The July 2021 issue of the academic journal ” Artificial Intelligence ” published by the Japanese Society for Artificial Intelligence has been published.
” Artificial Intelligence ” is an academic journal in which the editorial committee decides the theme and articles related to AI are brought together and published, and is published once every two months. It is an academic journal with a history of more than 30 years that covers fields and topical AI research that are familiar to us, and allows you to obtain current AI issues and the latest AI information. AINOW introduces the special features of each issue through interviews with researchers.
Click here for the previous introductory article in the journal (May issue)▼
This time, we interviewed Mr. Yoshimasa Tawatsuji (Waseda University) and Mr. Sho Yamamoto (Kinki University), who were in charge of the planning of the small special issue ” On ‘ Interaction through Virtual Space ‘”.
AI research and interaction are closely related, and there are fields such as HAI (Human Agent Interaction) and HCI (Human Computer Interaction) that study how humans, robots, and computers should interact. When it comes to interaction through virtual space, it is common to say that interaction in experimental settings built by introducing technologies such as robots, computers, VR, and sometimes AI technology is common. It was .
However, due to the spread of the new coronavirus, online communication has increased in various situations. In particular, video conferencing tools such as “Zoom” have come to be widely used in companies and universities, and interactions through virtual spaces have become familiar.
In particular, face-to-face communication between students and teachers was emphasized in the learning environment. There are issues such as On the contrary, it is changing to a different environment from the face-to-face learning that has been done so far, such as the casualness of being able to participate in the class just by accessing it from the web link.
How does communication in virtual space affect people’s thoughts and actions, and what kind of learning environment can be said to be suitable for learning? I asked the two of you.
In 2020, completed doctoral course at Waseda University Graduate School of Human Sciences. Ph.D. (Human Sciences). After working as an assistant at the University’s Faculty of Human Sciences, he is currently a lecturer at the University’s Global Education Center. His interests include Human-Agent Interaction, models of other people, and neural mechanisms that realize the uncanny valley. At the Japanese Society for Artificial Intelligence, he was an editorial committee member, an expert committee member, and was involved in past conference management.
In 2014, completed the doctoral course at the Graduate School of Engineering, Hiroshima University. Doctor of Engineering. After working as a specially appointed assistant professor at the same university’s graduate school and as an assistant professor at the department of information science at Kinki University, he became a lecturer at the department of information science at the same university in 2017 (current position). Next-Generation Fundamental Technology Research Laboratories and Informatics Research Laboratories. His main research fields are knowledge engineering, learning engineering, and application of AR /VR. He is engaged in research into technology for processing human learning and its processes, with a view to emotions and physical information. At the Japanese Society for Artificial Intelligence, he was also engaged in the secretary of research groups and the management of past conferences.
Table of Contents
- A project to reconsider “what is learning?” | Challenges of online education in the corona crisis
- The Future of Interaction in Education
- Collaborative learning support by separating what each person and agent can do
- Thoughts put into the special feature
- in conclusion
A project to reconsider “what is learning?” | Challenges of online education in the corona crisis
In this special feature, we will focus on interactions through such virtual spaces.
|1. Teacher-learner interaction
2. Learner-learner interaction
3. Person-agent interaction
Contributions by 7 researchers who are active in each of the three interaction spaces are posted. The differences from face-to-face interaction, the challenges of interaction in virtual space, usefulness, and future prospects are explained from each perspective.
ーーPlease tell us how this project came about.
Tawatsuji: With the spread of the new coronavirus infection, online communication has increased significantly at universities and companies. It’s completely different from face-to-face interaction, and it’s hard to understand emotional aspects such as “I wonder what the other person is thinking.” For example, when I ask in class, “Is this child understanding?”
This problem itself has long been pointed out as a problem of communication through online video, but what kind of situation and what kind of problems are occurring especially in education due to the corona crisis? I felt that I wanted to organize various things and deliver it to the readers, which is the beginning of things.
ーーNowadays, online is the mainstream way for students to take classes. However, there are many cases where the students turn off the video and the professor continues to talk alone while sharing the screen, which feels like a one-way street. When I think about this situation, I think, “What is interaction?” What do you think?
Mr. Tawatsuji: That’s right. When you meet someone in person, you can start a casual conversation, or you can sense that this person may be concentrating on something else at the moment. However, I feel that it is very difficult to obtain that information online. In this way (even when interviewing), you will be looking at the other person’s face through the camera, so it may be difficult to understand where the other person is actually looking, and it is difficult to guess the other person’s intentions. I think I’m in a situation.
There are many situations where the camera is turned off with video conferencing tools, but humans rely on non-verbal information for interaction, so I think it is quite difficult to interact without that information. In the field of education, I feel this even more clearly.
–Please tell us about the contents and perspectives that left an impression on you as the planners of the small feature.
Mr. Yamamoto: Everyone’s stories were very interesting. Taking a bird’s eye view of the contents, there were also stories that seemed to be enjoyed by those who are interested in EdTech (Education x Technology).
In addition, as Mr. Tawatsuji said earlier, there is information that is in a normal space but disappears in a virtual space. The way people look and behave. Perhaps the way I’m acting right now feels the same as usual when I’m on the screen, but since I’m only unconsciously moving my hands and face, the parts that are off the screen I don’t think it’s been communicated. Such information depends on the system and functions.
Focusing on this, for example, how to make up for it with VR, how to do turn-taking using information technology, how to interact with people’s information sharing, how to do input behavior, etc. Some people were there.
Other than that, it is difficult for a camera to convey the senses that a person has learned, such as the emotional information that a student seems to be bored in class. There was also a teacher who focused on whether to tell.
In the first place, when we think about virtual space, we tend to think of it as a space (limited by technology) like Zoom. However, in reality, virtual space is also a space where you can do anything. From now on, how should we think about interaction when it comes to online, various behaviors, psychological effects, social influences, and how to perceive the virtual in the first place. It’s planned.
Mr. Tawatsuji: As Mr. Yamamoto said, many of the articles were full of questions like, “What is learning in the first place?” I think that learning is a very fundamental thing, so I think there were many articles that were described based on what kind of aspect it is.
The Future of Interaction in Education
ーーWhat direction will interaction in education take in the future?
Mr. Tawatsuji: In many face-to-face classes, I think that education is often considered based on the assumption that there is only one teacher and many students, but that is not the case. I think that it will become more and more important in the future to design a system based on a firm understanding of what is being learned. It is important to provide more individualized support .
In that case, I think it will be necessary to provide support that considers how the child feels. However, of course, as technology progresses and becomes more sophisticated, I think that there will be various problems along with it, so I think it would be good if AI technology behaves well in education while carefully assessing that side. increase.
ーーThat’s true. I feel a little scared about the future of the result of truly exhaustive research on interaction.
Mr. Tawatsuji: That’s right. If you push it to the limit, you can also use it in a very dangerous way. I think it will be necessary to have a proper understanding of the technology and how far it can be incorporated.
ーーMr. Yamamoto, what do you think about educational interaction in the future?
Yamamoto: I also think that if all interaction research is really exhausted, all human behavior will be analyzed exhaustively, and in the end everyone will be able to naturally create interesting interactions.
In the end, it is also an interaction that the machine makes suggestions and people act accordingly. Therefore, it is probably possible to analyze why the mechanism occurs, and then consider how to design the system based on the mechanism.
A mechanical designer said that I like what he said, “It’s not as accurate as a machine as people think.” For example, if the arm of the machine decides to put the cup here, the machine can always put it in the same place. However, when I asked people to put it here, no one cared if it was just a little out of place. That amount of “play” will definitely remain, and there must be some places where you can interact well because of that play.
Collaborative learning support by separating what each person and agent can do
ーーI see, so you have to think about what it means to be human. In connection with how we relate to machines, I would also like to ask about the relationship with agents. What is the appropriate relationship between us humans and educational agents?
Mr. Tawatsuji: In my article this time, I am targeting characters called “Pedagogical Agents” that exist as computer programs in educational learning, but I am talking about dating learners. I believe that this kind of existence will appear in various forms of learning in reality in the future.
At that time, it is very important to know in what position the agent exists. When you are alone in space and there is a teacher (a character in a computer environment) and other learners, it is necessary to add functions that are fully conscious of the role that the agent will be there as an existence. I think it will come.
I think it’s important to add functionality, and technology can’t do everything, so it’s necessary to clearly separate what it can do and what it can’t do. So, in my personal opinion, agents will become something that makes us look back on what kind of existence we should be there, rather than helping us to support education and learning. I’m thinking about it.
ーーTeachers have a role to teach, and students have a role to listen and listen. Is that what you mean?
Tawatsuji: Yes, that is very important. When I was watching education, I thought that human teachers are amazing. After all, when I’m doing face-to-face classes, I can sense that “Is this child in trouble?” You can realize that “this child may not understand a little now”. It is still impossible for AI to have the ability that teachers can perceive. Therefore , I think it would be good if we could divide the functions into what only teachers can do and what AI is good at, and provide cooperative learning support in the future.
ーーHow can we measure the learning effect as a result of using such an agent?
Tawatsuji: I think it’s difficult to measure learning effects.
This is because I believe that what can be done to improve the efficiency of a child’s learning largely depends on the person. I don’t know when the learning effect will appear. For example, even if you are taking a class at university, you may find out later that “I didn’t understand at all when I was listening to the lecture, but what that teacher actually meant at that time”. . In that sense, the learning effect may appear with a delay, so I think it is difficult to evaluate the system itself.
As I mentioned in this article, I think I will need emotional support myself. For example, it’s like a partner who can snuggle up to you when you think learning is hard. It may not be possible to measure the effect, but I believe that agents that support learners will become necessary in the future.
ーーMr. Yamamoto, what do you think about the relationship with these agents and the learning effect?
Mr. Yamamoto: Listening to what you two have to say, I think there are probably two ways of getting involved when an agent is involved in learning in a learning setting. One is related to the learning itself, and the other is an emotional method such as encouragement.
In the case of the former, I personally feel that it has not yet reached practical use. When a professional teacher teaches a subject one-on-one, it can be done with a certain amount of cost, but when the number of learners increases to 10, 100, or 1,000, the teaching side becomes more difficult. Therefore, AI can make learning more efficient and faster, like some apps for individual learning that have become popular recently. However, I probably haven’t gotten to the point where I think , ” If AI can speed things up that much, maybe we can change the way we study in the first place .” I personally think that such a part can be made more efficient, and I am currently researching it.
So, when it comes to the learning effect, I think it will hit a ceiling if you think about how to learn efficiently with the current learning method. If you can think of how to change shifts efficiently with the way you do it, I think you can increase the learning effect even more.
On the other hand, it is the latter AI that is emotional and close to people. When a learner stumbles, he says, “Do your best.” Or talks to him when he’s worried. I think that learning efficiency will increase more and more if such AI is completed.
Thoughts put into the special feature
ーーLastly, what kind of people would you like to read this special issue?
Mr. Tawatsuji: Since we are familiar with education and learning on a regular basis, we tend to think that we have understood it, but I would like you to look back once more and ask yourself, “What is education and learning?” If you are asked what education/learning is, and you can’t say it well, or if you have even a slight sense of what education or learning is, I would definitely recommend reading this book.
I would be very happy if our articles could give you an opportunity to think about what learning is. The theme is “via virtual space”, so I hope that people who are doing online education will read it.
Mr. Yamamoto: As Mr. Tawatsuji said at the beginning, the focus is on rethinking how people learn in the first place. I think that we can provide a new perspective to reconsider the question, “Isn’t this kind of activity?” So, if students read it, they will come up with hints when thinking about how they will study in the future, and researchers will also see the novelty of their own research from a different perspective. you may be found.
Or if you are a teacher, I think that you can handle classes intuitively to a certain extent when you are teaching normally, but this (due to the corona disaster) has imposed restrictions on interaction, so it is somewhat intuitive. I think I had to rethink what I was doing. As these points become clearer, having educators read it may also help them dig deeper into the activity of learning.
ーーThank you very much!
Learning environments and methods of learning have changed significantly over time.
I myself felt that due to the influence of the coronavirus, classes at the university were moved to the “virtual space” of online, and learning was restricted. However, the online environment has created flexibility in the time for discussions, and various digital tools are now being used. In other words, communication and tools that try to compensate for what is lacking in real-world interactions are diversifying.
We take the word and action of “learning” for granted. However, by re-examining the way of learning in the wake of the corona crisis, there may be changes in teaching methods and learning attitudes. The future may soon be one in which more efficient learning becomes possible through collaboration with agents who are close to people, as the two of you mentioned.
It is packed with cutting-edge research on the future of education.
The Japanese Society for Artificial Intelligence Please take a look at the July issue of the academic journal ” Artificial Intelligence “.