Breaking the stereotype of 'AI≠all-round solution' - AI content developer Kim Chang-hwan
Q. Hello. Mr. Changhwan. Please introduce yourself.
I’m Changhwan Kim, an AI content developer at Elice. I graduated from KAIST with an undergraduate degree in electronics and a master’s degree in knowledge service engineering, and then joined Elice.
Q. Mr. Changhwan, you joined Elice after graduating from graduate school, please tell us about your career path. Why did you choose to pursue a career as an A.I. content developer?
I’ve been interested in education since my master’s program. I was part of an artificial intelligence lab, where students managed their own GPU servers and needed computing knowledge. My undergraduate major was electronics, so I had a lot of opportunities to explain things to others, and I found teaching rewarding.
I had worked at Elice as a content creator or educator during my undergraduate and master’s degrees, and I was impressed by the collaborative, responsible, and passionate way that Eliceians work, so it was a natural decision to join the team.
Q. Please describe your current role.
I create or modify content that is used as lecture materials in AI education. On the one hand, I’m in charge of the Elice track, which is a developer training course, or corporate training. I meet real learners and design the curriculum to meet their needs.
Q. What is the attraction of AI learning?
I always like to learn new things, so I studied AI. I started with a vague idea that it would be good to learn, but the more I studied, the more I enjoyed the process of creatively building models, and it gave me a great sense of personal fulfillment.
There are various machine learning and deep learning algorithms that have been published, but you have to keep learning because it is a field that can constantly create new formulas based on various data.
Q. What has been your most fulfilling work at Elice?
First and foremost, I am proud to see the improvement in learning satisfaction with the AI content I created. When I first started training, I used existing content, but when I used it in the field, I found that there were parts that were difficult to understand or needed to be supplemented. In order to understand new technologies, you need to understand the essential concepts. If you only know them individually, you have to study each new technology from scratch, and you can’t apply them properly. Therefore, I organized the content into a flow to understand the latest technologies based on these concepts. Through these efforts, I received positive responses and high satisfaction from learners. Through this experience, I also felt a sense of pride in contributing to the cultivation of AI talent, like a farmer who plants and grows wheat grains.
In the process of creating AI content, I learn more by myself. In fact, when studying, it is easy to miss the basically important parts, and there are not many materials that point out these parts, and it is difficult to find them. I always thought that it would be nice to have content that summarizes everything at once for learners to understand. In order to create such content, you need to be sure of yourself first, so there are many things that you learn while studying.
Q. What are some of your challenges in teaching AI?
Recently, there are more and more examples of AI being used in real life, such as video recommendation algorithms and consultations through chatbots, but there is still a lack of understanding of AI. The first thing most people think of when they think of AI is robots, so whenever I teach, I explain what exactly AI is. When I tell them that AI is not about building robots, it’s about using data to create models, they’re disappointed at first, but then they get excited when they see it learn automatically. The end goal is to help learners see AI for what it is, and to help them create complex models that think like the human brain.
The problem with training is that learners have different levels of understanding and learning about AI, so we take a test before training, analyze the results, and then train at an average level to get a sense of the overall level.
In addition, Elice operates an advanced CMS (Content Management System), Elice Library, which is constantly updated to meet the rapidly changing technology and environment. Just as you can download a verified app from a mobile app store and get regular updates, you can get the latest and greatest training courses with Elice Library, which is constantly updated.
All Elice content developers view learning data, adjust and manage content difficulty, update lab versions, and deliver the highest quality content using the Elice Library.
Q. Do you have any tips for teaching AI to beginners?
For beginners, I try to explain the concepts as easily as possible. I don’t approach them mathematically, but find and use analogies and terms that can be understood, and lower the psychological barrier to learning. In the case of racers who are training to become developers, they study AI after web development, so I use programming languages that they are relatively familiar with. For example, for the AI term "Vector," we explain in detail how to implement it as an Array in Python with examples. This helps them translate mathematical concepts and AI theories into code.
Q. AI technology is evolving very fast, from convolutional neural networks (CNNs) to recurrent neural networks (RNNs) to generative adversarial nets (GANs). How do you keep up with the latest trends?
I’m very interested in new technologies, so I try to keep up with them. The classic way is to study the latest papers, but realistically, it’s hard to do that while working, so I keep up with tech blogs from famous companies like Google, Meta, Microsoft, and Apple. This is because these global big tech companies invest heavily in research and release world-leading technologies, which are then detailed in their tech blogs.
I often visit a site called "Papers with Code". It’s a place where you can see papers and code together, and I use it when I want to see the implementation code of a prominent paper. There are also sites called Hacker News and Techmeme, which are very useful for keeping up with the most current issues in IT beyond development, so I often visit them to research the most recent technology trends or what companies are looking for in technology.
Q. Is there any content that you would like to develop in the future and why?"
The content I would like to develop in the future is a basic course on natural language processing (NLP) models. Natural language processing models are gaining a lot of attention with the emergence of AIs writing columns and AI announcers, but basic curriculum is lacking. In particular, Korean has much less content than English. The recently released *GPT model is also based on the Transformer model, which tracks relationships in data to learn context and meaning. Therefore, I think that in order to understand new natural language processing models, we need to solidify the basic concepts.
- GPT (Generative Pre-trained Transformer): AI language generation model, rated as the world’s strongest AI after AlphaGo, trained on 300 billion data sets and has 175 billion parameters.
Q. What qualities do you think are needed as an A.I content developer?
Since A.I. content developers are also experts in creating educational content, I’d like to focus on this aspect. The most important quality for a content developer is the ability to make it easy for anyone to understand. In that sense, it was a good experience for me to utilize the content I created myself, because it gave me a chance to think about how to meet the learners’ eyes while conducting training.
First and foremost, you need to have a deep understanding of AI. You need to know and study a lot so that you can make logical and smooth connections. I think you can feel confident and proud of the content you create. Furthermore, if you utilize language and expressions that suit the learners, I think effective training will be completed.
Q. What is your vision of an A.I. content developer?
AI is revolutionizing all fields and is being recognized as the future food industry, so I think the outlook will continue to be bright. Although Korea is leading the world in A.I., it is true that the core data and manpower are lacking, and the high-level manpower is also leaking overseas.
Therefore, AI education is being emphasized, and the growth potential for AI content developers is endless.
Q. Please tell us about your dreams for the future at Elice. What are your personal goals?
I am a person who believes that the direction of life is more important than big goals. I believe that if you live sincerely every day, good results will come. In terms of my dream direction, I want to expand the learning base by developing content that is well versed in A.I. Ultimately, I want to grow into an expert who can be thought of first when it comes to A.I. content.
Q. What would you like to say to someone who is applying to be an A.I. content developer?
I would like someone who looks at AI with a broad perspective. Unlike researchers who focus on one thing, content developers have to educate learners on various things. It would be even better if you study AI deeply enough to understand the logical flow, but don’t lose sight of the point of delivering education. In addition, someone who is interested in the latest technology and wants to learn while having fun is always welcome.
Technically, the Elice platform is very good. We developed the platform in-house, and once you use it, it’s so useful and convenient that it’s no different from offline training. In fact, I can teach a class with a lot of students and not have any interruptions at all. There are many companies that make their own products or services, but I think it’s hard to be confident that the quality is good. Therefore, I’m sure that the experience of creating success in a company with competitive technology and know-how will be a good foundation for personal career growth.
*This content is a work protected by copyright law and is copyrighted by Elice.
*The content is prohibited from secondary processing and commercial use without prior consent.