Risks of Cognitive Substitution and the Redesign of Learning Assessment in Generative AI-Assisted Learning
Authors
Abstract
Generative AI can improve the efficiency and quality of students’ task completion, but successful AI-assisted performance does not necessarily demonstrate independently acquired competence. Drawing on cognitive load theory, cognitive offloading, self-regulated learning, and assessment validity, this conceptual paper synthesizes findings from experimental studies, meta-analyses, and educational assessment research. It argues that the risk of cognitive substitution arises when AI performs cognitive operations central to the intended learning outcomes, learners do not meaningfully process the feedback, and assessment practices infer independent competence from AI-assisted products. Evidence of both beneficial AI tutoring and adverse effects of unrestricted use indicates that tool design, prior knowledge, and outcome measures matter. The paper distinguishes AI-assisted task performance, learners’ cognitive responsibility during the process, and independent transfer performance, and develops an assessment rationale that aligns these sources of evidence with specific learning goals. The framework links learning goals, task conditions, and assessment evidence to guide the design and examination of assessment practices in AI-mediated educational settings.
References
[1] YAN L, GREIFF S, TEUBER Z, et al. Promises and challenges of generative artificial intelligence for human learning[J]. Nature Human Behaviour, 2024, 8: 1839-1850. DOI: 10.1038/s41562-024-02004-5.
[2] HAN X, PENG H, LIU M. The impact of GenAI on learning outcomes: A systematic review and meta-analysis of experimental studies[J]. Educational Research Review, 2025, 48: 100714. DOI: 10.1016/j.edurev.2025.100714.
[3] KESTIN G, MILLER K, KLALES A, et al. AI tutoring outperforms in-class active learning: An RCT introducing a novel research-based design in an authentic educational setting[J]. Scientific Reports, 2025, 15: 17458. DOI: 10.1038/s41598-025-97652-6.
[4] BASTANI H, BASTANI O, SUNGU A, et al. Generative AI without guardrails can harm learning: Evidence from high school mathematics[J]. Proceedings of the National Academy of Sciences, 2025, 122(26): e2422633122. DOI: 10.1073/pnas.2422633122.
[5] CHEN S, CHEUNG A C K. Effect of generative artificial intelligence on university students learning outcomes: A systematic review and meta-analysis[J]. Educational Research Review, 2025, 49: 100737. DOI: 10.1016/j.edurev.2025.100737.
[6] SWELLER J. Cognitive load during problem solving: Effects on learning[J]. Cognitive Science, 1988, 12(2): 257-285. DOI: 10.1016/0364-0213(88)90023-7.
[7] RISKO E F, GILBERT S J. Cognitive offloading[J]. Trends in Cognitive Sciences, 2016, 20(9): 676-688. DOI: 10.1016/j.tics.2016.07.002.
[8] CHI M T H, WYLIE R. The ICAP framework: Linking cognitive engagement to active learning outcomes[J]. Educational Psychologist, 2014, 49(4): 219-243. DOI: 10.1080/00461520.2014.965823.
[9] PANADERO E. A review of self-regulated learning: Six models and four directions for research[J]. Frontiers in Psychology, 2017, 8: 422. DOI: 10.3389/fpsyg.2017.00422.
[10] WOOD D, BRUNER J S, ROSS G. The role of tutoring in problem solving[J]. Journal of Child Psychology and Psychiatry, 1976, 17(2): 89-100. DOI: 10.1111/j.1469-7610.1976.tb00381.x.
[11] SHUTE V J. Focus on formative feedback[J]. Review of Educational Research, 2008, 78(1): 153-189. DOI: 10.3102/0034654307313795.
[12] HATTIE J, TIMPERLEY H. The power of feedback[J]. Review of Educational Research, 2007, 77(1): 81-112. DOI: 10.3102/003465430298487.
[13] BARCAUI A. ChatGPT as a cognitive crutch: Evidence from a randomized controlled trial on knowledge retention[J]. Social Sciences & Humanities Open, 2025, 12: 102287. DOI: 10.1016/j.ssaho.2025.102287.
[14] DENG R, JIANG M, YU X, et al. Does ChatGPT enhance student learning? A systematic review and meta-analysis of experimental studies[J]. Computers & Education, 2025, 227: 105224. DOI: 10.1016/j.compedu.2024.105224.
[15] KANE M T. Validating the interpretations and uses of test scores[J]. Journal of Educational Measurement, 2013, 50(1): 1-73. DOI: 10.1111/jedm.12000.
[16] 沈书生, 祝智庭. ChatGPT类产品:内在机制及其对学习评价的影响[J]. 中国远程教育, 2023, 43(4): 8-15. DOI: 10.13541/j.cnki.chinade.20230223.001.
[17] BLACK P, WILIAM D. Assessment and classroom learning[J]. Assessment in Education: Principles, Policy & Practice, 1998, 5(1): 7-74. DOI: 10.1080/0969595980050102.
[18] BARNETT S M, CECI S J. When and where do we apply what we learn? A taxonomy for far transfer[J]. Psychological Bulletin, 2002, 128(4): 612-637. DOI: 10.1037/0033-2909.128.4.612.
[19] KOFINAS A K, TSAY C H H, PIKE D. The impact of generative AI on academic integrity of authentic assessments within a higher education context[J]. British Journal of Educational Technology, 2025, 56(6): 2522-2549. DOI: 10.1111/bjet.13585.
[20] MIAO F, HOLMES W. Guidance for Generative AI in Education and Research[M]. Paris: UNESCO, 2023. DOI: 10.54675/EWZM9535.
[21] OECD. OECD Digital Education Outlook 2026: Exploring Effective Uses of Generative AI in Education[M]. Paris: OECD Publishing, 2026. DOI: 10.1787/062a7394-en.
[22] 戴岭, 胡姣, 祝智庭. ChatGPT赋能教育数字化转型的新方略[J]. 开放教育研究, 2023, 29(4): 41-48. DOI: 10.13966/j.cnki.kfjyyj.2023.04.005.
