The impact of AI-teachers on the human teachers who use them in schools

人工智能教师对学校中使用人工智能教师的影响

• 批准号: 20K13900
• 负责人: ヤマモト.ラヴェナー ロクサナ
• 金额: $ 2.66万
• 依托单位: Ochanomizu University
• 依托单位国家: 日本
• 项目类别: Grant-in-Aid for Early-Career Scientists
• 财政年份: 2020
• 资助国家: 日本
• 起止时间: 2020-04-01 至 2024-03-31
• 项目状态: 已结题
• 来源: https://kaken.nii.ac.jp/en/grant/KAKENHI-PROJECT-20K13900/
• 关键词: AI in education;; AI tools for teachers;; Affective domain; AI in education; AI for teachers; teacher - AI cooperation; AI-teacher; AI-education; digital sociology; sociology of education

The project continued with investigations on the usefulness of AI tools adopted by or designed for teachers. There are three lines of research which report achievements:1. The study surveyed Japanese schoolteachers of English language who, with the scope of teaching pronunciation, adopted an AI tool which uses machine learning algorithms to analyse, and natural language processing techniques to understand users' speech. Findings show that such teaching tool not only provides a better teaching-learning experience, but may render traditional English pronunciation teaching techniques obsolete.2. Investigation on teachers' need to appraise student affect and their desire to use facial emotion recognition (FER) AI to complement their natural methods yielded positive results.3. The developed FER application was improved to allow detection of deviations from default student emotional state. Emotions are now modelled in terms of valence (positive/negative nature of emotion) and arousal (emotion intensity) by training a multi-task convolutional neural network, in order to create a continuous spectrum of emotional experiences, rather than simply categorising emotions into discrete classes (happy, neutral, sad etc.), as it did before.

该项目继续调查教师采用或为教师设计的人工智能工具的有用性。有三条研究路线报告的成就：1。该研究调查了日本学校的英语教师，他们在教学发音的范围内，采用了一种人工智能工具，该工具使用机器学习算法进行分析，并使用自然语言处理技术来理解用户的语音。研究结果表明，这种教学工具不仅提供了更好的教学体验，但可能会使传统的英语语音教学技巧过时.对教师评估学生情感的需求以及他们使用面部情绪识别（FER）AI来补充他们的自然方法的愿望的调查取得了积极的结果。开发的FER应用程序进行了改进，允许检测偏离默认的学生情绪状态。现在，通过训练多任务卷积神经网络，根据效价（情绪的积极/消极性质）和唤醒（情绪强度）对情绪进行建模，以创建连续的情绪体验谱，而不是简单地将情绪分类为离散类（快乐，中性，悲伤等），就像以前一样。

项目成果

Intelligent Agents Influx in Schools: Teacher Cultures, Anxiety Levels and Predictable Variations

智能代理涌入学校：教师文化、焦虑水平和可预测的变化

• 发表时间: 2021
• 期刊: Proceedings of the 22nd international conference on Artificial intelligence in education. AIED 2021.
• 作者: ASADA;S.;NAZEER-IKEDA;R.;KARTIKA;D.;木下寛子;胡中孟徳;板橋雅則;R. Yamamoto Ravenor
• 通讯作者: R. Yamamoto Ravenor