生成式人工智能支持教学决策的实践困境与关键进路

Practical Dilemmas and Key Approaches of Generative Artificial Intelligence in Supporting Teaching Decision-making

教学决策具备较强的道德属性,强调向善性和利他性。随着生成式人工智能被应用于教学,其强大的学习、理解和生成能力使之承担起大量非程序化决策任务。这也引发了关键问题:生成式人工智能是否具备支持教学决策的基本能力?未来的教育会受到什么样的影响?文章通过回顾社会学、心理学、教育学、机器道德领域的既往研究,发现生成式人工智能的介入推动了教学决策从经验主义向人机协作模式的转变,同时揭示出其自主性提升过程中伴随着道德敏感性的不稳定。这可能引发人机间的误会、欺骗、责任推卸等风险,并潜移默化地改变教育场景中的道德图景。为应对这些挑战,文章提出注入具备情境适应性的机器美德、构建教学决策专用的提示词与智能体以及推动前瞻性社会实验等策略,以促进道德机器在教育领域的建设。

Teaching decision-making possesses strong moral attributes,emphasizing goodness and altruism.As generative artificial intelligence is applied to teaching,its powerful learning,understanding and generating capabilities enable it to undertake numerous non-programmed decision-making tasks.This also raises key questions:does generative artificial intelligence have the basic ability to support teaching decision-making?How will the future of education be affected?By reviewing existing research in sociology,psychology,pedagogy,and machine ethics,this paper finds that generative artificial intelligence has propelled teaching decision-making from an empirical approach to a collaborative human-machine model,while revealing that the enhancement of its autonomy is accompanied by the instability of moral sensitivity.This may lead to misunderstandings,deception,and shirking of responsibility between humans and machines,and subtly change the moral landscape in educational settings.To address these challenges,this paper proposes strategies such as embedding context-adaptive machine virtues,developing teaching decision-specific prompts and agents,and promoting forward-looking social experiments to foster the construction of moral machines in education.