Professor Ryuichiro Higashinaka

Researchers’ Voice No.45

Graduate School of Informatics

My favorite saying: Persistence pays off

Q: Why did you choose this saying?

Even if things don’t go right at first, I believe that if you keep trying, you’ll often get it right in the end. I’ve found that even when something seems impossible at first, you may find a solution if you keep trying. Giving up halfway means it’s over, but persisting through trial and error and making gradual progress will allow you to achieve your goal. Whether it’s studying, researching, or doing a hobby, giving up too soon means it’s all over. If there’s something you want to do, I think it’s important to keep doing it with dedication.

Q: What kind of research are you doing?

I’m doing research on dialogue systems — those computer systems that have conversations with people. It seems like more and more people are becoming familiar with these systems, especially with the increasing use of voice agents on smartphones and communication robots, and recently, those using large language models. One of the goals of my research is to make systems communicate more naturally, like humans do. Right now, I’m working on systems that offer interactive services, especially in customer sales, and that augment our interactions and collaborate with us through dialogue.

Q: How did you get into this research?

When I started going to college, I never thought about becoming a researcher. It was during graduate school that I got involved in a long-term internship at a company where I was exposed to research in natural language processing and interactive robots. I found that research fascinating, so I joined a company’s basic research lab. Initially, I wanted to research machine translation, but since it was already in the commercialization phase, there wasn’t a dedicated department for it. Instead, I was assigned to the dialogue systems group. Since then, I’ve been working on research related to dialogue systems and natural language processing.

Q: What sorts of things make you think that research is interesting or rewarding?

I find it really interesting to get feedback from users after they’ve actually used our system. Of course, it’s great when our research succeeds, but even failures often lead to new discoveries. In general, research goes wrong more often than not, but I still enjoy it. Personally, I love experimenting and discussing ideas with students, as well as writing manuscripts together. Turning a vague concept into something concrete can be tough, but it’s also really rewarding.

Q: The demand for spoken dialogue systems, where you talk to a robot and it responds or tells you the answer, has grown in recent years. So, it is expected that the technology for dialogue systems will continue to evolve. Is there anything you keep in mind when developing the systems?

I always think about whether the systems I create will contribute to gaining new insights into human beings. While engineering practical solutions is important, designing dialogue systems is about understanding how people use language. I’m fascinated by why and how people are able to communicate with each other. So, I always aim to do research that not only makes our systems function well but also provides new insights into human communication.

Q: We’ve heard that students can attend your seminars and study groups online, and that some of the seminars are held in a virtual reality (VR) space. Using the VR goggles you provide, students can participate from home as if they were in the lab. How have the students responded?

I’ve found that they responded quite well. Using VR for my seminars creates an immersive experience, making remote students feel like they’re actually there, even if they are actually far away. I can see where they’re looking and understand their intentions through nods and raised hands. Another reason why I use VR is to educate about VR itself and to develop VR dialogue systems. By actually using VR, students see the differences from regular web conferencing and explore the unique interactions VR offers. This experience also aids in creating CG agents and can be used for social gatherings as well.

Q: Please tell us about an experience that you can only talk about now.

Since I was a child, I’ve not been good at writing, even though these days I correct students’ papers in my lab almost every day. When I started working at a company, I had to write a lot, but I struggled to put my thoughts into words. I found myself writing just a few lines when I should have written several pages, which must have surprised my boss. Then, one day, my boss was too busy and so I had to write a paper by myself. Through trial and error, I managed to finish it. When it got accepted for publication, it boosted my confidence a bit. Since then, I feel like I’ve been gradually improving in expressing what I want to say. People aren’t born with writing skills; it’s something you develop through daily practice and reviewing your writing over and over again.

Q: How do you spend your days off? What do you do to recharge?

I try to watch movies regularly. I also like to play golf, although I’m not very good at it and haven’t been able to do it since the COVID-19 pandemic. I also want to start going to the gym again, as I was a little worried by the results of my recent medical checkup.

Q: What are your goals and ambitions for the future?

I’ve been in my current position for four years now. Our lab has expanded in both size and diversity. The rise of large language models presents an opportunity to dramatically improve dialogue system performance. I’m eager to push forward and accelerate our research in dialogue systems.