AEL AI Experience Lab

About

Welcome to the AI Experience Lab (AEL) founded in 2021 by prof. Tak Yeon Lee at the Industrial Design department, KAIST. We create innovation solutions for real-world problems by integrating the power of design, data, and AI-technology.

Data-Centric Approach

Digital transformation is fundamentally changing how products and services operate and deliver value to people. Data now has become the lens to understand people, and the oil that runs AI-powered machines. AEL's primary goal is to create innovative yet practical solutions for challenging problems. To achieve the goal, we leverage individual researchers' creative and constructive mindsets for finding insights from data, and building prototypes with data and AI technologies.

Pioneering the real-world applications of Generative AI

Generative AI models such as LLM (Large Language Model) or Image Models have been a hot topic in the research community. However, the real-world applications of generative AI models are still limited. We are interested in exploring the real-world applications of generative AI models, and developing novel design methods to leverage the power of generative AI models.

Pursuing the next design paradigm

The world is evolving into a giant network of human/non-human stakeholders. It gets harder and harder for designers to satisfy individual end-users. While holding the basic principles of HCD, we treat end-users with the same level of importance as other nodes such as other user groups, businesses, and even AI models. That being said, we try to mediate interests and tensions within the network rather than to satisfy a specific target users.

Who We Are

Current Members

  • Tak Yeon, Lee
    Assistant Professor
    website
  • Seon Gyeom, Kim
    PhD Student
  • HyunSeung, Moon
    Master Student
  • JaeYoung, Choi
    Master Student
  • Hyun, Lee
    Master Student
  • Jaeryung, Chung
    Master Student

Alumni

  • Juhyeong, Park
    MSc.
    UX Designer @ Hyundai Card
  • Byoungjae, Kim
    MSc.
    PhD Student @ KAIST
  • Yoosang, Yoon
    MSc.
    UX Researcher
  • Minsun, Kim
    MSc.
    Job Searching
  • Jin, Jeong
    Master Student

Life @ AEL

How to Join

We are always looking for students interested in AI-infused products and data-driven design. If you're interested in working with me, please read below and send me an email (takyeonlee at kaist dot ac dot kr).

Ph.D. student

An ideal candidate should have some prior research experience and technical skills such as prototyping or machine learning. If you are interested in, please send me an email containing some of the following information,

  • Cover letter that summarizes your experience and research interest
  • A link to materials that demonstrate your technical capabilities such as source code repositories, a personal website, etc
  • List of publications
  • Any other information that demonstrates your potential as a researcher

Master student

In KAIST, master’s program is research-oriented, which means that every student needs to join a research group and writes a thesis. In fact, after admission, students are allowed to select their preferred lab, and the department makes the best effort to accommodate these requests (though acceptance to a specific lab is not guaranteed, as it depends on the number of applications received).

Undergraduate Internship

If you are currently registered as a KAIST student (including a visiting student) and are interested in an internship, independent research in our lab or in starting a URP (Undergraduate Research Program) with me, please send an email with the following information,

  • What do you want to achieve at the end of your internship?
  • What research topic are you interested in?
  • How long will your internship be? I prefer students to work with me at least for six months, but it's not a hard requirement.
  • Any materials that demonstrate your expertise and capabilities such as design brief, video of working prototype, etc

Courses

ID403 System Design    Spring 2022 -

ID403 System Design completes the undergraduate curriculum of the Industrial Design department. The entire course focuses on conducting a design project where each project group consists of 4-5 students. We assume that students have equipped most (if not all) design skills and knowledge for conducting user research, visual communication, building low and high-fidelity prototypes, and evaluate design outcomes. Historically the course had a specific theme for each year. We expect the final outcomes will not only look great but also provide innovative solutions and/or intriguing (or even provocative) messages to the audience. There is no predefined set of lectures, but we will have several sessions for sharing ideas and feedbacks. While we have no plan to get feedback from external audiences, each group will weekly present their progress to the instructor, teaching assistants, and other groups - and get feedback from them.

2022. Home for Future Family  Supported by TaeJae Foundation  Term Projects

2023. NAVER for Generation Z  Supported by NAVER  Term Projects

ID408B Data Analytics for Designers    Fall 2021 -

Design is often considered as an art domain, which excludes objective data-driven analytic reasonings. Such dichotomy is inadequate in the age of data where most products and services are directly / indirectly connected through digital network. Data-driven design is a set of techniques to better understand the problem (e.g. what people want, how people behave), and employ data-driven functionalities to create innovative solutions. Failing to use data at all / properly may end up creating products that are useful for only a fraction of people, overlooking critical signals from people's behavior. In this course students will (1) learn basic concepts of data, (2) get their hands dirty while cleaning, transform, analyze, and visualizing data. Since knowing the dark side of data-driven design is another goal of this course, students will also learn (3) common limitations of data such as bias, ambiguity, sparsity, insufficiency, and how to fight against them.

ID503 Design Project 1    2021 Spring

How can we design products for the age of data? The goal of this course is to learn basic concepts, potentials, and limitations of data and AI, and to exercise developing AI-infused data-driven products.

Projects

Data Visualization for Alleviating Stress of Emotional Labor Over the Phone
AEL in collaboration with many labs @ KAIST, Yonsei University, and industry.
2022 -
As part of a large research team, AEL has been working on an IITP-funded project. The project's goal is to alleviate the stress of emotional workers at call centers. Our team's responsibility is to visualize digital twin of emotional workers' work-related stress factors.
Future Home for Family
AEL in collaboration with many labs @ KAIST.
2022 -
As part of a large research team at KAIST, AEL has been working on how future home should support family members. During the first year the team has led the System Design class, and created some interesting design concepts.
Interactive Gender Estimation of Human Skull
SeungHo Baek, and Tak Yeon lee.
March 2022 - November 2022
Gender estimation is the first step when an unidentified skull is found. Gender estimation is currently done by forensic experts. However, we developed a novel web appication that allows non-expert users to estimate gender of any skull with only three images. We trained a computer vision model from rendered images of 800 3D skull models, which exceeds state-of-the-art
AI-supported Tools for Authoring Immersive Data Storytelling
Seon Gyeom Kim, and Tak Yeon lee.
March 2021 -
This project aims to build an AI-assisted authoring tool of immersive data storytelling (IDS). Our tool focuses on maximizing the following benefits of IDS: (1) Meaningful and engaging composition of charts in 3D space, (2) Embodied interaction between presenter and charts, (3) Enabling collaborative data exploration in a virtual space. Our tool makes IDS authoring easier and more effective by automating low-level specifications and recommending semantically meaningful chart arrangements.
Interactive Dashboard for High-Speed Train Safety Management
March 2021 - September 2021
Sensor data is a critical source of information for high-speed train safety management. However, it is not easy to understand the overall status of the train from the raw sensor data. This project aims to design, implement, and deliver an interactive dashboard for train operators. Operators can use the dashboard to understand the overall status of the train, and to quickly identify abnormal situations.
MaeHwaSoo - Interactive Curriculum Explorer for College Students
Minwoo Kim. Guided by Seok-Hyoung Bae, Tak Yeon lee.
June 2021 -
College students regularly spend time on checking their progress toward graduation. MaeHwaSoo is a web-based interactive platform for students to quickly check whether they are ready to graduate, and how many / what classes they need to take until graduation. This project was originally initiated by a undergraduate student Minwoo Kim, and now became an open-source project maintained and improved by other contributors in KAIST.Visit the site

Publications

2024

  • RECIPE4U: Student-ChatGPT Interaction Dataset in EFL Writing Education
    Jieun Han, Haneul Yoo, Junho Myung, Minsun Kim, Tak Yeon Lee, So-Yeon Ahn and Alice Oh
    Accepted for LREC-COLING 2024
  • Virfie: Enhancing Remote Togetherness with User-Created Scenarios for Virtual Group Selfie
    Hyerin Im, Taewan Kim, Eunhee Jung, Bonhee Ku, Seungho Baek, Youn-kyung Lim, Tek-Jin Nam, and Tak Yeon Lee
    Accepted for Intelligent Systems Conference (IntelliSys) 2024, published in Lecture Notes in Networks and Systems

2023

  • GUIDE for GAIED: Exploring Student-ChatGPT Dialogue in EFL Writing Education
    Jieun Han, Haneul Yoo, Junho Myung, Minsun Kim, Tak Yeon Lee, So-Yeon Ahn, and Alice Oh
    GAIED Workshop at NeurIPS 2023[Extended Abstract]
  • Visual Insight Recommendation: From Ranking Insight Visualizations to Insight Types
    Camille Harris, Ryan Rossi, Sana Malik, Jane Hoffswell, Fan Du, Tak Yeon Lee, Eunyee Koh, and Handong Zhao
    2023 IEEE International Conference on Big Data Industry and Government Program[Full Paper]
  • SpotLight: Visual Insight Recommendation
    Camille Harris, Ryan Rossi, Sana Malik, Jane Hoffswell, Fan Du, Tak Yeon Lee, Eunyee Koh, and Handong Zhao.
    WWW '23 Companion: Companion Proceedings of the ACM Web Conference 2023[Full Paper]
  • RECIPE: How to Integrate ChatGPT into EFL Writing Education
    Jieun Han, Haneul Yoo, Yoonsu Kim, Junho Myung, Minsun Kim, Hyunseung Lim, Juho Kim, Tak Yeon Lee, Hwajung Hong, So-Yeon Ahn, and Alice Oh
    L@S '23: Proceedings of the Tenth ACM Conference on Learning @ Scale[Short Paper]

2022

  • Personalized visualization recommendation
    Xin Qian, Ryan A Rossi, Fan Du, Sungchul Kim, Eunyee Koh, Sana Malik, Tak Yeon Lee, Nesreen K Ahmed
    ACM Transactions on the Web (TWEB)[Full Paper]
  • Is It Really Useful?: An Observation Study of How Designers Use CLIP-based Image Generation For Moodboards
    Seungho Baek, Hyerin Im, Uran Oh, Youn-kyung Lim, and Tak Yeon Lee
    HCAI Workshop at NeurIPS 2022[Workshop]
  • Virfie: Virtual Group Selfie Station for Remote Togetherness
    Hyerin Im, Taewan Kim, Eunhee Jung, Bonhee Ku, Seungho Baek, and Tak Yeon Lee
    In Extended Abstracts of the 2022 CHI Conference on Human Factors in Computing Systems (CHI EA '22)[Extended Abstract]

2021

  • An Evaluation-Focused Framework for Visualization Recommendation Algorithms
    Zehua Zeng, Phoebe Moh, Fan Du, Jane Hoffswell, Tak Yeon Lee, Sana Malik, Eunyee Koh, Leilani Battle
    Proceedings of the 32nd IEEE VIS 2021, Visualization ConferenceHonorable Mention[Full Paper]
  • Generating Accurate Caption Units for Figure Captioning
    Xin Qian, Eunyee Koh, Fan Du, Sungchul Kim, Joel Chan, Ryan A Rossi, Sana Malik, Tak Yeon Lee
    Proceedings of the Web Conference 2021 (WWW '21) Association for Computing Machinery, New York, NY, USA, 2792–2804. DOI:https://doi.org/10.1145/3442381.3449923[Full Paper]
  • EXACTA: Explainable Column Annotation
    Yikun Xian, Handong Zhao, Tak Yeon Lee, Sungchul Kim, Ryan Rossi, Zuohui Fu, Gerard de Melo and S. Muthukrishnan
    KDD '21: Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining (ACM) Association for Computing Machinery, New York, NY, USA, 2792–2804.[Full Paper]
  • Learning to Recommend Visualizations from Data
    Xin Qian, Ryan Rossi, Fan Du, Sungchul Kim, Eunyee Koh, Sana Malik, Tak Yeon Lee, and Joel Chan
    KDD '21: Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining (ACM) Association for Computing Machinery, New York, NY, USA, 2792–2804.[Full Paper]

Prior to 2021

Check Tak Yeon Lee's Google Scholar page for publications before 2021