Recent Activity


１．Educational application of large language models (LLMs)

1) at Hong Kong Baptist University

I have accumulated experience in workshop courses aimed at fostering creativity, such as the Design School at Kyoto University and the SHIP program at Waseda University. Designing and implementing these courses come with various challenges. For instance, while the importance of holistic evaluations in workshop courses is increasing, the burden on faculty members is significant. Therefore, achieving efficient and fair evaluations is necessary. Among these, evaluating reflection essays written by students is particularly troublesome.

In response to this, I started research on the educational application of large language models (LLMs) at Hong Kong Baptist University while I was a visiting professor from 2023 to 2024. Specifically, two experiments using an LLM (ChatGPT 4.0) were conducted for workshop courses targeted at undergraduate students.

Experiment 1: Facilitation of integrating evaluation results from multiple instructors. This involved scenario-based case studies, and the results are summarized below.

  • Toru Ishida, Tongxi Liu, Hailong Wang, and William K. Cheung. Facilitating Holistic Evaluations with LLMs: Insights from Scenario-Based Experiments. arXiv:2405.17728, 2024. (Presented at the 32nd International Conference on Computers in Education (ICCE 2024))

Experiment 2: Automated essay evaluation using various methods. This included quantitative and qualitative analysis using actual data, and the results are as follows.

  • Toru Ishida, Tongxi Liu, Hailong Wang, and William K. Cheung. Large Language Models as Partners in Student Essay Evaluation. arXiv:2405.18632, 2024. (Essays and analysis results used in the research is here.)
  • Toru Ishida, Tongxi Liu, Hailong Wang, and William K. Cheung. Harnessing Generative Diversity: An LLM Framework for Consistent Qualitative Assessment. TechRxiv. December 02, 2025. DOI: 10.36227/techrxiv.176107757.72471376/v2 (This is a rewrite of the above paper for artificial intelligence researchers.)

Through these two experiments, LLMs demonstrated the capability to become members of evaluation committees composed of faculty members.

2) at Telkom University, Indonesia

From 2024 to 2025, I serve as a visiting professor at Telkom University in Indonesia. Together with a team of 7 to 8 faculty members, we will apply generative AI to several regular courses. In countries referred to as the Global South, such as Indonesia, the student-to-teacher ratio is exceedingly high at 30:1. Considering that advanced universities in Europe and America have ratios of less than 10:1, innovation is essential for enhancing education. The utilization of generative AI is highly likely to become a breakthrough. We have just begun our research, but we have summarized the plan in the following article.

  • Kemas Muslim L, Toru Ishida, Aditya Firman Ihsan and Rikman Aherliwan Rudawan. Advancing Global South University Education with Large Language Models. arXiv:2410.07139, 2024.

In the process of promoting a project to utilize generative AI in education, we found that undergraduate students were not able to use LLMs effectively. We also observed that some faculty members were teaching students the techniques of LLM usage in industry, which does not necessarily support their learning. Therefore, in the context of higher education, I gave a lecture emphasizing the importance of four principles: Let Yourself Learn from LLM (Don’t try to “use” it), Ask Simple Questions and Keep the Conversation Going (No Need for Prompt Engineering), Enjoy Exploring Beyond the Topic (LLM is not a Google search), Respect LLM as Your Lifelong Mentor (Like Kongming in Ya Boy Kongming). The lecture slides are here (Lecture 1, Lecture 2, Audio Overviews).

2．Rethinking Autonomous Agents

The emergence of large language models (LLMs) is significantly advancing the study of autonomous agents, a field that has long faced major challenges. In response, I am currently pursuing the following lines of research:

1) Impact of LLMs on Conversational Translation

This research explores the impact of LLMs on conversational translation. Unlike formal text translation, conversational translation involves colloquial expressions, relies heavily on context and cultural background, and demands a wide range of general knowledge—factors that have traditionally made it a particularly difficult task. LLMs have remarkably addressed many of these long-standing issues, which have been discussed over the past two decades. A summary of our findings is presented below:

  • Toru Ishida, Yohei Murakami and Arif Bramantoro. Impact of Large Language Models on Conversational Translation: Towards Translation Agents. in the 7th International Conference on Applied Computational Intelligence in Information Systems (ACIIS 2025), invited paper, October 28, 2025. (also in TechRxiv. December 02, 2025.
    DOI: 10.36227/techrxiv.173014408.89705418/v2). 

2) Survey on AI Agent Research

For decades, the field of autonomous agents and multiagent systems (AAMAS) has developed foundational theories of autonomy and coordination. While LLMs have recently reignited interest in AI agents, their integration with these well-established frameworks remains limited. This survey examines AAMAS’s theoretical foundations (Concept), their implementations using LLMs (Code), and their deployment in society (Commerce), offering an integrated framework for understanding the current trajectory of AI agents. The aim is to bridge the groundbreaking practical advances driven by LLMs with the theoretical insights from AAMAS, thereby catalyzing future progress in the field.

  • Toru Ishida, Yohei Murakami, Donghui Lin, and Kemas Muslim Lhaksmana. AI Agents: From Concept to Code to Commerce. International Conference on Information and Communication Technology (ICoICT 2025), pp. 1-7, invited paper, IEEE eXplore, 2025.
  • Toru Ishida, Yohei Murakami, Donghui Lin, and Kemas Muslim Lhaksmana. AI Agents: From Concept to Code to Commerce – A Survey Bridging AAMAS and LLMs. TechRxiv. August 25, 2025. DOI: 10.36227/techrxiv.175616139.97637301/v1. August 25, 2025. (This is the extended version of the ICoICT paper, the Japanese translation is also available)

3. Lectures (primarily) in Southeast Asia

Additionally, I am engaged in giving lectures primarily in Southeast Asia, including the Philippines, Indonesia and Vietnam. Recent lectures are as follows:

  • Impact of Large Language Models on Conversational Translation. Keynote, International Conference on Applied Computational Intelligence in Information Systems (ACIIS 2025). Brunei Darussalam. October 20-22, 2025.
  • AI Agents: From Concept to Code to Commerce. Keynote, International Conference on Information and Communication Technology (ICoICT 2025). Telkom University, Bandung, Indonesia. July 30-31, 2025.
  • Design as the Core of Transdisciplinary Education: Reflections on Two Decades of Practice. Pacific Asia Summit on Transdisciplinary Education (PASTE 2025). Hong Kong Baptist University. June 25, 2025.
  • LLM as a Learning Buddy: Rethinking LLM in an Educational Context. Online lecture, Telkom University, Indonesia. April 25 and May 2, 2025.
  • Facilitating Holistic Evaluations with LLMs: Insights from Scenario-Based Experiments. The 32nd International Conference on Computers in Education (ICCE 2024), Manila, November 28, 2024.
  • From Babel to Harmony: The Evolution of Intercultural Collaboration with Machine Translation. Vietnam National University Ho Chi Minh City (VNUHCM) – University of Information Technology. September 27, 2024.
  • Designing a Sustainable World: An Ideation Workshop Using the Drawdown Solutions Library. Vietnam National University Ho Chi Minh City (VNUHCM) – University of Information Technology. September 25 – 26, 2024.
  • From Babel to Harmony: The Evolution of Intercultural Collaboration with Machine Translation. Keynote, International Conference on Artificial Intelligence, Blockchain, and Cloud Computing (ICoABCD). Bali, August 20, 2024, Bali, Indonesia.
  • Digital City, Smart City and Beyond. Keynote (with Mika Yasuoka), 20th International Conference on Intelligent Environments (IE 2024) (online), June 20, 2024.
  • Converging Research on Agents and Avatars: The Emerging Evolution in the Metaverse. Keynote, De La Salle University Research Congress 2024 (online), June 21, 2024.
  • Agents and Avatars in the Metaverse. Vietnam National University Ho Chi Minh City (VNUHCM) – University of Information Technology. March 22, 2024.
  • Agents and Avatars in the Metaverse. De La Salle University, Manila, Philippines. September 26, 2023.
  • Experiences at Kyoto University Design School. Department of Computer Science, University of the Philippines Diliman. September 25, 2023.
  • Agents and Avatars in the Metaverse. NVAITC Symposium 2023, Singapore. February 24, 2023.