AI
-
KAIST Develops AI to Easily Find Promising Materials That Capture Only CO₂
< Photo 1. (From left) Professor Jihan Kim, Ph.D. candidate Yunsung Lim and Dr. Hyunsoo Park of the Department of Chemical and Biomolecular Engineering >
In order to help prevent the climate crisis, actively reducing already-emitted CO₂ is essential. Accordingly, direct air capture (DAC) — a technology that directly extracts only CO₂ from the air — is gaining attention. However, effectively capturing pure CO₂ is not easy due to water vapor (H₂O) present in the air. KAIST researchers have successfully used AI-driven machine learning techniques to identify the most promising CO₂-capturing materials among metal-organic frameworks (MOFs), a key class of materials studied for this technology.
KAIST (President Kwang Hyung Lee) announced on the 29th of June that a research team led by Professor Jihan Kim from the Department of Chemical and Biomolecular Engineering, in collaboration with a team at Imperial College London, has developed a machine-learning-based simulation method that can quickly and accurately screen MOFs best suited for atmospheric CO₂ capture.
< Figure 1. Concept diagram of Direct Air Capture (DAC) technology and carbon capture using Metal-Organic Frameworks (MOFs). MOFs are promising porous materials capable of capturing carbon dioxide from the atmosphere, drawing attention as a core material for DAC technology. >
To overcome the difficulty of discovering high-performance materials due to the complexity of structures and the limitations of predicting intermolecular interactions, the research team developed a machine learning force field (MLFF) capable of precisely predicting the interactions between CO₂, water (H₂O), and MOFs. This new method enables calculations of MOF adsorption properties with quantum-mechanics-level accuracy at vastly faster speeds than before.
Using this system, the team screened over 8,000 experimentally synthesized MOF structures, identifying more than 100 promising candidates for CO₂ capture. Notably, this included new candidates that had not been uncovered by traditional force-field-based simulations. The team also analyzed the relationships between MOF chemical structure and adsorption performance, proposing seven key chemical features that will help in designing new materials for DAC.
< Figure 2. Concept diagram of adsorption simulation using Machine Learning Force Field (MLFF). The developed MLFF is applicable to various MOF structures and allows for precise calculation of adsorption properties by predicting interaction energies during repetitive Widom insertion simulations. It is characterized by simultaneously achieving high accuracy and low computational cost compared to conventional classical force fields. >
This research is recognized as a significant advance in the DAC field, greatly enhancing materials design and simulation by precisely predicting MOF-CO₂ and MOF-H₂O interactions.
The results of this research, with Ph.D. candidate Yunsung Lim and Dr. Hyunsoo Park of KAIST as co-first authors, were published in the international academic journal Matter on June 12.
※Paper Title: Accelerating CO₂ direct air capture screening for metal–organic frameworks with a transferable machine learning force field
※DOI: 10.1016/j.matt.2025.102203
This research was supported by the Saudi Aramco-KAIST CO₂ Management Center and the Ministry of Science and ICT's Global C.L.E.A.N. Project.
2025.06.29 View 134 -
Military Combatants Usher in an Era of Personalized Training with New Materials
< Photo 1. (From left) Professor Steve Park of Materials Science and Engineering, Kyusoon Pak, Ph.D. Candidate (Army Major) >
Traditional military training often relies on standardized methods, which has limited the provision of optimized training tailored to individual combatants' characteristics or specific combat situations. To address this, our research team developed an e-textile platform, securing core technology that can reflect the unique traits of individual combatants and various combat scenarios. This technology has proven robust enough for battlefield use and is economical enough for widespread distribution to a large number of troops.
On June 25th, Professor Steve Park's research team at KAIST's Department of Materials Science and Engineering announced the development of a flexible, wearable electronic textile (E-textile) platform using an innovative technology that 'draws' electronic circuits directly onto fabric.
The wearable e-textile platform developed by the research team combines 3D printing technology with new materials engineering design to directly print flexible and highly durable sensors and electrodes onto textile substrates. This enables the collection of precise movement and human body data from individual combatants, which can then be used to propose customized training models.
Existing e-textile fabrication methods were often complex or limited in their ability to provide personalized customization. To overcome these challenges, the research team adopted an additive manufacturing technology called 'Direct Ink Writing (DIW)' 3D printing.
< Figure 1. Schematic diagram of e-textile manufactured with Direct Ink Writing (DIW) printing technology on various textiles, including combat uniforms >
This technology involves directly dispensing and printing special ink, which functions as sensors and electrodes, onto textile substrates in desired patterns. This allows for flexible implementation of various designs without the complex process of mask fabrication. This is expected to be an effective technology that can be easily supplied to hundreds of thousands of military personnel.
The core of this technology lies in the development of high-performance functional inks based on advanced materials engineering design. The research team combined styrene-butadiene-styrene (SBS) polymer, which provides flexibility, with multi-walled carbon nanotubes (MWCNT) for electrical conductivity. They developed a tensile/bending sensor ink that can stretch up to 102% and maintain stable performance even after 10,000 repetitive tests. This means that accurate data can be consistently obtained even during the strenuous movements of combatants.
< Figure 2. Measurement of human movement and breathing patterns using e-textile >
Furthermore, new material technology was applied to implement 'interconnect electrodes' that electrically connect the upper and lower layers of the fabric. The team developed an electrode ink combining silver (Ag) flakes with rigid polystyrene (PS) polymer, precisely controlling the impregnation level (how much the ink penetrates the fabric) to effectively connect both sides or multiple layers of the fabric. This secures the technology for producing multi-layered wearable electronic systems integrating sensors and electrodes.
< Figure 3. Experimental results of recognizing unknown objects after machine learning six objects using a smart glove >
The research team proved the platform's performance through actual human movement monitoring experiments. They printed the developed e-textile on major joint areas of clothing (shoulders, elbows, knees) and measured movements and posture changes during various exercises such as running, jumping jacks, and push-ups in real-time.
Additionally, they demonstrated the potential for applications such as monitoring breathing patterns using a smart mask and recognizing objects through machine learning and perceiving complex tactile information by printing multiple sensors and electrodes on gloves. These results show that the developed e-textile platform is effective in precisely understanding the movement dynamics of combatants.
This research is an important example demonstrating how cutting-edge new material technology can contribute to the advancement of the defense sector. Major Kyusoon Pak of the Army, who participated in this research, considered required objectives such as military applicability and economic feasibility for practical distribution from the research design stage.
< Figure 4. Experimental results showing that a multi-layered e-textile glove connected with interconnect electrodes can measure tensile/bending signals and pressure signals at a single point >
Major Pak stated, "Our military is currently facing both a crisis and an opportunity due to the decrease in military personnel resources caused by the demographic cliff and the advancement of science and technology. Also, respect for life in the battlefield is emerging as a significant issue. This research aims to secure original technology that can provide customized training according to military branch/duty and type of combat, thereby enhancing the combat power and ensuring the survivability of our soldiers."
He added, "I hope this research will be evaluated as a case that achieved both scientific contribution and military applicability."
This research, where Kyusoon Pak, Ph.D. Candidate (Army Major) from KAIST's Department of Materials Science and Engineering, participated as the first author and Professor Steve Park supervised, was published on May 27, 2025, in `npj Flexible Electronics (top 1.8% in JCR field)', an international academic journal in the electrical, electronic, and materials engineering fields.
* Paper Title: Fabrication of Multifunctional Wearable Interconnect E-textile Platform Using Direct Ink Writing (DIW) 3D Printing
* DOI: https://doi.org/10.1038/s41528-025-00414-7
This research was supported by the Ministry of Trade, Industry and Energy and the National Research Foundation of Korea.
2025.06.25 View 177 -
KAIST to Lead the Way in Nurturing Talent and Driving S&T Innovation for a G3 AI Powerhouse
* Focusing on nurturing talent and dedicating to R&D to become a G3 AI powerhouse (Top 3 AI Nations).
* Leading the realization of an "AI-driven Basic Society for All" and developing technologies that leverage AI to overcome the crisis in Korea's manufacturing sector.
* 50 years ago, South Korea emerged as a scientific and technological powerhouse from the ashes, with KAIST at its core, contributing to the development of scientific and technological talent, innovative technology, national industrial growth, and the creation of a startup innovation ecosystem.
As public interest in AI and science and technology has significantly grown with the inauguration of the new government, KAIST (President Kwang Hyung Lee) announced its plan, on June 24th, to transform into an "AI-centric, Value-Creating Science and Technology University" that leads national innovation based on science and technology and spearheads solutions to global challenges.
At a time when South Korea is undergoing a major transition to a technology-driven society, KAIST, drawing on its half-century of experience as a "Starter Kit" for national development, is preparing to leap beyond being a mere educational and research institution to become a global innovation hub that creates new social value.
In particular, KAIST has presented a vision for realizing an "AI-driven Basic Society" where all citizens can utilize AI without alienation, enabling South Korea to ascend to the top three AI nations (G3). To achieve this, through the "National AI Research Hub" project (headed by Kee Eung Kim), led by KAIST representing South Korea, the institution is dedicated to enhancing industrial competitiveness and effectively solving social problems based on AI technology.
< KAIST President Kwang Hyung Lee >
KAIST's research achievements in the AI field are garnering international attention. In the top three machine learning conferences (ICML, NeurIPS, ICLR), KAIST ranked 5th globally and 1st in Asia over the past five years (2020-2024). During the same period, based on the number of papers published in top conferences in machine learning, natural language processing, and computer vision (ICML, NeurIPS, ICLR, ACL, EMNLP, NAACL, CVPR, ICCV, ECCV), KAIST ranked 5th globally and 4th in Asia. Furthermore, KAIST has consistently demonstrated unparalleled research capabilities, ranking 1st globally in the average number of papers accepted at ISSCC (International Solid-State Circuits Conference), the world's most prestigious academic conference on semiconductor integrated circuits, for 19 years (2006-2024).
KAIST is continuously expanding its research into core AI technologies, including hyper-scale AI models (Korean LLM), neuromorphic semiconductors, and low-power AI processors, as well as various application areas such as autonomous driving, urban air mobility (UAM), precision medicine, and explainable AI (XAI).
In the manufacturing sector, KAIST's AI technologies are also driving on-site innovation. Professor Young Jae Jang's team has enhanced productivity in advanced manufacturing fields like semiconductors and displays through digital twins utilizing manufacturing site data and AI-based prediction technology. Professor Song Min Kim's team developed ultra-low power wireless tag technology capable of tracking locations with sub-centimeter precision, accelerating the implementation of smart factories. Technologies such as industrial process optimization and equipment failure prediction developed by INEEJI Co., Ltd., founded by Professor Jaesik Choi, are being rapidly applied in real industrial settings, yielding results. INEEJI was designated as a national strategic technology in the 'Explainable AI (XAI)' field by the government in March.
< Researchers performing data analysis for AI research >
Practical applications are also emerging in the robotics sector, which is closely linked to AI. Professor Jemin Hwangbo's team from the Department of Mechanical Engineering garnered attention by newly developing RAIBO 2, a quadrupedal robot usable in high-risk environments such as disaster relief and rough terrain exploration. Professor Kyoung Chul Kong's team and Angel Robotics Co., Ltd. developed the WalkOn Suit exoskeleton robot, significantly improving the quality of life for individuals with complete lower body paralysis or walking disabilities.
Additionally, remarkable research is ongoing in future core technology areas such as AI semiconductors, quantum cryptography communication, ultra-small satellites, hydrogen fuel cells, next-generation batteries, and biomimetic sensors. Notably, space exploration technology based on small satellites, asteroid exploration projects, energy harvesting, and high-speed charging technologies are gaining attention.
Particularly in advanced bio and life sciences, KAIST is collaborating with Germany's Merck company on various research initiatives, including synthetic biology and mRNA. KAIST is also contributing to the construction of a 430 billion won Merck Bio-Center in Daejeon, thereby stimulating the local economy and creating jobs.
Based on these cutting-edge research capabilities, KAIST continues to expand its influence not only within the industry but also on the global stage. It has established strategic partnerships with leading universities worldwide, including MIT, Stanford University, and New York University (NYU). Notably, KAIST and NYU have established a joint campus in New York to strengthen human exchange and collaborative research. Active industry-academia collaborations with global companies such as Google, Intel, and TSMC are also ongoing, playing a pivotal role in future technology development and the creation of an innovation ecosystem.
These activities also lead to a strong startup ecosystem that drives South Korean industries. The flow of startups, which began with companies like Qnix Computer, Nexon, and Naver, has expanded to a total of 1,914 companies to date. Their cumulative assets amount to 94 trillion won, with sales reaching 36 trillion won and employing approximately 60,000 people. Over 90% of these are technology-based startups originating from faculty and student labs, demonstrating a model that makes a tangible economic contribution based on science and technology.
< Students at work >
Having consistently generated diverse achievements, KAIST has already produced approximately 80,000 "KAISTians" who have created innovation through challenge and failure, and is currently recruiting new talent to continue driving innovation that transforms South Korea and the world.
President Kwang Hyung Lee emphasized, "KAIST will establish itself as a global leader in science and technology, designing the future of South Korea and humanity and creating tangible value." He added, "We will focus on talent nurturing and research and development to realize the new government's national agenda of becoming a G3 AI powerhouse."
He further stated, "KAIST's vision for the AI field, in which it places particular emphasis, is to strive for a society where everyone can freely utilize AI. We will contribute to significantly boosting productivity by recovering manufacturing competitiveness through AI and actively disseminating physical AI, AI robots, and AI mobility technologies to industrial sites."
2025.06.24 View 502 -
KAIST Researchers Unveil an AI that Generates "Unexpectedly Original" Designs
< Photo 1. Professor Jaesik Choi, KAIST Kim Jaechul Graduate School of AI >
Recently, text-based image generation models can automatically create high-resolution, high-quality images solely from natural language descriptions. However, when a typical example like the Stable Diffusion model is given the text "creative," its ability to generate truly creative images remains limited. KAIST researchers have developed a technology that can enhance the creativity of text-based image generation models such as Stable Diffusion without additional training, allowing AI to draw creative chair designs that are far from ordinary.
Professor Jaesik Choi's research team at KAIST Kim Jaechul Graduate School of AI, in collaboration with NAVER AI Lab, developed this technology to enhance the creative generation of AI generative models without the need for additional training.
< Photo 2. Gayoung Lee, Researcher at NAVER AI Lab; Dahee Kwon, Ph.D. Candidate at KAIST Kim Jaechul Graduate School of AI; Jiyeon Han, Ph.D. Candidate at KAIST Kim Jaechul Graduate School of AI; Junho Kim, Researcher at NAVER AI Lab >
Professor Choi's research team developed a technology to enhance creative generation by amplifying the internal feature maps of text-based image generation models. They also discovered that shallow blocks within the model play a crucial role in creative generation. They confirmed that amplifying values in the high-frequency region after converting feature maps to the frequency domain can lead to noise or fragmented color patterns. Accordingly, the research team demonstrated that amplifying the low-frequency region of shallow blocks can effectively enhance creative generation.
Considering originality and usefulness as two key elements defining creativity, the research team proposed an algorithm that automatically selects the optimal amplification value for each block within the generative model.
Through the developed algorithm, appropriate amplification of the internal feature maps of a pre-trained Stable Diffusion model was able to enhance creative generation without additional classification data or training.
< Figure 1. Overview of the methodology researched by the development team. After converting the internal feature map of a pre-trained generative model into the frequency domain through Fast Fourier Transform, the low-frequency region of the feature map is amplified, then re-transformed into the feature space via Inverse Fast Fourier Transform to generate an image. >
The research team quantitatively proved, using various metrics, that their developed algorithm can generate images that are more novel than those from existing models, without significantly compromising utility.
In particular, they confirmed an increase in image diversity by mitigating the mode collapse problem that occurs in the SDXL-Turbo model, which was developed to significantly improve the image generation speed of the Stable Diffusion XL (SDXL) model. Furthermore, user studies showed that human evaluation also confirmed a significant improvement in novelty relative to utility compared to existing methods.
Jiyeon Han and Dahee Kwon, Ph.D. candidates at KAIST and co-first authors of the paper, stated, "This is the first methodology to enhance the creative generation of generative models without new training or fine-tuning. We have shown that the latent creativity within trained AI generative models can be enhanced through feature map manipulation."
They added, "This research makes it easy to generate creative images using only text from existing trained models. It is expected to provide new inspiration in various fields, such as creative product design, and contribute to the practical and useful application of AI models in the creative ecosystem."
< Figure 2. Application examples of the methodology researched by the development team. Various Stable Diffusion models generate novel images compared to existing generations while maintaining the meaning of the generated object. >
This research, co-authored by Jiyeon Han and Dahee Kwon, Ph.D. candidates at KAIST Kim Jaechul Graduate School of AI, was presented on June 16 at the International Conference on Computer Vision and Pattern Recognition (CVPR), an international academic conference.* Paper Title: Enhancing Creative Generation on Stable Diffusion-based Models* DOI: https://doi.org/10.48550/arXiv.2503.23538
This research was supported by the KAIST-NAVER Ultra-creative AI Research Center, the Innovation Growth Engine Project Explainable AI, the AI Research Hub Project, and research on flexible evolving AI technology development in line with increasingly strengthened ethical policies, all funded by the Ministry of Science and ICT through the Institute for Information & Communications Technology Promotion. It also received support from the KAIST AI Graduate School Program and was carried out at the KAIST Future Defense AI Specialized Research Center with support from the Defense Acquisition Program Administration and the Agency for Defense Development.
2025.06.20 View 580 -
‘InnoCORE Research Group’ Launched to Lead AI Convergence Innovation
KAIST announced on the 16th of June that it has launched the ‘InnoCORE (Innovation-Core) Research Group,’ which will lead advanced strategic research in AI convergence (AI+S&T), in cooperation with the Ministry of Science and ICT (Minister Yoo Sang-im, hereinafter referred to as MSIT) and DGIST, GIST, and UNIST*. Through this, the group plans to actively recruit up to 200 world-class postdoctoral researchers.
DGIST (Daegu Gyeongbuk Institute of Science & Technology), GIST (Gwangju Institute of Science & Technology), UNIST (Ulsan National Institute of Science and Technology)
The ‘InnoCORE Research Group’ aims to foster core research personnel who will lead innovation in the field of AI convergence, focusing on nurturing and attracting high-level research talent in AI+Science & Technology. This is a strategic response to prevent brain drain of domestic talent and attract excellent overseas talent amidst the accelerating global competition for AI talent.
Through this initiative, our university plans to accelerate AI-based science and technology innovation and disseminate research achievements across industries and the economy by supporting top domestic and international postdoctoral researchers to dedicate themselves to developing AI convergence technologies in an advanced collaborative research environment.
The InnoCORE project for advanced AI+S&T convergence research and global talent attraction is jointly promoted by four science and technology institutes, including KAIST. It is structured around AI core technologies (such as hyper-scale language models, AI semiconductors) and AI convergence technologies (such as bio, manufacturing, energy, and aerospace).
As the leading institution, our university operates the following four research groups:
Hyper-scale Language Model Innovation Research Group: Advancement of LLM technology and research on generative AI, multimodal AI, and ensuring reliability.
AI-based Intelligent Design-Manufacturing Integration Research Group: Establishment of an AI platform for the entire lifecycle of the manufacturing industry and innovation in design and processes.
AI-Innovation Drug Research Group: Securing AI-based drug development technologies across the entire lifecycle and overcoming intractable diseases.
AI-Transformed Aerospace Research Group: AI transformation of aerospace systems throughout their lifecycle and development of new technologies such as autonomous flight and space communication.
< Poster on the InnoCORE Global Jobfair for Recruitment of Postdoctoral Researchers >
In addition, a total of eight research groups are formed to promote global collaborative convergence research, including those led by DGIST, GIST, and UNIST: ▲Bio-Integrated Physical AI, ▲Early Diagnosis of Brain Diseases AI+Nano Convergence, ▲Intelligent Hydrogen Technology Innovation, and ▲AI-Space Solar Power Research Group.
Starting in 2025, the four science and technology institutes, including KAIST, will officially begin recruiting 400 postdoctoral researchers in the AI+S&T fields. Selected postdoctoral researchers will be guaranteed high-level treatment with an annual salary of over 90 million KRW, and additional support through matching with companies and research projects is also planned.
In particular, global recruitment fairs will be held in major US regions to expand the attraction of excellent overseas talent. Local recruitment fairs will be held in Boston (Harvard, MIT), New York (NYU), and Silicon Valley (Stanford) in June, along with promotions through global academic journals such as Nature and Science, and LinkedIn.
KAIST plans to provide multiple mentor programs, global joint research opportunities, and excellent infrastructure (such as supercomputers, semiconductor fabs, and AI research platforms) within the research groups to enable postdoctoral researchers to collaborate with experts from various academic and industrial fields.
President Kwang Hyung Lee emphasized, “Through this InnoCORE project, KAIST will leap forward as a Global Hub for AI+S&T convergence research. Young researchers from around the world will challenge themselves and grow at KAIST, and our country will play a pivotal role in establishing itself as a leading nation in global AI convergence research and industry. To achieve this, we will spare no effort in providing the best research environment and active support.”
KAIST plans to actively pursue the InnoCORE project to secure global competitiveness in AI convergence research and contribute to the development of advanced industries. The eight selected research groups will finalize their detailed research plans by the end of June and commence full-scale research in July.
2025.06.19 View 807 -
Simultaneous Analysis of 21 Chemical Reactions... AI to Transform New Drug Development
< Photo 1. (From left) Professor Hyunwoo Kim and students Donghun Kim and Gyeongseon Choi in the Integrated M.S./Ph.D. program of the Department of Chemistry >
Thalidomide, a drug once used to alleviate morning sickness in pregnant women, exhibits distinct properties due to its optical isomers* in the body: one isomer has a sedative effect, while the other causes severe side effects like birth defects. As this example illustrates, precise organic synthesis techniques, which selectively synthesize only the desired optical isomer, are crucial in new drug development. Overcoming the traditional methods that struggled with simultaneously analyzing multiple reactants, our research team has developed the world's first technology to precisely analyze 21 types of reactants simultaneously. This breakthrough is expected to make a significant contribution to new drug development utilizing AI and robots.
*Optical Isomers: A pair of molecules with the same chemical formula that are mirror images of each other and cannot be superimposed due to their asymmetric structure. This is analogous to a left and right hand, which are similar in form but cannot be perfectly overlaid.
KAIST's Professor Hyunwoo Kim's research team in the Department of Chemistry announced on the 16th that they have developed an innovative optical isomer analysis technology suitable for the era of AI-driven autonomous synthesis*. This research is the world's first technology to precisely analyze asymmetric catalytic reactions involving multiple reactants simultaneously using high-resolution fluorine nuclear magnetic resonance spectroscopy (19F NMR). It is expected to make groundbreaking contributions to various fields, including new drug development and catalyst optimization.
*AI-driven Autonomous Synthesis: An advanced technology that automates and optimizes chemical substance synthesis processes using artificial intelligence (AI). It is gaining attention as a core element for realizing automated and intelligent research environments in future laboratories. AI predicts and adjusts experimental conditions, interprets results, and designs subsequent experiments independently, minimizing human intervention in repetitive experiments and significantly increasing research efficiency and innovativeness.
Currently, while autonomous synthesis systems can automate everything from reaction design to execution, reaction analysis still relies on individual processing using traditional equipment. This leads to slower speeds and bottlenecks, making it unsuitable for high-speed repetitive experiments.
Furthermore, multi-substrate simultaneous screening techniques proposed in the 1990s garnered attention as a strategy to maximize reaction analysis efficiency. However, limitations of existing chromatography-based analysis methods restricted the number of applicable substrates. In asymmetric synthesis reactions, which selectively synthesize only the desired optical isomer, simultaneously analyzing more than 10 types of substrates was nearly impossible.
< Figure 1. Conventional organic reaction evaluation methods follow a process of deriving optimal reaction conditions using a single substrate, then expanding the substrate scope one by one under those conditions, leaving potential reaction areas unexplored. To overcome this, high-throughput screening is introduced to broadly explore catalyst reactivity for various substrates. When combined with multi-substrate screening, this approach allows for a much broader and more systematic understanding of reaction scope and trends. >
To overcome these limitations, the research team developed a 19F NMR-based multi-substrate simultaneous screening technology. This method involves performing asymmetric catalytic reactions with multiple reactants in a single reaction vessel, introducing a fluorine functional group into the products, and then applying their self-developed chiral cobalt reagent to clearly quantify all optical isomers using 19F NMR.
Utilizing the excellent resolution and sensitivity of 19F NMR, the research team successfully performed asymmetric synthesis reactions of 21 substrates simultaneously in a single reaction vessel and quantitatively measured the product yield and optical isomer ratio without any separate purification steps.
Professor Hyunwoo Kim stated, "While anyone can perform asymmetric synthesis reactions with multiple substrates in one reactor, accurately analyzing all the products has been a challenging problem to solve until now. We expect that achieving world-class multi-substrate screening analysis technology will greatly contribute to enhancing the analytical capabilities of AI-driven autonomous synthesis platforms."
< Figure 2. A method for analyzing multi-substrate asymmetric catalytic reactions, where different substrates react simultaneously in a single reactor, using fluorine nuclear magnetic resonance has been implemented. By utilizing the characteristics of fluorine nuclear magnetic resonance, which has a clean background signal and a wide chemical shift range, the reactivity of each substrate can be quantitatively analyzed. It is also shown that the optical activity of all reactants can be simultaneously measured using a cobalt metal complex. >
He further added, "This research provides a technology that can rapidly verify the efficiency and selectivity of asymmetric catalytic reactions essential for new drug development, and it is expected to be utilized as a core analytical tool for AI-driven autonomous research."
< Figure 3. It can be seen that in a multi-substrate reductive amination reaction using a total of 21 substrates, the yield and optical activity of the reactants according to the catalyst system were simultaneously measured using a fluorine nuclear magnetic resonance-based analysis platform. The yield of each reactant is indicated by color saturation, and the optical activity by numbers. >
Donghun Kim (first author, Integrated M.S./Ph.D. program) and Gyeongseon Choi (second author, Integrated M.S./Ph.D. program) from the KAIST Department of Chemistry participated in this research. The study was published online in the Journal of the American Chemical Society on May 27, 2025.※ Paper Title: One-pot Multisubstrate Screening for Asymmetric Catalysis Enabled by 19F NMR-based Simultaneous Chiral Analysis※ DOI: 10.1021/jacs.5c03446
This research was supported by the National Research Foundation of Korea's Mid-Career Researcher Program, the Asymmetric Catalytic Reaction Design Center, and the KAIST KC30 Project.
< Figure 4. Conceptual diagram of performing multi-substrate screening reactions and utilizing fluorine nuclear magnetic resonance spectroscopy. >
2025.06.16 View 1068 -
KAIST Professor Jee-Hwan Ryu Receives Global IEEE Robotics Journal Best Paper Award
- Professor Jee-Hwan Ryu of Civil and Environmental Engineering receives the Best Paper Award from the Institute of Electrical and Electronics Engineers (IEEE) Robotics Journal, officially presented at ICRA, a world-renowned robotics conference.
- This is the highest level of international recognition, awarded to only the top 5 papers out of approximately 1,500 published in 2024.
- Securing a new working channel technology for soft growing robots expands the practicality and application possibilities in the field of soft robotics.
< Professor Jee-Hwan Ryu (left), Nam Gyun Kim, Ph.D. Candidate (right) from the KAIST Department of Civil and Environmental Engineering and KAIST Robotics Program >
KAIST (President Kwang-Hyung Lee) announced on the 6th that Professor Jee-Hwan Ryu from the Department of Civil and Environmental Engineering received the 2024 Best Paper Award from the Robotics and Automation Letters (RA-L), a premier journal under the IEEE, at the '2025 IEEE International Conference on Robotics and Automation (ICRA)' held in Atlanta, USA, on May 22nd.
This Best Paper Award is a prestigious honor presented to only the top 5 papers out of approximately 1,500 published in 2024, boasting high international competition and authority.
The award-winning paper by Professor Ryu proposes a novel working channel securing mechanism that significantly expands the practicality and application possibilities of 'Soft Growing Robots,' which are based on soft materials that move or perform tasks through a growing motion similar to plant roots.
< IEEE Robotics Journal Award Ceremony >
Existing soft growing robots move by inflating or contracting their bodies through increasing or decreasing internal pressure, which can lead to blockages in their internal passages. In contrast, the newly developed soft growing robot achieves a growing function while maintaining the internal passage pressure equal to the external atmospheric pressure, thereby successfully securing an internal passage while retaining the robot's flexible and soft characteristics.
This structure allows various materials or tools to be freely delivered through the internal passage (working channel) within the robot and offers the advantage of performing multi-purpose tasks by flexibly replacing equipment according to the working environment.
The research team fabricated a prototype to prove the effectiveness of this technology and verified its performance through various experiments. Specifically, in the slide plate experiment, they confirmed whether materials or equipment could pass through the robot's internal channel without obstruction, and in the pipe pulling experiment, they verified if a long pipe-shaped tool could be pulled through the internal channel.
< Figure 1. Overall hardware structure of the proposed soft growing robot (left) and a cross-sectional view composing the inflatable structure (right) >
Experimental results demonstrated that the internal channel remained stable even while the robot was growing, serving as a key basis for supporting the technology's practicality and scalability.
Professor Jee-Hwan Ryu stated, "This award is very meaningful as it signifies the global recognition of Korea's robotics technology and academic achievements. Especially, it holds great significance in achieving technical progress that can greatly expand the practicality and application fields of soft growing robots. This achievement was possible thanks to the dedication and collaboration of the research team, and I will continue to contribute to the development of robotics technology through innovative research."
< Figure 2. Material supplying mechanism of the Soft Growing Robot >
This research was co-authored by Dongoh Seo, Ph.D. Candidate in Civil and Environmental Engineering, and Nam Gyun Kim, Ph.D. Candidate in Robotics. It was published in IEEE Robotics and Automation Letters on September 1, 2024.
(Paper Title: Inflatable-Structure-Based Working-Channel Securing Mechanism for Soft Growing Robots, DOI: 10.1109/LRA.2024.3426322)
This project was supported simultaneously by the National Research Foundation of Korea's Future Promising Convergence Technology Pioneer Research Project and Mid-career Researcher Project.
2025.06.09 View 1549 -
KAIST Introduces ‘Virtual Teaching Assistant’ That can Answer Even in the Middle of the Night – Successful First Deployment in Classroom
- Research teams led by Prof. Yoonjae Choi (Kim Jaechul Graduate School of AI) and Prof. Hwajeong Hong (Department of Industrial Design) at KAIST developed a Virtual Teaching Assistant (VTA) to support learning and class operations for a course with 477 students.
- The VTA responds 24/7 to students’ questions related to theory and practice by referencing lecture slides, coding assignments, and lecture videos.
- The system’s source code has been released to support future development of personalized learning support systems and their application in educational settings.
< Photo 1. (From left) PhD candidate Sunjun Kweon, Master's candidate Sooyohn Nam, PhD candidate Hyunseung Lim, Professor Hwajung Hong, Professor Yoonjae Choi >
“At first, I didn’t have high expectations for the Virtual Teaching Assistant (VTA), but it turned out to be extremely helpful—especially when I had sudden questions late at night, I could get immediate answers,” said Jiwon Yang, a Ph.D. student at KAIST. “I was also able to ask questions I would’ve hesitated to bring up with a human TA, which led me to ask even more and ultimately improved my understanding of the course.”
KAIST (President Kwang Hyung Lee) announced on June 5th that a joint research team led by Prof. Yoonjae Choi of the Kim Jaechul Graduate School of AI and Prof. Hwajeong Hong of the Department of Industrial Design has successfully developed and deployed a Virtual Teaching Assistant (VTA) that provides personalized feedback to individual students even in large-scale classes.
This study marks one of the first large-scale, real-world deployments in Korea, where the VTA was introduced in the “Programming for Artificial Intelligence” course at the KAIST Kim Jaechul Graduate School of AI, taken by 477 master’s and Ph.D. students during the Fall 2024 semester, to evaluate its effectiveness and practical applicability in an actual educational setting.
The AI teaching assistant developed in this study is a course-specialized agent, distinct from general-purpose tools like ChatGPT or conventional chatbots. The research team implemented a Retrieval-Augmented Generation (RAG) architecture, which automatically vectorizes a large volume of course materials—including lecture slides, coding assignments, and video lectures—and uses them as the basis for answering students’ questions.
< Photo 2. Teaching Assistant demonstrating to the student how the Virtual Teaching Assistant works>
When a student asks a question, the system searches for the most relevant course materials in real time based on the context of the query, and then generates a response. This process is not merely a simple call to a large language model (LLM), but rather a material-grounded question answering system tailored to the course content—ensuring both high reliability and accuracy in learning support.
Sunjun Kweon, the first author of the study and head teaching assistant for the course, explained, “Previously, TAs were overwhelmed with repetitive and basic questions—such as concepts already covered in class or simple definitions—which made it difficult to focus on more meaningful inquiries.” He added, “After introducing the VTA, students began to reduce repeated questions and focus on more essential ones. As a result, the burden on TAs was significantly reduced, allowing us to concentrate on providing more advanced learning support.”
In fact, compared to the previous year’s course, the number of questions that required direct responses from human TAs decreased by approximately 40%.
< Photo 3. A student working with VTA. >
The VTA, which was operated over a 14-week period, was actively used by more than half of the enrolled students, with a total of 3,869 Q&A interactions recorded. Notably, students without a background in AI or with limited prior knowledge tended to use the VTA more frequently, indicating that the system provided practical support as a learning aid, especially for those who needed it most.
The analysis also showed that students tended to ask the VTA more frequently about theoretical concepts than they did with human TAs. This suggests that the AI teaching assistant created an environment where students felt free to ask questions without fear of judgment or discomfort, thereby encouraging more active engagement in the learning process.
According to surveys conducted before, during, and after the course, students reported increased trust, response relevance, and comfort with the VTA over time. In particular, students who had previously hesitated to ask human TAs questions showed higher levels of satisfaction when interacting with the AI teaching assistant.
< Figure 1. Internal structure of the AI Teaching Assistant (VTA) applied in this course. It follows a Retrieval-Augmented Generation (RAG) structure that builds a vector database from course materials (PDFs, recorded lectures, coding practice materials, etc.), searches for relevant documents based on student questions and conversation history, and then generates responses based on them. >
Professor Yoonjae Choi, the lead instructor of the course and principal investigator of the study, stated, “The significance of this research lies in demonstrating that AI technology can provide practical support to both students and instructors. We hope to see this technology expanded to a wider range of courses in the future.”
The research team has released the system’s source code on GitHub, enabling other educational institutions and researchers to develop their own customized learning support systems and apply them in real-world classroom settings.
< Figure 2. Initial screen of the AI Teaching Assistant (VTA) introduced in the "Programming for AI" course. It asks for student ID input along with simple guidelines, a mechanism to ensure that only registered students can use it, blocking indiscriminate external access and ensuring limited use based on students. >
The related paper, titled “A Large-Scale Real-World Evaluation of an LLM-Based Virtual Teaching Assistant,” was accepted on May 9, 2025, to the Industry Track of ACL 2025, one of the most prestigious international conferences in the field of Natural Language Processing (NLP), recognizing the excellence of the research.
< Figure 3. Example conversation with the AI Teaching Assistant (VTA). When a student inputs a class-related question, the system internally searches for relevant class materials and then generates an answer based on them. In this way, VTA provides learning support by reflecting class content in context. >
This research was conducted with the support of the KAIST Center for Teaching and Learning Innovation, the National Research Foundation of Korea, and the National IT Industry Promotion Agency.
2025.06.05 View 1346 -
RAIBO Runs over Walls with Feline Agility... Ready for Effortless Search over Mountaineous and Rough Terrains
< Photo 1. Research Team Photo (Professor Jemin Hwangbo, second from right in the front row) >
KAIST's quadrupedal robot, RAIBO, can now move at high speed across discontinuous and complex terrains such as stairs, gaps, walls, and debris. It has demonstrated its ability to run on vertical walls, leap over 1.3-meter-wide gaps, sprint at approximately 14.4 km/h over stepping stones, and move quickly and nimbly on terrain combining 30° slopes, stairs, and stepping stones. RAIBO is expected to be deployed soon for practical missions such as disaster site exploration and mountain searches.
Professor Jemin Hwangbo's research team in the Department of Mechanical Engineering at our university announced on June 3rd that they have developed a quadrupedal robot navigation framework capable of high-speed locomotion at 14.4 km/h (4m/s) even on discontinuous and complex terrains such as walls, stairs, and stepping stones.
The research team developed a quadrupedal navigation system that enables the robot to reach its target destination quickly and safely in complex and discontinuous terrain.
To achieve this, they approached the problem by breaking it down into two stages: first, developing a planner for planning foothold positions, and second, developing a tracker to accurately follow the planned foothold positions.
First, the planner module quickly searches for physically feasible foothold positions using a sampling-based optimization method with neural network-based heuristics and verifies the optimal path through simulation rollouts.
While existing methods considered various factors such as contact timing and robot posture in addition to foothold positions, this research significantly reduced computational complexity by setting only foothold positions as the search space. Furthermore, inspired by the walking method of cats, the introduction of a structure where the hind feet step on the same spots as the front feet further significantly reduced computational complexity.
< Figure 1. High-speed navigation across various discontinuous terrains >
Second, the tracker module is trained to accurately step on planned positions, and tracking training is conducted through a generative model that competes in environments of appropriate difficulty.
The tracker is trained through reinforcement learning to accurately step on planned plots, and during this process, a generative model called the 'map generator' provides the target distribution.
This generative model is trained simultaneously and adversarially with the tracker to allow the tracker to progressively adapt to more challenging difficulties. Subsequently, a sampling-based planner was designed to generate feasible foothold plans that can reflect the characteristics and performance of the trained tracker.
This hierarchical structure showed superior performance in both planning speed and stability compared to existing techniques, and experiments proved its high-speed locomotion capabilities across various obstacles and discontinuous terrains, as well as its general applicability to unseen terrains.
Professor Jemin Hwangbo stated, "We approached the problem of high-speed navigation in discontinuous terrain, which previously required a significantly large amount of computation, from the simple perspective of how to select the footprint positions. Inspired by the placements of cat's paw, allowing the hind feet to step where the front feet stepped drastically reduced computation. We expect this to significantly expand the range of discontinuous terrain that walking robots can overcome and enable them to traverse it at high speeds, contributing to the robot's ability to perform practical missions such as disaster site exploration and mountain searches."
This research achievement was published in the May 2025 issue of the international journal Science Robotics.
Paper Title: High-speed control and navigation for quadrupedal robots on complex and discrete terrain, (https://www.science.org/doi/10.1126/scirobotics.ads6192)YouTube Link: https://youtu.be/EZbM594T3c4?si=kfxLF2XnVUvYVIyk
2025.06.04 View 2247 -
KAIST Develops Virtual Staining Technology for 3D Histopathology
Moving beyond traditional methods of observing thinly sliced and stained cancer tissues, a collaborative international research team led by KAIST has successfully developed a groundbreaking technology. This innovation uses advanced optical techniques combined with an artificial intelligence-based deep learning algorithm to create realistic, virtually stained 3D images of cancer tissue without the need for serial sectioning nor staining. This breakthrough is anticipated to pave the way for next-generation non-invasive pathological diagnosis.
< Photo 1. (From left) Juyeon Park (Ph.D. Candidate, Department of Physics), Professor YongKeun Park (Department of Physics) (Top left) Professor Su-Jin Shin (Gangnam Severance Hospital), Professor Tae Hyun Hwang (Vanderbilt University School of Medicine) >
KAIST (President Kwang Hyung Lee) announced on the 26th that a research team led by Professor YongKeun Park of the Department of Physics, in collaboration with Professor Su-Jin Shin's team at Yonsei University Gangnam Severance Hospital, Professor Tae Hyun Hwang's team at Mayo Clinic, and Tomocube's AI research team, has developed an innovative technology capable of vividly displaying the 3D structure of cancer tissues without separate staining.
For over 200 years, conventional pathology has relied on observing cancer tissues under a microscope, a method that only shows specific cross-sections of the 3D cancer tissue. This has limited the ability to understand the three-dimensional connections and spatial arrangements between cells.
To overcome this, the research team utilized holotomography (HT), an advanced optical technology, to measure the 3D refractive index information of tissues. They then integrated an AI-based deep learning algorithm to successfully generate virtual H&E* images.* H&E (Hematoxylin & Eosin): The most widely used staining method for observing pathological tissues. Hematoxylin stains cell nuclei blue, and eosin stains cytoplasm pink.
The research team quantitatively demonstrated that the images generated by this technology are highly similar to actual stained tissue images. Furthermore, the technology exhibited consistent performance across various organs and tissues, proving its versatility and reliability as a next-generation pathological analysis tool.
< Figure 1. Comparison of conventional 3D tissue pathology procedure and the 3D virtual H&E staining technology proposed in this study. The traditional method requires preparing and staining dozens of tissue slides, while the proposed technology can reduce the number of slides by up to 10 times and quickly generate H&E images without the staining process. >
Moreover, by validating the feasibility of this technology through joint research with hospitals and research institutions in Korea and the United States, utilizing Tomocube's holotomography equipment, the team demonstrated its potential for full-scale adoption in real-world pathological research settings.
Professor YongKeun Park stated, "This research marks a major advancement by transitioning pathological analysis from conventional 2D methods to comprehensive 3D imaging. It will greatly enhance biomedical research and clinical diagnostics, particularly in understanding cancer tumor boundaries and the intricate spatial arrangements of cells within tumor microenvironments."
< Figure 2. Results of AI-based 3D virtual H&E staining and quantitative analysis of pathological tissue. The virtually stained images enabled 3D reconstruction of key pathological features such as cell nuclei and glandular lumens. Based on this, various quantitative indicators, including cell nuclear distribution, volume, and surface area, could be extracted. >
This research, with Juyeon Park, a student of the Integrated Master’s and Ph.D. Program at KAIST, as the first author, was published online in the prestigious journal Nature Communications on May 22.
(Paper title: Revealing 3D microanatomical structures of unlabeled thick cancer tissues using holotomography and virtual H&E staining.
[https://doi.org/10.1038/s41467-025-59820-0]
This study was supported by the Leader Researcher Program of the National Research Foundation of Korea, the Global Industry Technology Cooperation Center Project of the Korea Institute for Advancement of Technology, and the Korea Health Industry Development Institute.
2025.05.26 View 2545 -
KAIST to Develop a Korean-style ChatGPT Platform Specifically Geared Toward Medical Diagnosis and Drug Discovery
On May 23rd, KAIST (President Kwang-Hyung Lee) announced that its Digital Bio-Health AI Research Center (Director: Professor JongChul Ye of KAIST Kim Jaechul Graduate School of AI) has been selected for the Ministry of Science and ICT's 'AI Top-Tier Young Researcher Support Program (AI Star Fellowship Project).' With a total investment of ₩11.5 billion from May 2025 to December 2030, the center will embark on the full-scale development of AI technology and a platform capable of independently inferring and determining the kinds of diseases, and discovering new drugs.
< Photo. On May 20th, a kick-off meeting for the AI Star Fellowship Project was held at KAIST Kim Jaechul Graduate School of AI’s Yangjae Research Center with the KAIST research team and participating organizations of Samsung Medical Center, NAVER Cloud, and HITS. [From left to right in the front row] Professor Jaegul Joo (KAIST), Professor Yoonjae Choi (KAIST), Professor Woo Youn Kim (KAIST/HITS), Professor JongChul Ye (KAIST), Professor Sungsoo Ahn (KAIST), Dr. Haanju Yoo (NAVER Cloud), Yoonho Lee (KAIST), HyeYoon Moon (Samsung Medical Center), Dr. Su Min Kim (Samsung Medical Center) >
This project aims to foster an innovative AI research ecosystem centered on young researchers and develop an inferential AI agent that can utilize and automatically expand specialized knowledge systems in the bio and medical fields.
Professor JongChul Ye of the Kim Jaechul Graduate School of AI will serve as the lead researcher, with young researchers from KAIST including Professors Yoonjae Choi, Kimin Lee, Sungsoo Ahn, and Chanyoung Park, along with mid-career researchers like Professors Jaegul Joo and Woo Youn Kim, jointly undertaking the project. They will collaborate with various laboratories within KAIST to conduct comprehensive research covering the entire cycle from the theoretical foundations of AI inference to its practical application.
Specifically, the main goals include: - Building high-performance inference models that integrate diverse medical knowledge systems to enhance the precision and reliability of diagnosis and treatment. - Developing a convergence inference platform that efficiently combines symbol-based inference with neural network models. - Securing AI technology for new drug development and biomarker discovery based on 'cell ontology.'
Furthermore, through close collaboration with industry and medical institutions such as Samsung Medical Center, NAVER Cloud, and HITS Co., Ltd., the project aims to achieve: - Clinical diagnostic AI utilizing medical knowledge systems. - AI-based molecular target exploration for new drug development. - Commercialization of an extendible AI inference platform.
Professor JongChul Ye, Director of KAIST's Digital Bio-Health AI Research Center, stated, "At a time when competition in AI inference model development is intensifying, it is a great honor for KAIST to lead the development of AI technology specialized in the bio and medical fields with world-class young researchers." He added, "We will do our best to ensure that the participating young researchers reach a world-leading level in terms of research achievements after the completion of this seven-year project starting in 2025."
The AI Star Fellowship is a newly established program where post-doctoral researchers and faculty members within seven years of appointment participate as project leaders (PLs) to independently lead research. Multiple laboratories within a university and demand-side companies form a consortium to operate the program.
Through this initiative, KAIST plans to nurture bio-medical convergence AI talent and simultaneously promote the commercialization of core technologies in collaboration with Samsung Medical Center, NAVER Cloud, and HITS.
2025.05.26 View 3022 -
KAIST and Mainz Researchers Unveil 3D Magnon Control, Charting a New Course for Neuromorphic and Quantum Technologies
< Professor Se Kwon Kim of the Department of Physics (left), Dr. Zarzuela of the University of Mainz, Germany (right) >
What if the magnon Hall effect, which processes information using magnons (spin waves) capable of current-free information transfer with magnets, could overcome its current limitation of being possible only on a 2D plane? If magnons could be utilized in 3D space, they would enable flexible design, including 3D circuits, and be applicable in various fields such as next-generation neuromorphic (brain-mimicking) computing structures, similar to human brain information processing. KAIST and an international joint research team have, for the first time in the world, predicted a 3D magnon Hall effect, demonstrating that magnons can move freely and complexly in 3D space, transcending the conventional concept of magnons.
KAIST (President Kwang Hyung Lee) announced on May 22nd that Professor Se Kwon Kim of the Department of Physics, in collaboration with Dr. Ricardo Zarzuela of the University of Mainz, Germany, has revealed that the interaction between magnons (spin waves) and solitons (spin vortices) within complex magnetic structures (topologically textured frustrated magnets) is not simple, but complex in a way that enables novel functionalities.
Magnons (spin waves), which can transmit information like electron movement, are garnering attention as a next-generation information processing technology that transmits information without using current, thus generating no heat. Until now, magnon research has focused on simple magnets where spins are neatly aligned in one direction, and the mathematics describing this was a relatively simple 'Abelian gauge theory.'
The research team demonstrated, for the first time in the world, that in complex spin structures like frustrated magnets, magnons interact and become entangled in complex ways from various directions. They applied an advanced mathematical framework, 'non-Abelian gauge theory,' to describe this movement, which is a groundbreaking achievement.
This research presents the possibility of future applications in low-power logic devices using magnons and topology-based quantum information processing technologies, indicating a potential paradigm shift in future information technology.
In conventional linear magnetic materials, the value representing the magnetic state (order parameter) is given as a vector. In magnonics research based on this, it has been interpreted that a U(1) Abelian gauge field is induced when magnons move in soliton structures like skyrmions. This means that the interaction between solitons and magnons has a structure similar to quantum electrodynamics (QED), which has successfully explained various experimental results such as the magnon Hall effect in 2D magnets.
< Figure. Schematic diagram of non-Abelian magnon quantum chromodynamics describing the dynamics of three types of magnons discovered for the first time in this study.>
However, through this research, the team theoretically revealed that in frustrated magnets, the order parameter must be expressed not as a simple vector but as a quaternion. As a result, the gauge field experienced by magnons resembles an SU(3) non-Abelian gauge field, rather than a simple U(1) Abelian gauge field.
This implies that within frustrated magnets, there are not one or two types of magnons seen in conventional magnets, but three distinct types of magnons, each interacting and intricately entangled with solitons. This structure is highly significant as it resembles quantum chromodynamics (QCD) that describes the strong interaction between quarks mediated by gluons rather than quantum electrodynamics (QED) that describes electromagnetic forces.
Professor Se Kwon Kim stated, "This research presents a powerful theoretical framework to explain the dynamics of magnons occurring within the complex order of frustrated magnets," adding, "By pioneering non-Abelian magnonics, it will be a conceptual turning point that can influence quantum magnetism research as a whole."
The research results, with Dr. Ricardo Zarzuela of the University of Mainz, Germany, as the first author, were published in the world-renowned physics journal Physical Review Letters on May 6th.※ Paper title: "Non-Abelian Gauge Theory for Magnons in Topologically Textured Frustrated Magnets," Phys. Rev. Lett. 134, 186701 (2025)DOI: https://doi.org/10.1103/PhysRevLett.134.186701
This research was supported by the Brain Pool Plus program of the National Research Foundation of Korea.
2025.05.22 View 2438