AI tutoring systems could incorporate explicit prompts for verification and internalization steps to align with the theory.
  • The same delegation-monitoring-internalization-transfer sequence may apply when learners use other assistive tools such as search engines or calculators.
  • Curriculum design could shift from measuring final outputs to measuring growth in capability after AI support is withdrawn.

Load-bearing premise

Existing learning theories remain important but do not directly explain when AI-assisted performance becomes durable human capability.

What would settle it

An experiment in which learners complete tasks with AI support, engage in monitoring and internalization, yet show no increase in independent capability when AI support is later withdrawn would falsify the central claim.

read the original abstract

Learning theories have historically changed when the conditions of learning evolved. Generative and agentic AI create a new condition by allowing learners to delegate explanation, writing, problem solving, and other cognitive work to systems that can generate, recommend, and sometimes act on the learner's behalf. This creates a fundamental challenge for learning theory: successful performance can no longer be assumed to indicate learning. Learners may complete tasks effectively with AI support while developing less understanding, weaker judgment, and limited transferable capability. We argue that this problem is not fully captured by existing learning theories. Behaviourism, cognitivism, constructivism, and connectivism remain important, but they do not directly explain when AI-assisted performance becomes durable human capability. We propose Agentivism, a learning theory for human-AI interaction. Agentivism defines learning as durable growth in human capability through selective delegation to AI, epistemic monitoring and verification of AI contributions, reconstructive internalization of AI-assisted outputs, and transfer under reduced support. The importance of Agentivism lies in explaining how learning remains possible when intelligent delegation is easy and human-AI interaction is becoming a persistent and expanding part of human learning.

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

2 major / 2 minor

Summary. The manuscript argues that existing learning theories (behaviourism, cognitivism, constructivism, connectivism) do not directly explain when AI-assisted performance produces durable human capability. It proposes Agentivism as a new theory for human-AI interaction, defining learning as durable growth in capability through four processes: selective delegation to AI, epistemic monitoring and verification of AI contributions, reconstructive internalization of AI-assisted outputs, and transfer under reduced support.

Significance. If future work operationalizes and tests the four processes, the framework could usefully guide design of AI tools that promote transferable capability rather than mere task completion. The manuscript correctly identifies a timely distinction between supported performance and independent learning but supplies no comparative analysis, predictions, or data to establish that the proposed mechanisms add explanatory power beyond existing theories.

major comments (2)
  1. [Introduction / motivation for Agentivism] The central motivation (that the four listed theories 'do not directly explain when AI-assisted performance becomes durable human capability') is asserted without a section-by-section comparison showing specific explanatory gaps; this justification is load-bearing for introducing a new named theory.
  2. [Definition of Agentivism] The definition of learning under Agentivism is given as the conjunction of the four processes, yet no criteria are supplied for when each process has occurred or how their joint presence produces measurable 'durable growth'; this renders the central claim difficult to falsify or apply.
minor comments (2)
  1. [Abstract] The abstract and opening paragraphs use 'durable growth in human capability' repeatedly without an initial operational gloss or pointer to related constructs in the learning-sciences literature.
  2. [Related work / motivation] No references are supplied for prior work on 'agentic' AI or human-AI delegation in education, which would help situate the novelty claim.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive comments. We respond point by point to the major comments below and indicate planned revisions.

read point-by-point responses
  1. Referee: [Introduction / motivation for Agentivism] The central motivation (that the four listed theories 'do not directly explain when AI-assisted performance becomes durable human capability') is asserted without a section-by-section comparison showing specific explanatory gaps; this justification is load-bearing for introducing a new named theory.

    Authors: We agree that the motivation would be strengthened by explicit comparison. In revision we will add a new subsection that systematically contrasts how behaviourism, cognitivism, constructivism, and connectivism each treat the distinction between AI-supported task performance and the acquisition of durable, transferable human capability, thereby identifying the precise gaps Agentivism addresses. revision: yes

  2. Referee: [Definition of Agentivism] The definition of learning under Agentivism is given as the conjunction of the four processes, yet no criteria are supplied for when each process has occurred or how their joint presence produces measurable 'durable growth'; this renders the central claim difficult to falsify or apply.

    Authors: We accept that the present definition remains at a conceptual level and would benefit from greater specificity. We will revise the relevant section to supply preliminary, literature-grounded indicators for each process (for example, observable verification behaviours for epistemic monitoring) together with a discussion of how their conjunction could be linked to measurable durable growth. We will also state explicitly that full operationalisation and empirical validation are tasks for subsequent work, which will enable falsification. revision: yes

Circularity Check

0 steps flagged

No significant circularity identified

full rationale

The paper advances a purely conceptual proposal that defines Agentivism via four enumerated processes without any equations, fitted parameters, predictions, or formal derivations. The claim that prior theories are insufficient is asserted directly rather than derived from self-referential steps, and no load-bearing self-citation or ansatz is invoked to close a loop. The framework is offered as a guide for future operationalization, remaining self-contained against external benchmarks with no reduction of outputs to inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 1 invented entities

The paper rests on domain assumptions about the insufficiency of prior theories and introduces the new theory as an invented framework without quantitative elements or external validation.

axioms (2)
  • domain assumption Successful performance can no longer be assumed to indicate learning when using AI systems.
    Presented as the fundamental challenge created by generative and agentic AI.
  • domain assumption Existing learning theories do not directly explain durable human capability under AI assistance.
    Used to justify the need for a new theory.
invented entities (1)
  • Agentivism no independent evidence
    purpose: New learning theory defining mechanisms for durable capability in human-AI settings.
    Introduced as the proposed solution without prior existence or independent evidence.

pith-pipeline@v0.9.0 · 5730 in / 1334 out tokens · 27550 ms · 2026-05-25T06:42:06.806568+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

85 extracted references · 85 canonical work pages · 1 internal anchor

  1. [1]

    Psychology as the Behaviorist Views It

    Watson JB. Psychology as the Behaviorist Views It. Psychological Review. 1913;20(2):158–177. https://doi.org/10.1037/h0074428

  2. [2]

    The Behavior of Organisms: An Experimental Analysis

    Skinner BF. The Behavior of Organisms: An Experimental Analysis. New York: Appleton-Century; 1938

  3. [3]

    Liu, T., Huynh, N., and van der Schaar, M

    Miller GA. The Magical Number Seven, Plus or Minus Two: Some Limits on Our Capacity for Processing Information. Psychological Review. 1956;63(2):81–97. https://doi.org/10.1037/h0043158

  4. [4]

    Human Memory: A Proposed System and Its Control Processes

    Atkinson RC, Shiffrin RM. Human Memory: A Proposed System and Its Control Processes. In: Spence KW, Spence JT, editors. The Psychology of Learning and Motivation. vol. 2. Academic Press; 1968. p. 89–195

  5. [5]

    Experience and Education

    Dewey J. Experience and Education. New York: Macmillan; 1938

  6. [6]

    The Origins of Intelligence in Children

    Piaget J. The Origins of Intelligence in Children. New York: International Universities Press; 1952

  7. [7]

    Mind in Society: The Development of Higher Psychological Processes

    Vygotsky LS. Mind in Society: The Development of Higher Psychological Processes. Cambridge, MA: Harvard University Press; 1978

  8. [8]

    Connectivism: A Learning Theory for the Digital Age

    Siemens G. Connectivism: A Learning Theory for the Digital Age. International Journal of Instructional Technology and Distance Learning. 2005;2(1):3–10

  9. [9]

    Large Language Models Challenge the Future of Higher Education

    Milano S, McGrane JA, Leonelli S. Large Language Models Challenge the Future of Higher Education. Nature Machine Intelligence. 2023;5(4):333–334. https: //doi.org/10.1038/s42256-023-00644-2

  10. [10]

    Large AI Models Are Cultural and Social Technologies

    Farrell H, Gopnik A, Shalizi C, Evans J. Large AI Models Are Cultural and Social Technologies. Science. 2025;387(6739):1153–1156. https://doi.org/10. 1126/science.adt9819. 17

  11. [11]

    Build- ing Machines That Learn and Think with People

    Collins KM, Sucholutsky I, Bhatt U, Chandra K, Wong L, Lee M, et al. Build- ing Machines That Learn and Think with People. Nature Human Behaviour. 2024;8(10):1851–1863. https://doi.org/10.1038/s41562-024-01991-9

  12. [12]

    ChatGPT Has Entered the Classroom: How LLMs Could Trans- form Education

    Extance A. ChatGPT Has Entered the Classroom: How LLMs Could Trans- form Education. Nature. 2023;623(7987):474–477. https://doi.org/10.1038/ d41586-023-03507-3

  13. [13]

    Experimental Evidence on the Productivity Effects of Genera- tive Artificial Intelligence

    Noy S, Zhang W. Experimental Evidence on the Productivity Effects of Genera- tive Artificial Intelligence. Science. 2023;381(6654):187–192. https://doi.org/10. 1126/science.adh2586

  14. [14]

    Empowering student self-regulated learning and science education through ChatGPT: A pioneering pilot study

    Ng DTK, Tan CW, Leung JKL. Empowering student self-regulated learning and science education through ChatGPT: A pioneering pilot study. British Journal of Educational Technology. 2024;55(4):1328–1353. https://doi.org/10.1111/bjet. 13454

  15. [15]

    Evaluation of ChatGPT’s Real-Life Implementation in Under- graduate Dental Education: Mixed Methods Study

    Kavadella A, Dias da Silva MA, Kaklamanos EG, Stamatopoulos V, Gian- nakopoulos K. Evaluation of ChatGPT’s Real-Life Implementation in Under- graduate Dental Education: Mixed Methods Study. JMIR Medical Education. 2024;10:e51344. https://doi.org/10.2196/51344

  16. [16]

    Integrating ChatGPT for vocabulary learning and retention: A classroom-based study of Saudi EFL learners

    Abdelhalim SM, Alsehibany R. Integrating ChatGPT for vocabulary learning and retention: A classroom-based study of Saudi EFL learners. Language Learning & Technology. 2025;p. 1–24. https://doi.org/10.64152/10125/73635

  17. [17]

    From Chalkboards to Chatbots: Evaluating the Impact of Generative AI on Learning Outcomes in Nigeria

    De Simone M, Tiberti F, Barron Rodriguez M, Manolio F, Mosuro W, Dikoru EJ. From Chalkboards to Chatbots: Evaluating the Impact of Generative AI on Learning Outcomes in Nigeria. Washington, DC: World Bank; 2025

  18. [18]

    Beware of metacognitive lazi- ness: Effects of generative artificial intelligence on learning motivation, processes, and performance

    Fan Y, Tang L, Le H, Shen K, Tan S, Zhao Y, et al. Beware of metacognitive lazi- ness: Effects of generative artificial intelligence on learning motivation, processes, and performance. British Journal of Educational Technology. 2025;56(2):489–530. https://doi.org/10.1111/bjet.13544

  19. [20]

    Cognitive ease at a cost: LLMs reduce mental effort but compromise depth in student scientific inquiry

    Stadler M, Bannert M, Sailer M. Cognitive ease at a cost: LLMs reduce mental effort but compromise depth in student scientific inquiry. Computers in Human Behavior. 2024;160:108386. https://doi.org/10.1016/j.chb.2024.108386

  20. [21]

    AI makes you smarter but none the wiser: The disconnect between performance and metacognition

    Fernandes D, Villa S, Nicholls S, Haavisto O, Buschek D, Schmidt A, et al. AI makes you smarter but none the wiser: The disconnect between performance and metacognition. Computers in Human Behavior. 2025;p. 108779. https://doi.org/ 18 10.1016/j.chb.2025.108779

  21. [22]

    Distinguishing performance gains from learning when using generative AI

    Yan L, Greiff S, Lodge JM, Gaˇ sevi´ c D. Distinguishing performance gains from learning when using generative AI. Nature Reviews Psychology. 2025;4(7):435–

  22. [23]

    https://doi.org/10.1038/s44159-025-00467-5

  23. [24]

    Generative artificial intelligence-supported programming education: Effects on learning performance, self-efficacy and processes

    Li S, Liu J, Dong Q. Generative artificial intelligence-supported programming education: Effects on learning performance, self-efficacy and processes. Aus- tralasian Journal of Educational Technology. 2025;https://doi.org/10.14742/ajet. 9932

  24. [25]

    Machine Culture

    Brinkmann L, Baumann F, Bonnefon JF, Derex M, M¨ uller TF, Nussberger AM, et al. Machine Culture. Nature Human Behaviour. 2023;7(11):1855–1868. https: //doi.org/10.1038/s41562-023-01742-2

  25. [26]

    Cultural Tendencies in Generative AI

    Lu JG, Song LL, Zhang LD. Cultural Tendencies in Generative AI. Nature Human Behaviour. 2025;9(11):2360–2369. https://doi.org/10.1038/s41562-025-02242-1

  26. [27]

    High heels, compass, spider-man, or drug? Metaphor analysis of generative artificial intelligence in academic writing

    Jin F, Sun L, Pan Y, Lin CH. High heels, compass, spider-man, or drug? Metaphor analysis of generative artificial intelligence in academic writing. Computers & Education. 2025;228:105248. https://doi.org/10.1016/j.compedu.2025.105248

  27. [28]

    Generative AI enhances individual creativity but reduces the collective diversity of novel content

    Doshi AR, Hauser OP. Generative AI enhances individual creativity but reduces the collective diversity of novel content. Science Advances. 2024;10(28):eadn5290. https://doi.org/10.1126/sciadv.adn5290

  28. [29]

    Interaction between students and artificial intelligence in the con- text of creative potential development

    Xu M. Interaction between students and artificial intelligence in the con- text of creative potential development. Interactive Learning Environments. 2025;33(7):4460–4475. https://doi.org/10.1080/10494820.2025.2465439

  29. [30]

    Evaluating AI-assisted creative ideation: A crossover study in higher education

    Balta-Salvador R, Braso-Vives E, Pena M. Evaluating AI-assisted creative ideation: A crossover study in higher education. Thinking Skills and Creativity. 2026;59:101958. https://doi.org/10.1016/j.tsc.2025.101958

  30. [31]

    The effects of generative AI on collaborative problem-solving and team creativity performance in digital story creation: an experimental study

    Wei X, Wang L, Lee LK, Liu R. The effects of generative AI on collaborative problem-solving and team creativity performance in digital story creation: an experimental study. International Journal of Educational Technology in Higher Education. 2025;22(1). https://doi.org/10.1186/s41239-025-00526-0

  31. [32]

    When machines join the moral circle: The persona effect of generative AI agents in collaborative reasoning

    Jin Y, Martinez-Maldonado R, Shi W, Huang S, Zheng M, Han X, et al. When machines join the moral circle: The persona effect of generative AI agents in collaborative reasoning. British Journal of Educational Technology. 2026;00:1–24. https://doi.org/https://doi.org/10.1111/bjet.70067

  32. [33]

    The homogenizing effect of large lan- guage models on human expression and thought

    Sourati Z, Ziabari AS, Dehghani M. The homogenizing effect of large lan- guage models on human expression and thought. Trends in Cognitive Sciences. 2026;00:1–12. https://doi.org/https://doi.org/10.1016/j.tics.2026.01.003. 19

  33. [34]

    Artificial Intelligence and Illusions of Understanding in Scientific Research

    Messeri L, Crockett MJ. Artificial Intelligence and Illusions of Understanding in Scientific Research. Nature. 2024;627(8002):49–58. https://doi.org/10.1038/ s41586-024-07146-0

  34. [35]

    Extending Minds with Generative AI

    Clark A. Extending Minds with Generative AI. Nature Communications. 2025;16:4627. https://doi.org/10.1038/s41467-025-59906-9

  35. [36]

    Sycophantic AI decreases prosocial intentions and promotes dependence

    Cheng M, Lee C, Khadpe P, Yu S, Han D, Jurafsky D. Sycophantic AI decreases prosocial intentions and promotes dependence. Science. 2026;391(6792):eaec8352

  36. [37]

    Social theory and social structure

    Merton RK. Social theory and social structure. Simon and Schuster; 1968

  37. [39]

    AI chatbots as reading companions in self-directed out-of-class reading: A self-determination theory perspective

    Pan M, Lai C, Guo K. AI chatbots as reading companions in self-directed out-of-class reading: A self-determination theory perspective. British Journal of Educational Technology. 2025;https://doi.org/10.1111/bjet.70002

  38. [40]

    LLM-based collaborative program- ming: impact on students’ computational thinking and self-efficacy

    Yan YM, Chen CQ, Hu YB, Ye XD. LLM-based collaborative program- ming: impact on students’ computational thinking and self-efficacy. Humanities and Social Sciences Communications. 2025;12(1). https://doi.org/10.1057/ s41599-025-04471-1

  39. [41]

    Interactions with generative AI chat- bots: unveiling dialogic dynamics, students’ perceptions, and practical competen- cies in creative problem-solving

    Song Y, Huang L, Zheng L, Fan M, Liu Z. Interactions with generative AI chat- bots: unveiling dialogic dynamics, students’ perceptions, and practical competen- cies in creative problem-solving. International Journal of Educational Technology in Higher Education. 2025;22(1). https://doi.org/10.1186/s41239-025-00508-2

  40. [42]

    Fan G, Liu D, Zhang R, Pan L. The impact of AI-assisted pair programming on student motivation, programming anxiety, collaborative learning, and program- ming performance: a comparative study with traditional pair programming and individual approaches. International Journal of STEM Education. 2025;12(1). https://doi.org/10.1186/s40594-025-00537-3

  41. [43]

    To trust or to think: Cognitive forcing functions can reduce over-reliance on AI in AI-assisted decision-making

    Buˇ cinca Z, Malaya M, Gajos KZ. To trust or to think: Cognitive forcing functions can reduce over-reliance on AI in AI-assisted decision-making. Proceedings of the ACM on Human-Computer Interaction. 2021;5(CSCW1):1–21. https://doi.org/ 10.1145/3449287

  42. [44]

    Measuring Actual Learning Versus Feeling of Learning in Response to Being Actively Engaged in the Classroom

    Deslauriers L, McCarty LS, Miller K, Callaghan K, Kestin G. Measuring Actual Learning Versus Feeling of Learning in Response to Being Actively Engaged in the Classroom. Proceedings of the National Academy of Sciences of the United States of America. 2019;116(39):19251–19257. https://doi.org/10.1073/ pnas.1821936116. 20

  43. [45]

    Generative AI Without Guardrails Can Harm Learning: Evidence from High School Mathemat- ics

    Bastani H, Bastani O, Sungu A, Ge H, Kabakci O, Mariman R. Generative AI Without Guardrails Can Harm Learning: Evidence from High School Mathemat- ics. Proceedings of the National Academy of Sciences of the United States of America. 2025;122(26):e2422633122. https://doi.org/10.1073/pnas.2422633122

  44. [47]

    A conversational agent based on contin- gent teaching model to support collaborative learning activities: impacts on students’ learning performance, self-efficacy and perceptions

    Hu W, Gong R, Wu S, Li Y. A conversational agent based on contin- gent teaching model to support collaborative learning activities: impacts on students’ learning performance, self-efficacy and perceptions. Educational Tech- nology Research and Development. 2025;73(5):3341–3372. https://doi.org/10. 1007/s11423-025-10526-6

  45. [48]

    Enhancing student engagement in online collaborative writ- ing through a generative AI-based conversational agent

    Hu W, Tian J, Li Y. Enhancing student engagement in online collaborative writ- ing through a generative AI-based conversational agent. The Internet and Higher Education. 2025;65:100979. https://doi.org/10.1016/j.iheduc.2024.100979

  46. [49]

    Investigating the tripartite interaction among teachers, students, and generative AI in EFL education: A mixed-methods study

    Guan L, Lee JCK, Zhang Y, Gu MM. Investigating the tripartite interaction among teachers, students, and generative AI in EFL education: A mixed-methods study. Computers and Education: Artificial Intelligence. 2025;8:100384. https: //doi.org/10.1016/j.caeai.2025.100384

  47. [50]

    Parent-led vs

    Xiao F, Zou EW, Lin J, Li Z, Yang D. Parent-led vs. AI-guided dialogic reading: Evidence from a randomized controlled trial in children’s e-book context. British Journal of Educational Technology. 2025;56(5):1784–1813. https://doi.org/10. 1111/bjet.13615

  48. [51]

    Supporting learner agency in collaborative writing with generative AI

    Kim S, So HJ, Park K. Supporting learner agency in collaborative writing with generative AI. British Journal of Educational Technology. 2026;https://doi.org/ 10.1111/bjet.70015

  49. [52]

    Bridging Learnersourcing and AI: Exploring the Dynamics of Student-AI Collaborative Feedback Genera- tion

    Singh A, Brooks C, Wang X, Li W, Kim J, Wilson D. Bridging Learnersourcing and AI: Exploring the Dynamics of Student-AI Collaborative Feedback Genera- tion. In: Proceedings of the 14th Learning Analytics and Knowledge Conference. New York, NY, USA: ACM; 2024. p. 742–748

  50. [53]

    Unveiling interaction patterns between students and generative AI teachable agents: Focusing on students’ agency and AI agents’ authority

    Xing W, Kim T, Song Y, Li H, Li C, Kim J. Unveiling interaction patterns between students and generative AI teachable agents: Focusing on students’ agency and AI agents’ authority. British Journal of Educational Technology. 2026;https://doi.org/10.1111/bjet.70038

  51. [54]

    On the conversational per- suasiveness of GPT-4

    Salvi F, Horta Ribeiro M, Gallotti R, West R. On the conversational per- suasiveness of GPT-4. Nature Human Behaviour. 2025;9(8):1645–1653. https: //doi.org/10.1038/s41562-025-02194-6. 21

  52. [55]

    Scientific Discovery in the Age of Artificial Intelligence

    Wang H, Fu T, Du Y, Gao W, Huang K, Liu Z, et al. Scientific Discovery in the Age of Artificial Intelligence. Nature. 2023;620(7972):47–60. https://doi.org/10. 1038/s41586-023-06221-2

  53. [56]

    Teachers’ professional agency in learning with AI: A case study of a generative AI-based knowledge-building learning companion for teachers

    Tan SC, Tan YY, Teo CL, Yuan G. Teachers’ professional agency in learning with AI: A case study of a generative AI-based knowledge-building learning companion for teachers. British Journal of Educational Technology. 2026;https://doi.org/10. 1111/bjet.70013

  54. [57]

    Large Language Models Encode Clinical Knowledge

    Singhal K, Azizi S, Tu T, Mahdavi SS, Wei J, Chung HW, et al. Large Language Models Encode Clinical Knowledge. Nature. 2023;620(7972):172–180. https:// doi.org/10.1038/s41586-023-06291-2

  55. [58]

    Artificial Intelligence for Modelling Infectious Dis- ease Epidemics

    Kraemer MUG, et al. Artificial Intelligence for Modelling Infectious Dis- ease Epidemics. Nature. 2025;638(8051):623–635. https://doi.org/10.1038/ s41586-024-08564-w

  56. [59]

    Multimodal Generative AI for Medical Image Interpretation

    Rao V, et al. Multimodal Generative AI for Medical Image Interpretation. Nature. 2025;639(8056):888–896. https://doi.org/10.1038/s41586-025-08675-y

  57. [61]

    Social Cognitive Theory: An Agentic Perspective

    Bandura A. Social Cognitive Theory: An Agentic Perspective. Annual Review of Psychology. 2001;52(1):1–26. https://doi.org/10.1146/annurev.psych.52.1.1

  58. [62]

    Impact of AI Assistance on Student Agency

    Darvishi A, Khosravi H, Sadiq S, Gaˇ sevi´ c D, Siemens G. Impact of AI Assistance on Student Agency. Computers & Education. 2024;210:104967. https://doi.org/ 10.1016/j.compedu.2023.104967

  59. [64]

    GPTs Are GPTs: Labor Market Impact Potential of LLMs

    Eloundou T, Manning S, Mishkin P, Rock D. GPTs Are GPTs: Labor Market Impact Potential of LLMs. Science. 2024;384(6702):1306–1308. https://doi.org/ 10.1126/science.adj0998

  60. [65]

    Biased AI writing assistants shift users’ attitudes on societal issues

    Williams-Ceci S, Jakesch M, Bhat A, Kadoma K, Zalmanson L, Naaman M. Biased AI writing assistants shift users’ attitudes on societal issues. Science Advances. 2026;12(11):eadw5578. https://doi.org/10.1126/sciadv.adw5578

  61. [66]

    AI Is Turning Research into a Scientific Monoculture

    Traberg CS, Roozenbeek J, van der Linden S. AI Is Turning Research into a Scientific Monoculture. Communications Psychology. 2026;4:37. https://doi.org/ 22 10.1038/s44271-026-00428-5

  62. [67]

    Artificial intelligence tools expand scientists’ impact but contract science’s focus

    Hao Q, Xu F, Li Y, Evans J. Artificial intelligence tools expand scientists’ impact but contract science’s focus. Nature. 2026;649:1237–1243. https://doi.org/10. 1038/s41586-025-09922-y

  63. [68]

    The effects of three different approaches to human-AI collaboration on online collaborative learning

    Gyasi JF, Zheng L, Love SF, Boateng FO. The effects of three different approaches to human-AI collaboration on online collaborative learning. Educational Technol- ogy & Society. 2025;28(2):373–392. https://doi.org/10.30191/ETS.202504 28(2) .TP07

  64. [69]

    Combining human and artificial intelligence for enhanced AI literacy in higher education

    Tzirides AO, Zapata G, Kastania NP, Saini AK, Castro V, Ismael SA, et al. Combining human and artificial intelligence for enhanced AI literacy in higher education. Computers and Education Open. 2024;6:100184. https://doi.org/10. 1016/j.caeo.2024.100184

  65. [70]

    Generative AI: A double-edged sword for creative thinking learning — Evidence from facial expressions and fNIRS

    Song X, Zhang Y, Lu Z, Xu L, Shen H. Generative AI: A double-edged sword for creative thinking learning — Evidence from facial expressions and fNIRS. Computers & Education. 2026;247:105578. https://doi.org/10.1016/j.compedu. 2026.105578

  66. [71]

    Generative AI as a reflective scaffold in a UAV-based STEM project: A mixed-methods study on students’ higher-order thinking and cognitive transformation

    Chen SY, Chen WC, Lai CF. Generative AI as a reflective scaffold in a UAV-based STEM project: A mixed-methods study on students’ higher-order thinking and cognitive transformation. Education and Information Technologies. 2025;30(17):24787–24814. https://doi.org/10.1007/s10639-025-13758-4

  67. [72]

    Beyond the “Wow” Fac- tor: Using Generative AI for Increasing Generative Sense-Making

    Makransky G, Shiwalia BM, Herlau T, Blurton S. Beyond the “Wow” Fac- tor: Using Generative AI for Increasing Generative Sense-Making. Educational Psychology Review. 2025;37(3). https://doi.org/10.1007/s10648-025-10039-x

  68. [73]

    Towards reliable generative AI-driven scaffolding: Reducing hallucinations and enhancing quality in self-regulated learning support

    Qian K, Liu S, Li T, Rakovi´ c M, Li X, Guan R, et al. Towards reliable generative AI-driven scaffolding: Reducing hallucinations and enhancing quality in self-regulated learning support. Computers & Education. 2026;240:105448. https://doi.org/10.1016/j.compedu.2025.105448

  69. [74]

    Development and implementation of a generative AI-enhanced simulation to enhance problem- solving skills for pre-service teachers

    Lim J, Lee U, Koh J, Jeong Y, Lee Y, Byun G, et al. Development and implementation of a generative AI-enhanced simulation to enhance problem- solving skills for pre-service teachers. Computers & Education. 2025;232:105306. https://doi.org/10.1016/j.compedu.2025.105306

  70. [75]

    AI Tutoring Outperforms In-Class Active Learning: An RCT Introducing a Novel Research-Based Design in an Authentic Educational Setting

    Kestin G, Miller K, Klales A, Milbourne T, Ponti G. AI Tutoring Outperforms In-Class Active Learning: An RCT Introducing a Novel Research-Based Design in an Authentic Educational Setting. Scientific Reports. 2025;15(1). https://doi. org/10.1038/s41598-025-97652-6

  71. [76]

    Human and Artificial Intelligence Collabora- tion for Socially Shared Regulation in Learning

    J¨ arvel¨ a S, Nguyen A, Hadwin A. Human and Artificial Intelligence Collabora- tion for Socially Shared Regulation in Learning. British Journal of Educational 23 Technology. 2023;54(5):1057–1076. https://doi.org/10.1111/bjet.13325

  72. [77]

    A Qualitative Systematic Review on AI Empowered Self- Regulated Learning in Higher Education

    Lan M, Zhou X. A Qualitative Systematic Review on AI Empowered Self- Regulated Learning in Higher Education. npj Science of Learning. 2025;10(1):21. https://doi.org/10.1038/s41539-025-00319-0

  73. [78]

    Practical and Ethical Challenges of Large Language Models in Education: A Systematic Scoping Review

    Yan L, Sha L, Zhao L, Li Y, Martinez-Maldonado R, Chen G, et al. Practical and Ethical Challenges of Large Language Models in Education: A Systematic Scoping Review. British Journal of Educational Technology. 2024;55(1):90–112. https://doi.org/10.1111/bjet.13370

  74. [79]

    Agency as a system property in human—AI interaction in edu- cation

    Cukurova M. Agency as a system property in human—AI interaction in edu- cation. British Journal of Educational Technology. 2026;p. in press. https: //doi.org/10.1111/bjet.70060

  75. [80]

    Towards hybrid human-AI learning technologies

    Molenaar I. Towards hybrid human-AI learning technologies. European Journal of Education. 2022;57(4):632–645. https://doi.org/10.1111/ejed.12527

  76. [81]

    Enhancing self-directed learning and Python mastery through integration of a large language model and learning analytics dashboard

    Liu M, Wu Z, Dai H, Su Y, Malik L, Liao J, et al. Enhancing self-directed learning and Python mastery through integration of a large language model and learning analytics dashboard. British Journal of Educational Technology. 2025;https:// doi.org/10.1111/bjet.70005

  77. [82]

    Masset, R

    Pozdniakov S, Brazil J, Banihashem SK, Noroozi O, Sadiq S, Khosravi H, et al. AI assistance in peer feedback provision: Pedagogically sound, but minimally adopted. Computers & Education. 2026;248:105591. https://doi.org/10.1016/j. compedu.2026.105591

  78. [83]

    3R Principle

    Rossi S, Fraccaro V, Manzotti R. The Brain Side of Human-AI Interactions in the Long-Term: The “3R Principle”. npj Artificial Intelligence. 2026;2(1):15. https://doi.org/10.1038/s44387-025-00063-1

  79. [84]

    ChatGPT in education: An effect in search of a cause

    Weidlich J, Gaˇ sevi´ c D, Drachsler H, Kirschner P. ChatGPT in education: An effect in search of a cause. Journal of Computer Assisted Learning. 2025;41(5):e70105

  80. [85]

    The effects of generative AI agents and scaffolding on enhancing students’ compre- hension of visual learning analytics

    Yan L, Martinez-Maldonado R, Jin Y, Echeverria V, Milesi M, Fan J, et al. The effects of generative AI agents and scaffolding on enhancing students’ compre- hension of visual learning analytics. Computers & Education. 2025;234:105322. https://doi.org/10.1016/j.compedu.2025.105322

Showing first 80 references.