Unsure about AGI, LG AI Research’s Junior Members Just Want to Succeed

The juniors at LG AI Research don’t entirely fit under the label “Gen Z.” They’re young, but they have a stronger sense of responsibility than anyone else, and they have a passion for research and learning. They’re the “most junior of juniors,” yet they also nag the senior researchers. Introducing the LG AI Research Life of these juniors united by their personalities!

Meet Our Junior Members Who Want to Succeed!

Data Intelligence Lab Dongmin Kim

#Younger_Brother_Image #The_Cutie_of_the_Senior_Researchers

I’m primarily working on a task called “anomaly detection.” To use a metaphor, it’s a kind of health checkup. When we go for a checkup, we use the results of blood tests, endoscopies, etc. to understand our health. The same is true for data. Observing unstructured data, or “tabular data,” from manufacturing processes can help identify which sensors are having problems, or if a specific part is defective. I’m working on detecting outliers in particular processes.

Multimodal Lab Hyunmin Lee

#Fearless_Junior #Do_It_Until_You_Know_It

I’m researching image editing technology using generative AI. You can think of it as research using AI to generate different elements within an image. For example, a designer might consider different colors or materials for a particular product. It would be pretty cumbersome and time-consuming to go back and forth between samples. Utilizing image editing technology, you can quickly create a variety of results. You can change and combine not only the product’s color, but also the material, texture, and other elements.

Language Lab Changho Lee

#Don’t_Be_Fooled_By_My_Sturdy_Appearance #I’m_Young

I’m working on instruction tuning research. Instruction tuning is a way to improve the quality of a large language model (LLM) by fine-tuning the model with different sets of input and output instructions. My main focus is on researching which approach yields optimal model performance when learned with different approaches, and developing it to improve performance.

EXAONE Lab Heuiyeen Yeen

#The_Most_Junior_of_Juniors #Nagging_Seniors_Old-school_Style

I’m researching how to create instructional tuning data with human annotators (people who specialize in data labeling) and automatically generate it through machine generation. Through instruction tuning data, we’re building the world’s most powerful large-scale language model. It does everything well in both general and academic mass coding. 

The Brilliant LG AI Research Lives of the Juniors

Changho Lee : I’ve been living the LG AI Research life since I was an undergraduate student. As part of LG’s Industry-Academia Project Program, I interacted with researchers from LG AI Research every week during my project. As they shared deep insights into their research, I realized that I had so much to learn from them. Most of all, I could tell that they genuinely loved and enjoyed their research. At the end of my Industry-Academia Project, I applied to be an intern and was hired.

When I transitioned from intern to full-time employee, I think the thing I heard the most from people was, “You must be really smart to be working at LG AI Research.” In truth, I was the most junior member with a lot to learn and not even enough time to read papers every day. Still, I feel like I’ve grown every day since I joined the company. 

Hyunmin Lee : Being a member of LG AI Research has definitely changed the way I approach research compared to when I was a student. As a graduate student, writing papers and surpassing state-of-the-art (SOTA) performance were the biggest goals of my research. I think the difference now is that I’m expanding my thinking beyond the performance of AI technology in order to incorporate actual technology into services.

Also, the more I research, the more I realize that it’s really important to be in constant communication with my seniors and fellow researchers. I feel like I’m growing in both my work and my capabilities by making sure my research is on track and by talking with others about whether there’s a better approach.

Dongmin Kim : I also think it’s really important to talk to the researchers around you. When I get stuck while researching, instead of worrying about it alone, I reach out to other researchers. I explain what I’m doing and ask for their thoughts on how to proceed. Interestingly, I often get a lot of ideas in the process, and I find myself organizing my thoughts.

There are a lot of incredible people at LG AI Research. In fact, I can go to anyone in the company for research advice because the people we work with are so good at research. I feel pride when I talk to these brilliant people and grow my research skills. I feel like I’m growing in the process.

Heuiyeen Yeen : I also remember feeling proud of myself. When I first joined the company, the researchers seemed so amazing that I thought, “Could I ever get to their level?” After a while, I found myself at a meeting sharing insights with attendees based on my own research experiences. I thought, “Oh, I have my own research approach. I’m growing.”

What “Occupational Hazards” Are Unique to AI Researchers?

Dongmin Kim : Now that I think about it, I’ve developed an “occupational hazard” since becoming a researcher. I think all engineers are like this. When I see something new, I try to imagine how it was made. For example, remember the “red light, green light” scene in Squid Game? In that scene, when people move, the machine automatically aims at and shoots them.Everybody was saying that it was brutal and scary. Meanwhile, I was watching it, and I was like, “Woah, that’s some really good object detection.”

Changho Lee : I’ve had a lot of experiences like that too. I guess we’re all similar!

Hyunmin Lee : I keep calling exercise machines “machine learning.” I should be saying “I worked out on the exercise machine for an hour,” but I keep saying that I worked out on the “machine learning” for an hour.

Heuiyeen Yeen : I keep calling exercise machines “machine learning.” I should be saying “I worked out on the exercise machine for an hour,” but I keep saying that I worked out on the “machine learning” for an hour.

Messages from the Current Juniors to the Next Juniors

Dongmin Kim : It’s great to work from home, but I’d love to see you in the office too. You’ll learn a lot more.

Changho Lee : Get ready to give it your all! Don’t worry, we’ll take care of you!

Hyunmin Lee : It’s okay to make mistakes, so be brave and try.

Heuiyeen Yeen : You have to be as cute as me to be the next junior member!

▶ Introducing Juniors at LG AI Research!