Summaries of speaker input in French AI Seminar in Seoul: Educate AI before using AI Education!

These are the notes of the speakers’ input at the international conference held in Seoul on October 22, 2024 by the French Embassy of South Korea.

KS Park:  

Opening remark: We may need to educate AI before using AI in education 

Use of AI in education raises a much more sensitive set of ethical questions.  The current version of AI is machine learning.  AI learns to mimic human behavior by studying what humans do.  However, human behavior is full of bias, injustices, and ignorance.  Sanitizing the training data so that AI does not mimic these biases, injustices and ignorance is one of the most important ethical challenges of AI. Now, such ethical need is even greater for AI used for education as AI runs the risk of educating students into those biases, injustices and ignorance.  Even before AI, freedom of speech and diversity allowed in the real world has not been allowed in the same breadth and depth inside the classroom as people at young age are easily influenced by early exposures. Post AI, such need remains the same and therefore it is important to moderate and censor the training data used for training the AI used for education.  Simply, we need to make sure that MS Chatbot Tay does not teach our students

There are many initiatives to govern AI but not much on how to govern the universe of the training data.  Most initiatives are focusing on “slowing down” the AI’s learning than ethically training the AI. AI is at the pinnacle of automation, a trend that has steadily continued for centuries, freeing people from menial and often dehumanizing labor (think laundry machines and home makers! Or border officials and passport checks). 

We should spend more resources and time trying to make the training data ethical, especially for the AI used in education, which will influence our students.  How will the government create the training data free from bias and ignorance? 

Questions:

1. Korean constitution protects impartiality, independence, and professionalism of education.  It prohibits public education from becoming the tool of the incumbent government indoctrinating people into certain versions of thoughts and culture. Yet the government is bound to have some control on educational content.  Therefore, there have been many controversies and disputes, for instance, in accrediting history and social science textbooks. How will AI education governance address this age-old problem?   *Especially, Korea is developing AI textbooks with the goal of individualizing curriculum for each student from diverse versions of curriculum tailored to their capacity and interest.  How will we make sure that all these versions meet the requirements of impartiality, independence, and professionalism.

• How will the government create the training data for AI used in education, that is free from data protection law violations and copyright infringements? 

Bruno Galmar

I am French and have been working in Taiwanese universities for more than ten years. In Taiwan, the current wave of AIin education followed the wave of computational thinking and coding for all. In 2018, I started working in a private university that used “two languages : coding and foreign” (Python and English) as its marketing strategy. One or two years later, the marketing strategy changed to “不要叫我大學生 叫我AI生” “Don’t call me a university student, call me an AI student.” The emphasis on AI education has been growing stronger every year.

Currently, I work in a public university renowned for its scientific departments, where there is a strong effort to integrate AI in the humanities. I am involved in that effort. I am part of a group of professors of departments in the humanities (History, Philosophy, Chinese and French) aiming at fostering AI research and education within our faculty.

I am interested in AI development and policies—specially regarding AI education—in Taiwan, France and Quebec. I also keep an eye on AI research in Japan.

Session : 1 will be a straightforward presentation of each country’s governance system on AI in education

I will focus on the example of Taiwan, drawing comparisons with French-speaking countries where my students go for exchanges during their studies.

Taiwan, France and Quebec all have ambitions to become key leaders in AI innovation and AI education.

Of the three, I believe that Taiwan has the strongest ambition regarding AI.

The new president aims for Taiwan to become “a smart AI island”. Taiwan’s minister of economic affairs 郭智輝 aims for Taiwan to reach the top 3 in AI powers by 2028 !!! Taiwan has succeeded in the semiconductor industry, in the machinery industry and several other high-tech domains. Why not in the AI industry as well?

One of the key problems is training enough AI specialists.

One part of Taiwan’s plan is to train 100,000 foreign students and 100,000 domestic students to become AI engineers by 2028, to work locally.

The AI Taiwan Action Plan (2018-2021) resulted in the training of 30,000 AI specialists. The AI Taiwan Action Plan 2.0 (2023-2026) aims to accelerate training, starting AI education from primary schools. In Taiwan, the idea to start as early as possible appears to me a distinctive feature with France’s vision of AI education starting from secondary education.

Taiwan has recently built its own LLMs : Trustworthy AI Dialogue Engine (TAIDE) customized with Taiwanese features and supporting the local traditional Chinese characters. This is one of its first steps toward becoming a smart AI island and could also serve as a resource for AI education.

A concrete example of the current AI education effort : one of my students in our French department is also enrolled in an AI certificate program where she studies : AI and Law, Introductory Computer Vision, NLP, An Introduction to AI through Literature (my own class) and other AI courses. Once she will have graduated, I think she will have a strong interdisciplinary profile : trained both in humanities with very strong language skills in English and French and an overall vision of AI and its applications in different domains.

France has its own ambition regarding AI. I noticed that for several years the amount of AI and machine learning high-quality content in French for education is growing fast (e.g. CNRSFormationFIDLE). Highly renowned researchers and great professors (e.g. Stéphane Mallat) are strongly committed to organize France AI’s education and to make France one of the best places to become an AI specialist.

Quebec has also a strong offering in AI education and a vivid AI ecosystem connecting researchers and companies of all sizes and of various economical domains.

Appendix

https://www.info.gouv.fr/communique/comite-de-lintelligence-artificielle

https://www.economie.gouv.fr/strategie-nationale-intelligence-artificielle

President Lai Aims To Make Taiwan a ‘Smart AI Island’ | TaiwanPlus News

(https://www.rti.org.tw/news/view/id/2219857)

Taiwan Sets Export Record in August on Booming AI Demand｜TaiwanPlus News

Tech Giant Nvidia Chooses Taiwan for Its First R&D Center in Asia | TaiwanPlus News

Session 2 deals with substantive ethical issues that AI in education face or is likely to face in the future

Taiwan renowned NTU’s Electrical Engineering professor 葉丙成 was part of a team to respond quickly to current ethical issues of AI in education, proposing MOE guidelines to inform teachers, students and parents for primary and secondary education (中小學使用生成式人工智慧注意事項/Guidelines for Elementary, Junior High and Senior High School on Using Generative Artificial Intelligence). The same professor has been highly successful in gamification of education (https://www.pagamo.org/) and is a charismatic voice in debates regarding the future of education in Taiwan in the Gen AI era.

In the humanities departments, we are gradually evolving to offer AI classes to avoid being seen as lagging behind by students and parents, who are constantly reminded that AI skills are key to a bright professional future. We have started training colleagues willing to adopt the Generative AI tools, and discussing how to adapt our curriculum, class practice and assessment methods. There is a need for training in AI ethics for education.

The high-quality of text generation by AI has disrupted the way we conduct writing classes in foreign languages. Even for oral classes, in-class brainstorming activities can be disrupted by students using AI at the first stage of brainstorming. We need to establish guidelines on how AI should be used in classroom activities.

With several departments of the Humanities, we are brainstorming to launch research projects on AI ethics and philosophy of AI, where our expertise could shed new light on current debates.

Myoungshin Kim

• AI offers an opportunity to rethink the essence of education
• Few issues have garnered such extensive international attention and triggered concrete actions as quickly as AI.
• We should use these concerns as a catalyst to redesign our common future, and this opportunity must not be missed.
• However, there is no consensus on how to mitigate AI’s risks while maximizing its benefits. Sharing our successes and failures will be critical to finding better solutions.

• The AI Divide is a pressing concern
• A recent report by the UN and ILO, Mind the AI Divide, highlights the growing concerns over the AI divide and stresses the need for quality education.
• While interest in AI and demand for education are increasing, there is a significant gap in the availability of quality education.
• At LG AI Research, we offer free, high-quality AI education to over 30,000 students, including K-12, university students, and professionals, annually. All our programs include AI ethics education to help citizens become mature AI users and critical overseers.
• On a global scale, we are collaborating with UNESCO to develop an AI ethics MOOC, which will be launched in the first half of 2026.

• The necessity of AI ethics education for AI developers and researchers
• Since AI is ultimately created by humans, it is essential to focus on their awareness of AI ethics and human rights sensitivity.
• The embedded ethics approach is gaining traction in some universities abroad.
• At LG AI Research, we conduct an annual AI ethics awareness survey and hold bi-weekly AI ethics seminars where researchers themselves present, discuss, and propose improvements rather than relying on external speakers.
• Additionally, we provide education on the Universal Declaration of Human Rights and the Sustainable Development Goals to data workers and red-teaming members to improve their human rights sensitivity.

Jungmin Kwon*

What is an AI Digital Textbook?

• AI-based diagnostics and analysis:
→ For students: Provides the optimal learning path, personalized prescriptions, and support.
→ For parents: Offers information necessary for guiding their children, such as academic achievement and subject interest status.
→ For teachers: Provides necessary information for supporting students, such as academic engagement by class and by individual students.

When and for which subjects will AI digital textbooks be introduced?

• 2025 onward: Introduction of AI digital textbooks in Math, English, Information Technology, and Special Education (Korean).
• By 2028: Expansion to subjects like Korean, Technology and Home Economics, Social Studies, and Science.
  ★ Features such as audio descriptions, subtitles, and multilingual translation will be provided to ensure that everyone, regardless of physical disabilities or multicultural backgrounds, can receive personalized education.

Ministry of Education claims to aim to open the era of 1:1 personalized education with AI digital textbooks and create an educational environment that supports the human development of students.

Criticism

From a constructivist perspective, the use of AI digital textbooks could be problematic, as it might reinforce a traditional, one-way transmission of knowledge, which contrasts with the fundamental principles of constructivism. Constructivism emphasizes the importance of learners actively constructing their own understanding through experiences and interactions, rather than passively receiving information from static resources like textbooks.

In the AI era, where knowledge evolves rapidly, relying on textbooks—even AI-enhanced ones—risks presenting outdated or overly rigid information that doesn’t reflect the fluid nature of learning in the 21st century. AI’s potential should be harnessed to foster environments where students can explore, experiment, and create knowledge through problem-solving and critical thinking, not just absorb prepackaged content. Technology should be used as a tool for creativity rather than simply replicating traditional models of rote learning.

Korea’s current education system, with its emphasis on drill-and-practice, high competition, and rote memorization, contributes to chronic stress and depression among students. AI digital textbooks, if designed similarly to current educational methods, may further entrench these issues rather than solving them. Instead of helping students engage with material in dynamic, innovative ways, it may perpetuate the same competitive, standardized approach that stifles creativity and individual growth.

For AI to truly enhance education, it must move beyond the limitations of the textbook format. In fact, the words “AI” and “textbook” put together seems ironic. AI technology in schools should enable students to engage with diverse perspectives, create their own projects, collaborate with others, and develop critical thinking skills in meaningful, contextualized learning environments. AI should serve as a catalyst for creative exploration and deeper learning, rather than just another tool for reinforcing outdated educational paradigms.

An additional concern with AI digital textbooks is the issue of data privacy. These systems rely heavily on collecting and analyzing student data to provide personalized learning paths and recommendations. However, if students or parents do not consent to sharing personal data, this creates a significant challenge. Without access to this data, the AI system may not be able to provide the tailored support it promises, leading to a potential divide between students who consent to data sharing and those who don’t. Furthermore, there are broader ethical concerns regarding how this data is stored, used, and protected, especially in light of increasing cybersecurity threats. However, the current policy does not warrant rights of students and parents to opt-out without facing disadvantages.

* She also argued during the discussion that the South Korean government envisions AI digital textbook having all the answers and interrogating students to find out the fallacies of students. Besides the privacy concern over this process, she believes that it is the students who should interrogate AI, and we need the human teachers to provide those motivations to the students, and that the current national educational policy focused on injecting the correct and selected set of knowledge into the students cannot succeed in this motivating function.

Ikuko Yairi**

(1) Japanese Ministry documents

Ministry of Education, Culture, Sports, Science and Technology “Handling of generative AI in the educational aspects of universities and technical colleges” (2023-7)
https://www.mext.go.jp/b_menu/houdou/2023/mext_01260.html

(2) Japanese University initiatives

(i) Professor Yoshida Rui’s (University of Tokyo) website: ChatGPT/AI education-related information summary, methods of using ChatGPT/AI at domestic universities, examples of practice, and materials
https://edulab.t.u-tokyo.ac.jp/chatgpt-ai-resources/

(ii) AI Research Institute: 5 examples of generative AI use at universities (Tohoku University, Ritsumeikan University, Toyo University, Kinki University, Musashino University)
https://metaversesouken.com/ai/generative_ai/university/

(3) Elementary and secondary school initiatives

Ministry of Education, Culture, Sports, Science, and Technology: Leading DX School aiming to “create effective educational practices” using generative AI Generative AI pilot school examples
https://leadingdxschool.mext.go.jp/ai_school/

(4)Presentation materials by Professor Tamura Yasuhisa, Education Center, Sophia University (private documents below)

(i) Changes in learning and evaluation due to generative AI and expectations for educational technology research (Information Systems Society Research Meeting 2023-5)
(ii) Changes in education and learning brought about by generative AI (lecture at Sophia University’s Basic Education Center 2023-5)
(iii) Activating active learning using generative AI (Lecture at Hiroshima University 2024-2)
(iv) The future of learning with the introduction of information technology and AI (Sophia University Trial Class 2024-8)

In session 2, I will be able to answer questions about the current situation in Japan and participate in the discussion from the perspective of an AI researcher who implements deep learning models.

**This speaker could not make to the event due to unforeseen circumstances.