人工智能、认知卸载及其对教育的影响

March 2026 Prof Jason M. Lodge and Prof Leslie Loble AM This report forms part of the work programsupporting the Australian Network for Quality DigitalEducation. The Network brings together leadersfrom across education, industry, social purposeand philanthropic organisations, government andresearch, in the common purpose of ensuring that allAustralian students benefit from the best educationaltechnology (edtech), and the benefits of edtech areleveraged to tackle the persistent learning divide.Members of the Network have provided valuableengagement, input and feedback as part of thereport’s development, though the report does notrepresent a consensus or endorsed Network view. Accessibility This report is partially conformant with WCAG2.1level AA. Known limitations and alternatives arelisted below: ++Image descriptions — images and graphs donot have long image descriptions because wehave retrofitted accessibility requirementsonto an existing template and layout.++Main body text does not meet requirementof 12-point size because we have retrofittedaccessibility requirements onto an existingtemplate and layout. Alternatives areto zoom in or use a screen reader. The Network is chaired by Leslie Loble AM, whois Industry Professor at the UTS Centre for SocialJustice and Inclusion (socialjustice.uts.edu.au). Acknowledgements The authors are grateful that the Australian Networkfor Quality Digital Education is hosted by the Professor Jason M. Lodge is Professor of EducationalPsychology and Director of the Learning,Instruction, and Technology Lab in the Schoolof Education, The University of Queensland. UTS Centre for Social Justice and Inclusion,and supported by the Paul Ramsay Foundation.The authors appreciate the valuable expertiseand guidance of the Network members andothers, which has enhanced the report. How to cite this report Lodge J. M. and Loble L (2026).Artificialintelligence, cognitive offloading and implicationsfor education, University of Technology Sydney,doi:10.71741/4pyxmbnjaq.31302475. We acknowledge the Traditional Owners ofCountry throughout Australia and pay ourrespects to Elders past and present. Copyright information Paul Ramsay Foundation This report is published by the University of TechnologySydney © University of Technology Sydney 2026. 262 Liverpool St, Darlinghurst NSW 2010paulramsayfoundation.org.au With the exception of the UTS branding, contentprovided by third parties, and any materialprotected by a trademark, all material presentedin this publication is licensed under a CreativeCommons Attribution — Non Commercial,Derivative Works 4.0 (CC BY-NC-ND 4.0)licence. The full licence terms are available at:creativecommons.org/licenses/by-nc-nd/4.0. PRF is a philanthropic foundation. The latePaul Ramsay AO established the Foundationin his name in 2006 and, after his death in2014, left most of his estate to continue hisphilanthropy for generations to come. At PRF, we work for a future where people and placeshave what they need to thrive. With organisations andcommunities, we invest in, build, and influence theconditions needed to stop disadvantage in Australia. This research was funded by PRF (grant number 5040).Any opinions, findings, or conclusions expressedin this report are those of the authors and do notnecessarily reflect the views of the Foundation. Any enquiries about or comment on this publicationshould be directed toedtechnetwork@uts.edu.au. Table of Contents Unpacking the paradox: Bypassing “desirable difficulties”22Metacognitive laziness and the illusion of competence23The pedagogical solution: From cognitive atrophy to augmentation24 A new metacognitive equity gap27Augmenting the teacher to scale expertise28 Appendix: Annotated bibliography35 This report investigates a profoundnew challenge driven by AI’s power torapidly access information and providea semblance of thinking: the risk thatstudents will outsource too much ofthe cognitive work that is crucial toestablishing the knowledge, skill and‘thinking infrastructure’ that enables bothschooling success and lifelong capacityfor ongoing learning, understanding,reflection, creativity and achievement. Preface Artificial intelligence (AI) and especially generative AI are propelling a newdynamic for Australian education, simultaneously unlocking compellingopportunities and substantial challenges for teaching and learning.Teachers and students find themselves on the front line of this paradox asthey navigate in real time questions of how to best use AI for learning andknowledge gain. Today, in Australia, nearly 80 percent of studentsreport using artificial intelligence (Thomas et al.2025), and two-thirds of early secondary teachers—fourth highest usage in the OECD (OECD 2025) —merely three years since ChatGPT burst through. The AI dynamic is nuanced and complex ineducation. It can both counteract — or compound— non-technological factors that propel learninggaps and educational outcomes, from teachershortage