认知偏差调适导向的建设法规案例推理教学——基于人机协同质辨与判断校准

Construction-Law Case-Reasoning Instruction for Cognitive-Bias Mitigation—Human-AI Collaborative Deliberation and Judgment Calibration

摘要: 目的：针对建设法规课程中“记住条文但不会适用”、责任判断易受结果、立场和人工智能建议影响等问题，提出认知偏差调适导向的案例推理教学。方法：将生成式人工智能限定为平行案例生成者、反方质询者和待核验建议提供者，构建“初判–显偏–换位–质辨–核验–校准”六阶段过程。结果：形成“偏差程度–论证质量–信心校准”三维评价，给出教师操作、风险防控和模块化方案，并提出后续准实验路径。结论：该设计将人工智能的流畅性与不可靠性转化为可控教学条件，推动课程由条文记忆转向规范证据约束下的法规适用与可校准判断。

Abstract: Purpose: This study addresses a recurring problem in construction-law education: students may recall legal provisions yet fail to apply them or resist outcome-, stance-, and AI-induced bias. Methods: The study proposes six stages: initial judgment, bias exposure, perspective shift, collaborative deliberation, source verification, and judgment calibration. Generative AI is limited to generating isomorphic cases, asking counter-position questions, and providing claims that require verification. Results: The framework combines four evidence artifacts, a three-dimensional evaluation scheme, teacher prompts, contingency and risk controls, modular options, and a quasi-experimental validation plan. Conclusion: The design treats AI fluency and fallibility as controlled instructional conditions and shifts learning toward source-grounded rule application and calibrated judgment.