基于大模型增强的通用人工智能教师架构  被引量:26

General Artificial Intelligence Teacher Architecture Based on Enhanced Pre-trained Large Models

在线阅读下载全文

作  者:余胜泉 熊莎莎 YU Shengquan;XIONG Shasha(Advanced Innovation Center for Future Education,Beijing Normal University,Beijing 102206,China)

机构地区:[1]北京师范大学未来教育高精尖创新中心,北京102206

出  处:《开放教育研究》2024年第1期33-43,共11页Open Education Research

基  金:“十四五”国家重点研发计划项目“农村地区教师教学能力智能评测与教学精准辅助技术研究”(2022YFC3303600)。

摘  要:以ChatGPT为代表的预训练大模型在教育界产生了巨大影响,也为发展通用人工智能教师带来了曙光。预训练大模型应用于教育存在幻觉、深度逻辑缺失、社会情感缺失等局限,如果这些关键问题不解决,大模型不可能在教育中得到真正有价值的应用。本文提出通过增强大模型构建通用人工智能教师架构,其核心思路是精调训练增强场景知识、检索增强认知、外部智能组件编排增强推理、多模态融合增强感知、情感计算增强社会情感,再通过教育知识图谱对大模型输出进行监督。通用人工智能教师主要有六种应用场景:需要渊博知识的场景、洞察创意增强场景、约束与管理场景、社会情感互动场景、个性化指导与反馈场景、多模态内容表现场景。文章最后分析了通用智能时代人类教师面临的机遇与挑战,提出教师需要主动拥抱、使用、驾驭并控制智能技术,解放、增强、进化自我,并跨越陷阱。Pre-trained large models have had a huge impact in the field of education,bringing a new dawn to general artificial intelligence teachers.There are some limitations in the current application of large pre-trained models in education such as hallucinations,a lack of deep logical reasoning,and absence of social and emotional understanding.Without addressing these key issues,large models cannot be truly valuable in its educational applications.This paper proposes the general artificial intelligence teacher architecture by enhancing large models.The core idea includes finetuning training to enhance scenario knowledge,retrieval-augmented generation to enhance cognition,orchestration of external intelligent components to enhance reasoning,multi-modal integration to enhance perception,and emotional computing to enhance emotional understanding.Additionally,the output of large models is supervised using an educational knowledge graph.This paper describes six application scenarios for the general artificial intelligence teacher including scenarios requiring profound knowledge,insight and creativity enhancement,constraints and management,social-emotional interaction,personalized guidance and feedback,and multi-modal content presentation.Finally,the paper analyzes the opportunities and challenges faced by human teachers in the era of general intelligence and suggests that teachers to actively embrace,use,master,and control intelligent technology,thus liberating,enhancing,and developing themselves,while overcoming pitfalls.

关 键 词:人工智能教师 生成式人工智能 预训练大模型 人工智能教育应用 

分 类 号:G451[文化科学—教育学]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象