生成式人工智能提供的解释,如何转化为大学生自身的知识结构?本文属于概念性理论研究,以知识整合为主线,结合自我解释与生成性学习理论,区分外部表征、交互活动和个体理解。将尚未经学习者有依据接纳的概念、证据与适用条件联结界定为候选关系,提出原有理解显化、候选关系审查、自我解释重构和变式应用校验相互反馈的框架,并推导三个条件性命题。论证聚焦关系依据的判断、错误修订与近迁移校验,结果包括有依据的采纳、修订和审慎保留。本文从个体微观认知加工层面补充工具使用与交互记录研究,提出观测和教学评价方向。框架、命题及观测方向尚未经实证检验,全部课程例子均为假设情境。
生成式人工智能;大学生;知识整合;自我解释;人机协同学习
This conceptual study asks how university students come to understand the explanations they obtain from generative AI. Knowledge integration provides its main perspective, supplemented by research on self-explanation and generative learning. The analysis separates external representations and interaction activities from individual understanding. Candidate relations are defined as links among concepts, evidence, and conditions of applicability that learners have not yet accepted on justified grounds. To examine how these relations enter learners’ knowledge structures, the paper connects four processes: articulating prior understanding, examining candidate relations, reconstructing self-explanations, and testing understanding through near-transfer tasks. Learners may return to earlier judgments during this process; no fixed sequence is prescribed. The analysis addresses the grounds for acceptance, the correction of errors, and the use of near-transfer tasks to check understanding. Possible outcomes include justified acceptance, revision, and suspension of judgment. Three conditional propositions inform suggestions for future observation, instructional design, and assessment. By focusing on individual cognitive processing, the framework complements research on tool use and interaction records. The framework, propositions, and proposed observational measures have yet to be tested empirically. All classroom examples are hypothetical.
generative artificial intelligence; university students; knowledge integration; self-explanation; human-AI collaborative learning
王晨. 生成式人工智能情境下大学生知识整合机制[J]. 人文与社会科学学刊. 2026, 2 (11): 112-118. DOI: 10.70693/202610073465.
王晨. (2026). 生成式人工智能情境下大学生知识整合机制. 人文与社会科学学刊, 2 (11), 112-118. https://doi.org/10.70693/202610073465