%41%49
-
World’s First SoulMate AI Semiconductor: A Personalized Digital Soulmate Developed
< (From left) KAIST Professor Hoi-Jun Yoo and PhD candidate Seongyon Hong >
While Large Language Models (LLMs) like ChatGPT are adept at answering countless questions, they often remain unaware of a user's minor habits or previous conversational contexts. This is why AI, despite being deeply integrated into our daily lives, can still feel like a "stranger." Overcoming these limitations, researchers at KAIST have developed the world’s first AI semiconductor, dubbed "SoulMate," which learns and adapts to a user’s speech style, preferences, and emotions in real-time—becoming a true "digital soulmate."
KAIST announced on March 17th that a research team led by Professor Hoi-Jun Yoo from the Graduate School of AI Semiconductors has developed SoulMate, a personalized LLM accelerator that evolves according to the specific characteristics of the user.This technology is being hailed as a core semiconductor breakthrough that will accelerate the era of "Hyper-Personalized AI"—moving beyond "AI for everyone" to an AI that learns and responds to an individual's unique conversational style and preferences.
The core of SoulMate lies in On-Device AI technology, which processes data directly on the device without going through external servers (the cloud). The team directly implemented Retrieval-Augmented Generation (RAG), which generates customized answers based on remembered conversations, and Low-Rank Adaptation (LoRA), which immediately reflects and learns from user feedback, within the semiconductor itself.
< SoulMate AI Semiconductor Chip >
Through this, SoulMate has realized a real-time personalized AI system that responds to the user at a staggering speed of 0.2 seconds (216.4 ms) while simultaneously performing learning tasks.
< SoulMate Application Demo >
Furthermore, the team applied a Mixed-Rank architecture that optimizes processing methods based on the importance of information, drastically reducing power consumption. The semiconductor operates at an ultra-low power of just 9.8 milliwatts (mW)—approximately 1/500th of a typical smartphone processor's power consumption—allowing it to handle complex learning and inference simultaneously on mobile devices without battery concerns.
In particular, SoulMate features a "Security-Complete AI" structure where all personal data is processed internally within the device rather than being transmitted to external servers, fundamentally blocking any risk of personal information leaks. The research team expects this technology to pair with next-generation platforms such as smartphones, wearables, and personal AI devices to open a true era of personalized AI services.
< SoulMate Demo Screen >
"This research mimics the process of people building friendships, providing the technical foundation for AI to evolve into a true companion for the user," said Professor Hoi-Jun Yoo. "Future AI will move beyond being a mere tool to become a 'Best Friend' that understands me best anytime, anywhere, while perfectly protecting personal privacy."
The study, with PhD student Seongyon Hong as the first author, was selected as a "Highlight Paper" at the International Solid-State Circuits Conference (ISSCC) held in San Francisco this past February, garnering significant attention from the global academic community.
Paper Title: SoulMate: A 9.8mW Mobile Intelligence System-on-Chip with Mixed-Rank Architecture for On-Device LLM Personalization Authors: Seongyon Hong, Jiwon Choi, Jeonggyu So, Nayeong Lee, Wooyoung Jo, Zhamaliddin Kalzhan Link: https://ieeexplore.ieee.org/document/11409048
At the conference, the research team successfully demonstrated how the AI's response style changes in real-time according to user reactions using the actual semiconductor chip, proving the excellence of Korean AI semiconductor technology. The SoulMate AI semiconductor is planned for commercialization around 2027 through the faculty-led startup "OnNeuro AI."
< SoulMate Demonstration Photo >
This research was conducted with support from the Information and Communication Broadcast Innovation Talent Cultivation Program of the Ministry of Science and ICT and the Institute of Information & Communications Technology Planning & Evaluation (IITP).
2026.03.17 View 12 -
Secret to Drug Addiction Relapse Found: Brain's Addiction Circuit Identified
<(From Left) Dr. Minju Jeong,(UCSD), Prof. Byung Kook Lim (UCSD), Prof. Se-Bum Paik (KAIST)>
Drug addiction carries an extremely high risk of relapse, as cravings can be reignited by minor stimuli even long after one has stopped using. Previously, this phenomenon was attributed to a decline in the function of the prefrontal cortex (PFC), which regulates impulses. However, a joint international research team has recently revealed that the cause of addiction relapse is not a simple decline in brain function, but rather an imbalance in specific neural circuits.
KAIST announced on March 9th that a research team led by Prof. Se-Bum Paik from the Department of Brain and Cognitive Sciences and Prof. Byung Kook Lim from the University of California, San Diego (UCSD) has identified the core principle by which specific inhibitory neurons in the prefrontal cortex regulate cocaine-seeking behavior.
In particular, the research team focused on parvalbumin-positive (PV) inhibitory neurons, which regulate the balance of neural signals by suppressing the activity of other neurons in the brain. They confirmed that these cells act as a "brake gate" that controls excitatory signals in the brain and serve as a crucial factor in determining drug-seeking behavior that emerges after withdrawal.
The prefrontal cortex (PFC) of our brain can properly perform its "braking" function to suppress impulses when excitatory and inhibitory signals are in balance. To investigate how chronic drug exposure disrupts this balance, the research team conducted cocaine administration experiments on mice. During this process, they tracked when inhibitory neurons in the PFC were activated and how they sent signals to downstream brain regions.
The experimental results showed that parvalbumin (PV) cells, which account for about 60-70% of the inhibitory neurons in the PFC, were highly active when the mice attempted to seek cocaine. However, when "extinction training"—training to stop seeking the drug—was conducted, the activity of these cells significantly decreased. This demonstrates that the activity patterns of PV cells are not permanently fixed by addiction but can be readjusted through the extinction process.
<Figure 1. Experimental design illustrating cocaine self-administration and longitudinal tracking of prefrontal cortical neural activity during cocaine-seeking behavior>
The research team confirmed that artificially suppressing PV cell activity significantly reduced cocaine-seeking behavior in mice. Conversely, activating these cells caused the drug-seeking behavior to persist even after the extinction process. This effect was specifically observed in drug-addiction behavior and did not appear with general rewards like sugar water. Furthermore, this phenomenon was not observed in somatostatin (SOM) cells—another type of inhibitory neuron—indicating that PV cells selectively regulate drug addiction behavior.
<Figure 2. Comparison of single-neuron activity, population activity patterns, and behavioral modulation of prefrontal inhibitory neurons across different stages of cocaine-seeking behavior>
The team also identified the specific brain circuit through which these PV cells operate. Signals originating from the prefrontal cortex are transmitted to the reward circuit of the Ventral Tegmental Area (VTA), a key brain region related to reward. This pathway emerged as the central channel for regulating addiction behavior, determining whether or not to seek the drug again. In this process, PV neurons act as a "regulatory switch," controlling the flow of signals to influence dopamine signaling and deciding whether to maintain or suppress addictive behavior.
In short, the study revealed that addiction relapse is not due to an overall functional decline of the prefrontal cortex, but is determined by whether PV neurons regulate the neural pathway connecting the PFC to the reward circuit.
<Figure 3. Schematic illustrating the prefrontal–reward circuit mechanism that determines drug-seeking behavior>
Prof. Se-Bum Paik stated, "This research shows that drug addiction is a circuit-level problem arising from a collapse in the regulatory balance of specific neurons and downstream neural circuits. The discovery that parvalbumin (PV) cells act as a 'gate' for addictive behavior will provide a crucial lead for developing precision-targeted treatment strategies in the future."
This study was led by Dr. Minju Jeong (UCSD) as the first author, with Prof. Byung Kook Lim (UCSD) and Prof. Se-Bum Paik (KAIST) serving as co-corresponding authors. The findings were published online on February 26 in Neuron, a premier journal in the field of neuroscience.
Paper Title: Distinct Interneuronal Dynamics Selectively Gate Target-Specific Cortical Projections in Drug Seeking
DOI: 10.1016/j.neuron.2026.01.002
Full Author List: Minju Jeong, Seungdae Baek, Qingdi Wang, Li Yao, Eun Ji Lee, Arturo Marroquin Rivera, Joann Jocelynn Lee, Hyeonseok Jang, Dhananjay Bambah-Mukku, Christine Hyun-Seung Mun, Tyler Boesen, Sumit Nanda, Cheol Ryong Ku, Hong-wei Dong, Benoit Labonté, Se-Bum Paik, and Byung Kook Lim.
This research was conducted with the support of the Basic Research Program in Science and Engineering of the National Research Foundation of Korea.
2026.03.10 View 275 -
KAIST Develops mRNA Platform That Remains Effective Even in Aging and Obesity
<(From Left) Dr. Subin Yoon, Ph.D candidate Hyeonggon Cho, Prof. Jae-Hwan Nam, Prof. Young-suk Lee>
Since the COVID-19 pandemic, mRNA vaccines have gained attention as a next-generation pharmaceutical technology. mRNA therapeutics work by delivering genetic instructions that enable cells to produce specific proteins for therapeutic effects. However, their efficacy has been reported to decline in elderly individuals or patients with obesity. To address this limitation, Korean researchers have newly designed a key regulatory region of mRNA that improves therapeutic protein production efficiency, developing a next-generation mRNA platform that maintains effectiveness even in aging and obesity conditions.
KAIST (President Kwang Hyung Lee) announced on the 10th of March that a joint research team led by Professor Young-suk Lee of the Department of Bio and Brain Engineering and Professor Jae-Hwan Nam of The Catholic University of Korea (President Jun-Gyu Choi) has developed a new mRNA platform by precisely designing the sequence of the 5′ untranslated region (5′UTR)*, a key regulatory region of mRNA.*5′ untranslated region (5′UTR): A region of mRNA that initiates and regulates protein production. The design of this region influences both the amount and speed of protein synthesis.
The research team analyzed large-scale bioinformatics datasets to identify 5′UTR sequences that enable proteins to be produced more efficiently across diverse cellular environments. When applied, the designed sequences significantly enhanced protein production and immune responses even in preclinical models of aging and obesity.
mRNA is a long single-stranded RNA molecule that serves as the blueprint for producing proteins required by the body. It consists of several components: the 5′UTR, which initiates and regulates the rate of protein production; the coding sequence (CDS), which contains the genetic information for a specific protein; the 3′ untranslated region (3′UTR), which helps maintain mRNA stability within cells; and the poly(A) tail, which further enhances stability and supports protein synthesis.
Among these components, the 5′UTR and 3′UTR do not determine the type of protein produced, but they play a critical role in regulating how efficiently the protein is synthesized. For this reason, these regions are receiving increasing attention as key bioengineering platforms for improving the performance of various mRNA therapeutics, including vaccines and treatments.
<Schematic Diagram of mRNA Therapeutic Design and Validation Using Bioinformatics>
To identify highly efficient 5′UTR sequences capable of promoting protein production across multiple tissues and cellular environments, the team conducted an integrated analysis of large-scale biological datasets. This included multiple analytical approaches such as RNA sequencing (RNA-seq) for analyzing gene activity across tissues, single-cell RNA sequencing (scRNA-seq) for examining gene expression at the individual cell level, and ribosome profiling (Ribo-seq) for measuring actual protein translation efficiency.
The researchers also focused on the fact that in aging or obesity conditions, cells often experience high levels of stress—particularly oxidative stress—which can reduce their ability to synthesize proteins. When the newly designed mRNA therapeutics were applied to preclinical models of aging and obesity, the results showed significantly improved protein production and immune responses compared with existing approaches. This research is expected to be applicable not only to mRNA vaccines but also to a wide range of biopharmaceutical technologies, including gene therapies and immunotherapies.
<Multimodal Bio–Big Data Analysis–Based mRNA Therapeutic Design (AI-Generated Image)>
Professor Young-suk Lee of KAIST Department of Bio and Brain Engineering stated, “This study identified a design strategy that enables mRNA to produce proteins more efficiently by analyzing large-scale biological data,” adding, “This technology will provide an important foundation for ensuring that mRNA vaccines and therapeutics remain effective even in environments where drug efficacy may decline, such as in elderly or obese patients.”
In this study, Dr. Subin Yoon from The Catholic University of Korea and doctoral candidate Hyeonggon Cho from KAIST participated as co-first authors. The research findings were published online on January 2 in the internationally renowned journal Molecular Therapy (IF = 12.0), a leading journal in gene and cell therapy.
(Paper title: ”Designing 5′UTR sequences improves the capacity of mRNA therapeutics in preclinical models of aging and obesity” DOI: https://doi.org/10.1016/j.ymthe.2025.12.060)
This research was supported by the Excellent Young Researcher Program and the Bio-Medical Technology Development Program of the National Research Foundation of Korea funded by the Ministry of Science and ICT, the Infectious Disease Response Innovative Technology Support Program of the Ministry of Food and Drug Safety, and the Infectious Disease Prevention and Therapeutics Technology Development Program of the Korea Health Industry Development Institute.
2026.03.10 View 270 -
Earth’s Safety Limit Already Exceeded… Carbon Emissions More Than Double the Planetary Boundary
<(From Left) Professor Haewon McJeon, Dr. Paul Wolfram>
Earth is not infinite. Pollution beyond certain levels threatens the climate and ecosystems. To prevent this, scientists have proposed “Planetary Boundaries,” defining the safe operating limits of the Earth system. A KAIST research team recalculated climate change and nitrogen pollution using the same standard and found that current carbon emissions already exceed the planet’s sustainable limit by more than double.
KAIST (President Kwang Hyung Lee) announced on the 6th of March that Professor Haewon McJeon of the Graduate School of Green Growth and Sustainability, in collaboration with Dr. Paul Wolfram’s team at the Pacific Northwest National Laboratory (PNNL) of the U.S. Department of Energy, recalculated the carbon dioxide emission boundary using an annual emissions (flow) framework rather than the traditional cumulative carbon stock framework.
Until now, climate change has been evaluated based on how much CO₂ accumulates in the atmosphere (stock). In contrast, nitrogen and phosphorus pollution have been assessed based on how much is emitted each year (flow). Because these problems were measured using different metrics, it was difficult to fairly compare their relative severity. The research team therefore recalculated carbon emissions using the same annual emissions framework used for nitrogen pollution.
Based on the condition of limiting the rise in global average temperature to within 1.5°C, the analysis showed that the Earth’s safe limit for annual CO₂ emissions is approximately 4–17 gigatons (Gt CO₂ per year). However, humanity’s current annual emissions amount to about 37 gigatons (Gt CO₂ per year). This level exceeds the Earth’s safe operating space by more than twofold.
Professor Haewon McJeon stated, “When carbon emissions are compared using the same framework as nitrogen pollution, the severity of climate change becomes much clearer,” adding, “This study helps place different environmental problems on the same analytical basis, which can contribute to setting clearer policy priorities.”
<Comparative Measurement of Planetary Boundaries and Proposal for Flow-Based Carbon Emission Limits>
<Scope and Sensitivity of Flow-Based Carbon Emission Limits>
He further emphasized, “The need for integrated strategies that simultaneously consider carbon, nitrogen, and phosphorus pollution is growing,” adding that global efforts toward decarbonization must accelerate further.
The study was jointly led by Professor Haewon McJeon and Dr. Paul Wolfram as co-corresponding authors, with Hassan Niazi, Page Kyle, and other researchers from PNNL participating as collaborators. The research results were published on February 16 in the international journal Nature Sustainability.
※ Paper title: “Ensuring consistency between biogeochemical planetary boundaries”
DOI: https://doi.org/10.1038/s41893-026-01770-6
This research was supported by the project “Development of an AI-Based Next-Generation Integrated Assessment Model for Climate–Human Interactions” funded by the Ministry of Science and ICT and the National Research Foundation of Korea.
In a Science commentary published on March 5 titled “Thirty-six solutions to stabilize Earth’s climate,” Professor McJeon revisited the progress of climate technologies over the past 20 years. He pointed out that although humanity has possessed many of the necessary technologies, they have not been implemented quickly enough, allowing the climate crisis to intensify. He also emphasized that the pace of decarbonization must accelerate to achieve carbon neutrality.
※ Commentary: “Thirty-six solutions to stabilize Earth’s climate”
Link: https://doi.org/10.1126/science.aed5212
2026.03.06 View 408 -
KAIST Develops Brain-Like AI… Thinks One More Time Even When Predictions Are Wrong
<(From left) Professor Sang Wan Lee, Myoung Hoon Ha, and Dr. Yoondo Sung>
Artificial intelligence now plays Go, paints pictures, and even converses like a human. However, there remains a decisive difference: AI requires far more electricity than the human brain to operate. Scientists have long asked the question, “How can the brain learn so intelligently using so little energy?” KAIST researchers have moved one step closer to the answer.
KAIST (President Kwang Hyung Lee) announced on the 29th that a research team led by Distinguished Professor Sang Wan Lee of the Department of Brain and Cognitive Sciences has developed a new technology that applies the learning principles of the human brain to deep learning, enabling stable training even in deep artificial intelligence models.
Our brain does not passively receive the world. Instead of merely perceiving what is happening in the present, it first predicts what will happen next and, when reality differs from that prediction, adjusts itself to reduce the difference (i.e., prediction error). This is similar to anticipating an opponent’s next move in Go and changing strategy if the prediction turns out to be wrong. This mode of information processing is known as “Predictive Coding.”
< Predictive Coding (PC) Module >
Scientists have attempted to apply this principle to AI, but encountered difficulties. As neural networks become deeper, errors tend to concentrate in specific layers or vanish altogether, repeatedly leading to performance degradation.
The research team mathematically identified the cause of this problem and proposed a new solution. The key idea is simple: instead of predicting only the final outcome, the AI is designed to also predict how its prediction errors will change in the future. The team refers to this as “Meta Prediction.” In simple terms, it is an AI that “thinks once more about its mistakes.” When this method was applied, learning proceeded stably in deep neural networks without halting.
<Analysis of Instability in Predictive Coding Model Errors>
The experimental results were also impressive. In 29 out of 30 experiments, the proposed method achieved higher accuracy than the current standard AI training method, backpropagation. Backpropagation is the representative learning method in which AI “goes backward by the amount of error and corrects it.”
Conventional AI training methods (backpropagation) require tightly interconnected layers, meaning the entire network must be computed and updated simultaneously. In contrast, this new approach demonstrates that, like the brain, large AI models can be effectively trained even when learning occurs in a distributed and partially independent manner.
<Performance Comparison of Predictive Coding Models>
This technology is expected to expand into various fields where power efficiency is critical, including neuromorphic computing, robot AI that must adapt to changing environments, and edge AI operating within devices.
Distinguished Professor Sang Wan Lee stated, “The key to this research is not simply imitating the structure of the brain, but enabling AI to follow the brain’s learning principles themselves,” adding, “We have opened the possibility of artificial intelligence that learns efficiently like the brain.”
This study was conducted with Dr. Myoung Hoon Ha as the first author and Professor Sang Wan Lee as the corresponding author. The paper was accepted to the International Conference on Learning Representations (ICLR 2026) and was published online on January 26.
※ Paper title: “Stable and Scalable Deep Predictive Coding Networks with Meta Prediction Errors”Original paper: https://openreview.net/forum?id=kE5jJUHl9i¬eId=e6T5T9cYqO
This research was supported by the Ministry of Science and ICT and the Institute of Information & Communications Technology Planning & Evaluation (IITP) through the Digital Global Research Support Program (joint research with Microsoft Research), the Samsung Electronics SAIT NPRC Program, and the SW Star Lab Program.
2026.03.06 View 577 -
Developing Technology to Become the Joker in The Dark Knight
<(From left) Ph.D. candidate Taewoong Kang, Ph.D candidate Junha Hyung, Professor Jaegul Choo, and Ph.D. candidate Minho Park (From top right square, from left), Ph.D. candidate Kinam Kim, Seoul National University undergraduate researcher Dohyeon Kim>
What if, while watching The Dark Knight, you weren't just observing the Joker on screen, but actually seeing Gotham City through his eyes? The video technology that allows viewers to experience the world through a character's perspective, rather than as a mere observer, is becoming a reality. Researchers at our university have developed a new AI model that generates first-person viewpoint videos from standard footage.
KAIST announced on February 23rd that Professor Jaegul Choo’s research team at the Kim Jaechul Graduate School of AI has developed 'EgoX,' an AI model that utilizes observer-perspective (exocentric) video to precisely generate the scenes that a person in the video would actually be seeing.
With the rapid advancement of Augmented Reality (AR), Virtual Reality (VR), and AI robotics, the importance of "egocentric video"—which captures scenes as one directly sees them—is growing. However, obtaining high-quality first-person footage previously required users to wear expensive action cameras or smart glasses. Furthermore, there were significant technical limitations in naturally converting existing standard (third-person or exocentric) video into a first-person perspective.
A key feature of this technology is that it goes beyond simply rotating the screen; it comprehensively understands the person's position, posture, and the 3D structure of the surrounding space to reconstruct the first-person viewpoint.
< Example of converting a third-person perspective video into a first-person perspective video >
Existing technologies often only converted still images or required footage from four or more cameras. Additionally, they frequently suffered from awkward visual artifacts in videos with complex lighting or rapid movement.
In contrast, EgoX can generate high-quality first-person video from just a single third-person video source. Specifically, the research team succeeded in realistically implementing natural shifts in vision—such as when a person turns their head—by precisely modeling the correlation between head movement and the actual field of view.
This technology demonstrated stable performance across various daily scenarios, including cooking, exercising, and working, without being limited to specific environments. It is being evaluated as a breakthrough that opens new possibilities for securing high-quality first-person data from existing video archives without the need for wearable devices.
EgoX is expected to have a significant impact across various industries. In the fields of AR, VR, and the Metaverse, it can maximize user experience by transforming standard videos into immersive content that makes users feel as if they are experiencing the scene firsthand.
Furthermore, it is projected to contribute to the fields of robotics and AI training by serving as core data for "Imitation Learning," where robots learn by watching human actions. New types of video services, such as switching sports broadcasts or vlogs to the perspective of the athlete or the protagonist, are also anticipated.
< EgoX technology that converts a third-person perspective into a first-person perspective (AI-generated image) >
Distinguished Professor Jaegul Choo stated, "This research is significant in that AI has moved beyond simple video conversion to learning and reconstructing human 'vision' and 'spatial understanding.' We expect an environment to open up where anyone can create and experience immersive content using only previously recorded videos." He added, "KAIST will continue to secure global competitiveness in the field of generative AI-based video technology."
This research was led by first authors Taewoong Kang, Kinam Kim, and Dohyeon Kim . The paper was pre-released on arXiv on December 9, 2025, garnering significant attention from AI industry giants like NVIDIA and Meta, as well as academia. It is scheduled for official presentation at the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), an international academic conference to be held in Colorado, USA, on June 3, 2026.
Paper Title: EgoX: Egocentric Video Generation from a Single Exocentric Video Paper Link: https://keh0t0.github.io/EgoX/
Meanwhile, this research was supported by the Ministry of Science and ICT through the National Research Foundation of Korea's individual basic research project, "Research on User-Centered Content Generation and Editing Technology through Generative AI," and the Supercomputer No. 5 High-Performance Computing-based R&D Innovation Support project, "Research on Video Filming Viewpoint Conversion Based on Diffusion Models."
2026.02.23 View 454 -
KAIST Awakens dormant immune cells inside tumors to attack cancer
<(From Left) Professor Ji-Ho Park, Dr. Jun-Hee Han from the Department of Bio and Brain Engineering>
Within tumors in the human body, there are immune cells (macrophages) capable of fighting cancer, but they have been unable to perform their roles properly due to suppression by the tumor. KAIST researchers have overcome this limitation by developing a new therapeutic approach that directly converts immune cells inside tumors into anticancer cell therapies.
KAIST (President Kwang Hyung Lee) announced on the 30th that a research team led by Professor Ji-Ho Park of the Department of Bio and Brain Engineering has developed a therapy in which, when a drug is injected directly into a tumor, macrophages already present in the body absorb it, produce CAR (a cancer-recognizing device) proteins on their own, and are converted into anticancer immune cells known as “CAR-macrophages.”
Solid tumors—such as gastric, lung, and liver cancers—grow as dense masses, making it difficult for immune cells to infiltrate tumors or maintain their function. As a result, the effectiveness of existing immune cell therapies has been limited.
CAR-macrophages, which have recently attracted attention as a next-generation immunotherapy, have the advantage of directly engulfing cancer cells while simultaneously activating surrounding immune cells to amplify anticancer responses.
However, conventional CAR-macrophage therapies require immune cells to be extracted from a patient’s blood, followed by cell culture and genetic modification. This process is time-consuming, costly, and has limited feasibility for real-world patient applications.
To address this challenge, the research team focused on “tumor-associated macrophages” that are already accumulated around tumors.
They developed a strategy to directly reprogram immune cells in the body by loading lipid nanoparticles—designed to be readily absorbed by macrophages—with both mRNA encoding cancer-recognition information and an immunostimulant that activates immune responses.
In other words, in this study, CAR-macrophages were created by “directly converting the body’s own macrophages into anticancer cell therapies inside the body.”
<Figure . Schematic illustration of the strategy for in vivo CAR-macrophage generation and cancer cell eradication via co-delivery of CAR mRNA and immunostimulants using lipid nanoparticles (LNPs)>
When this therapeutic agent was injected into tumors, macrophages rapidly absorbed it and began producing proteins that recognize cancer cells, while immune signaling was simultaneously activated. As a result, the generated “enhanced CAR-macrophages” showed markedly improved cancer cell–killing ability and activated surrounding immune cells, producing a powerful anticancer effect.
In animal models of melanoma (the most dangerous form of skin cancer), tumor growth was significantly suppressed, and the therapeutic effect was shown to have the potential to extend beyond the local tumor site to induce systemic immune responses.
Professor Ji-Ho Park stated, “This study presents a new concept of immune cell therapy that generates anticancer immune cells directly inside the patient’s body,” adding that “it is particularly meaningful in that it simultaneously overcomes the key limitations of existing CAR-macrophage therapies—delivery efficiency and the immunosuppressive tumor environment.”
This research was led by Jun-Hee Han, Ph.D., of the Department of Bio and Brain Engineering at KAIST as the first author, and the results were published on November 18 in ACS Nano, an international journal in the field of nanotechnology.
※ Paper title: “In Situ Chimeric Antigen Receptor Macrophage Therapy via Co-Delivery of mRNA and Immunostimulant,” Authors: Jun-Hee Han (first author), Erinn Fagan, Kyunghwan Yeom, Ji-Ho Park (corresponding author), DOI: 10.1021/acsnano.5c09138
This research was supported by the Mid-Career Researcher Program of the National Research Foundation of Korea.
2025.12.31 View 2501 -
KAIST-KakaoBank Speeds Up 'Explainable AI' by 11 Times: "Boosts Financial AI Reliability
< (From left) Professor Jaesik Choi of the Kim Jaechul Graduate School of AI, Ph.D candidate Chanwoo Lee, Ph.D candidate Youngjin Park >
The research team led by Professor Jaesik Choi of KAIST's Kim Jaechul Graduate School of AI, in collaboration with KakaoBank Corp, announced that they have developed an accelerated explanation technology that can explain the basis of an Artificial Intelligence (AI) model's judgment in real-time. This research achievement significantly increases the practical applicability of Explainable Artificial Intelligence (hereinafter XAI) technology in fields requiring real-time decision-making, such as financial services, by achieving an average processing speed 8.5 times faster, and up to 11 times faster, than existing explanation algorithms for AI model predictions.
In the financial sector, a clear explanation for decisions made by AI systems is essential. Especially in services directly related to customer rights, such as loan screening and anomaly detection, regulatory demands to transparently present the basis for the AI model's judgment are increasingly stringent. However, conventional Explainable Artificial Intelligence (XAI) technologies required the repeated calculation of hundreds to thousands of baselines to generate accurate explanations, resulting in massive computational costs. This was a major factor limiting the application of XAI technology in real-time service environments.
To address this issue, Professor Choi's research team developed the 'ABSQR (Amortized Baseline Selection via Rank-Revealing QR)' framework for accelerating explanation algorithms. ABSQR noticed that the value function matrix generated during the AI model explanation process has a low-rank structure. It introduced a method to select only a critical few baselines from the hundreds available. This drastically reduced the computation complexity, which was previously proportional to the number of baselines, to be proportional only to the number of selected critical baselines, thereby maximizing computational efficiency while maintaining explanatory accuracy.
Specifically, ABSQR operates in two stages. The first stage systematically selects important baselines using Singular Value Decomposition (SVD) and Rank-Revealing QR decomposition techniques. Unlike existing random sampling methods, this is a deterministic selection method aimed at preserving information recovery, which guarantees the accuracy of the explanation while significantly reducing computation. The second stage introduces an amortized inference mechanism, which reuses the pre-calculated weights of the baselines through cluster-based search, allowing the system to provide an explanation for the model's prediction result in real-time service environments without repeatedly evaluating the model. The research team verified the superiority of ABSQR through experiments on various real-world datasets. Tests on standard datasets across five sectors—finance, marketing, and demographics—showed that ABSQR achieved an average processing speed 8.5 times faster than existing explanation algorithms that use all baselines, with a maximum speed improvement of over 11 times. Furthermore, the degradation of explanatory accuracy due to speed acceleration was minimized, maintaining up to 93.5% of the explanation accuracy compared to the baseline algorithm. This level is sufficient to meet the explanation quality required in real-world applications.
< ABSQR Framework Overview. (1) The baseline selection stage utilizes the low-rank structure of the value function matrix to select only a small number of key baselines, and (2) the accelerated search stage reuses the pre-calculated baseline weight coefficients based on clusters. This dramatically reduces the computation complexity, which was proportional to the number of baselines, to be proportional only to the number of selected key baselines. >
A KakaoBank official stated, "We will continue relentless research and development to enhance the reliability and convenience of financial services and introduce innovative financial technologies that customers can experience." Chanwoo Lee and Youngjin Park, co-first authors from KAIST, explained the significance of the research: "This methodology solves the crucial acceleration problem for real-time application in the financial sector, proving that it is possible to provide users with the reasons behind a learning model's decision in real-time." They added, "This research provides new insights into what constitutes unnecessary computation and the selection of important baselines in explanation algorithms, practically contributing to the improvement of explanation technology efficiency." This research, co-authored by PhD candidates Chanwoo Lee and Youngjin Park from the KAIST Kim Jaechul Graduate School of AI, and researchers Hyeongeun Lee and Yeeun Yoo from the KakaoBank Financial Technology Research Institute, was presented on November 12 at the 'CIKM 2025 (ACM International Conference on Information and Knowledge Management)', the world's highest-authority academic conference in the field of information and knowledge management. ※ Paper Title: Amortized Baseline Selection via Rank-Revealing QR for Efficient Model Explanation
※ Author Information:
※ Author Information: DOI: https://doi.org/10.1145/3746252.3761036
Co-First Authors: Chanwoo Lee (KAIST Kim Jaechul Graduate School of AI), Youngjin Park (KAIST Kim Jaechul Graduate School of AI), Hyeogeun Lee (KakaoBank), Yeeun Yoo (KakaoBank)
Co-Authors: Daehee Han (KakaoBank), Junho Choi (KAIST Kim Jaechul Graduate School of AI), Kunhyung Kim (KAIST Kim Jaechul Graduate School of AI)
Corresponding Authors: Nari Kim (KAIST Kim Jaechul Graduate School of AI), Jaesik Choi (KAIST Kim Jaechul Graduate School of AI)
Meanwhile, this research achievement was conducted through KakaoBank's industry-academia research project 'Advanced Research on Explainable Artificial Intelligence Algorithms in the Financial Sector' and the Ministry of Science and ICT/Institute for Information & Communications Technology Planning and Evaluation (IITP) supported project 'Development of Explainable Artificial Intelligence Technology Providing Explainability in a Plug-and-Play Manner and Verification of Explanation Provision for AI Systems.'
2025.12.12 View 1699 -
KAIST Predicts Human Group Behavior with AI! 1st Place at the World’s Top Conference… Major Success after 23 Years
<(From Left) Ph.D candidate Geon Lee, Ph.D candidate Minyoung Choe, M.S candidate Jaewan Chun, Professor Kijung Shin, M.S candidate Seokbum Yoon>
KAIST (President Kwang Hyung Lee) announced on the 9th of December that Professor Kijung Shin’s research team at the Kim Jaechul Graduate School of AI has developed a groundbreaking AI technology that predicts complex social group behavior by analyzing how individual attributes such as age and role influence group relationships.
With this technology, the research team achieved the remarkable feat of winning the Best Paper Award at the world-renowned data mining conference “IEEE ICDM,” hosted by the Institute of Electrical and Electronics Engineers (IEEE). This is the highest honor awarded to only one paper out of 785 submissions worldwide, and marks the first time in 23 years that a Korean university research team has received this award, once again demonstrating KAIST’s technological leadership on the global research stage.
Today, group interactions involving many participants at the same time—such as online communities, research collaborations, and group chats—are rapidly increasing across society. However, there has been a lack of technology that can precisely explain both how such group behavior is structured and how individual characteristics influence it at the same time.
To overcome this limitation, Professor Kijung Shin’s research team developed an AI model called “NoAH (Node Attribute-based Hypergraph Generator),” which realistically reproduces the interplay between individual attributes and group structure.
NoAH is an artificial intelligence that explains and imitates what kinds of group behaviors emerge when people’s characteristics come together. For example, it can analyze and faithfully reproduce how information such as a person’s interests and roles actually combine to form group behavior.
As such, NoAH is an AI that generates “realistic group behavior” by simultaneously reflecting human traits and relationships. It was shown to reproduce various real-world group behaviors—such as product purchase combinations in e-commerce, the spread of online discussions, and co-authorship networks among researchers—far more realistically than existing models.
< The process of generating group interactions using NoAH >
Professor Kijung Shin stated, “This study opens a new AI paradigm that enables a richer understanding of complex interactions by considering not only the structure of groups but also individual attributes together,” and added, “Analyses of online communities, messengers, and social networks will become far more precise.”
This research was conducted by a team consisting of Professor Kijung Shin and KAIST Kim Jaechul Graduate School of AI students: master’s students Jaewan Chun and Seokbum Yoon, and doctoral students Minyoung Choe and Geon Lee, and was presented at IEEE ICDM on November 18.
※ Paper title: “Attributed Hypergraph Generation with Realistic Interplay Between Structure and Attributes” Original paper: https://arxiv.org/abs/2509.21838
< Photo from the award ceremony held on November 14 at the International Spy Museum in Washington, D.C.>
Meanwhile, including this award-winning paper, Professor Shin’s research team presented a total of four papers at IEEE ICDM this year. In addition, in 2023, the team also received the Best Student Paper Runner-up (4th place) at the same conference.
This work was supported by Institute of Information & Communications Technology Planning & Evaluation (IITP) grant funded by the Korea government (MSIT) (No. RS-202400457882, AI Research Hub Project) (RS-2019-II190075, Artificial Intelligence Graduate School Program (KAIST)) (No. RS-2022-II220871, Development of AI Autonomy and Knowledge Enhancement for AI Agent Collaboration).
2025.12.09 View 2112 -
How Does AI Think? KAIST Achieves First Visualization of the Internal Structure Behind AI Decision-Making
<(From Left) Ph.D candidate Daehee Kwon, Ph.D candidate Sehyun lee, Professor Jaesik Choi>
Although deep learning–based image recognition technology is rapidly advancing, it still remains difficult to clearly explain the criteria AI uses internally to observe and judge images. In particular, technologies that analyze how large-scale models combine various concepts (e.g., cat ears, car wheels) to reach a conclusion have long been recognized as a major unsolved challenge.
KAIST (President Kwang Hyung Lee) announced on the 26th of November that Professor Jaesik Choi’s research team at the Kim Jaechul Graduate School of AI has developed a new explainable AI (XAI) technology that visualizes the concept-formation process inside a model at the level of circuits, enabling humans to understand the basis on which AI makes decisions.
The study is evaluated as a significant step forward that allows researchers to structurally examine “how AI thinks.”
Inside deep learning models, there exist basic computational units called neurons, which function similarly to those in the human brain. Neurons detect small features within an image—such as the shape of an ear, a specific color, or an outline—and compute a value (signal) that is transmitted to the next layer.
In contrast, a circuit refers to a structure in which multiple neurons are connected to jointly recognize a single meaning (concept). For example, to recognize the concept of cat ear, neurons detecting outline shapes, neurons detecting triangular forms, and neurons detecting fur-color patterns must activate in sequence, forming a functional unit (circuit).
Up until now, most explanation techniques have taken a neuron-centric approach based on the idea that “a specific neuron detects a specific concept.” However, in reality, deep learning models form concepts through cooperative circuit structures involving many neurons. Based on this observation, the KAIST research team proposed a technique that expands the unit of concept representation from “neuron → circuit.”
The research team’s newly developed technology, Granular Concept Circuits (GCC), is a novel method that analyzes and visualizes how an image-classification model internally forms concepts at the circuit level.
GCC automatically traces circuits by computing Neuron Sensitivity and Semantic Flow. Neuron Sensitivity indicates how strongly a neuron responds to a particular feature, while Semantic Flow measures how strongly that feature is passed on to the next concept. Using these metrics, the system can visualize, step-by-step, how basic features such as color and texture are assembled into higher-level concepts.
The team conducted experiments in which specific circuits were temporarily disabled (ablation). As a result, when the circuit responsible for a concept was deactivated, the AI’s predictions actually changed.
In other words, the experiment directly demonstrated that the corresponding circuit indeed performs the function of recognizing that concept.
This study is regarded as the first to reveal, at a fine-grained circuit level, the actual structural process by which concepts are formed inside complex deep learning models. Through this, the research suggests practical applicability across the entire explainable AI (XAI) domain—including strengthening transparency in AI decision-making, analyzing the causes of misclassification, detecting bias, improving model debugging and architecture, and enhancing safety and accountability.
The research team stated, “This technology shows the concept structures that AI forms internally in a way that humans can understand,” adding that “this study provides a scientific starting point for researching how AI thinks.”
Professor Jaesik Choi emphasized, “Unlike previous approaches that simplified complex models for explanation, this is the first approach to precisely interpret the model’s interior at the level of fine-grained circuits,” and added, “We demonstrated that the concepts learned by AI can be automatically traced and visualized.”
< Overview of the Conceptual Circuit Proposed by the Research Team >
This study, with Ph.D. candidates Dahee Kwon and Sehyun Lee from KAIST Kim Jaechul Graduate School of AI as co–first authors, was presented on October 21 at the International Conference on Computer Vision (ICCV).
Paper title: Granular Concept Circuits: Toward a Fine-Grained Circuit Discovery for Concept Representations
Paper link: https://openaccess.thecvf.com/content/ICCV2025/papers/Kwon_Granular_Concept_Circuits_Toward_a_Fine-Grained_Circuit_Discovery_for_Concept_ICCV_2025_paper.pdf
This research was supported by the Ministry of Science and ICT and the Institute for Information & Communications Technology Planning & Evaluation (IITP) under the “Development of Artificial Intelligence Technology for Personalized Plug-and-Play Explanation and Verification of Explanation” project, the AI Research Hub Project, and the KAIST AI Graduate School Program, and was carried out with support from the Defense Acquisition Program Administration (DAPA) and the Agency for Defense Development (ADD) at the KAIST Center for Applied Research in Artificial Intelligence.
2025.11.26 View 2499 -
Failure in the AI Era? The 3rd Failure Conference Held
< 2025 Failure Conference Poster >
KAIST announced on the 31st of October that it will be holding the '3rd Failure Conference' from Wednesday, November 5th to Friday, November 14th. The event is organized by the KAIST Center for Ambitious Failure (Director Sungho Jo), and, under the theme 'AI times Failure,' it will re-examine the value of humaneness through the sensibility of 'failure' in this era of great transformation led by AI technology.
Composed of lectures, competitions, exhibitions, and networking programs, this conference provides a venue for new introspection on the relationship between humanity, society, and technology through the lens of 'failure.'
Failure Seminar 'AI Era, Asking the Way of Humanity' will be held on November 6th at the Jeong Geun-mo Conference Hall in the Academic and Cultural Complex
Professor Juho Kim of the KAIST School of Computing will discuss the human sensibility and resilience needed in the AI era through the paradox that "AI learns how to fail less, but humans are losing the opportunity to fail. Following this, Professor Sang Wook Lee of the Hanyang University Department of Philosophy will present philosophical and ethical challenges and practical directions for the advancement of AI technology to lead to universal welfare for humanity. The 'AI times Failure Idea Contest' Finals will take place on November 7th at the John Hanner Hall in the Academic and Cultural Complex. 12 teams, selected from preliminaries that included 111 teams from universities and graduate schools nationwide, will demonstrate their ideas in booth form on the theme of 'The Future where AI and Humans Coexist.' Participants will explore AI errors, human limitations, and the possibility of trust and recovery, presenting attempts to convert technological failure into human introspection, and human failure into technological possibility. On the day of the finals, the Grand Prize (KAIST President’s Award), First Prize, and Second Prize will be selected through judging.
The Photography Exhibition '404: Perfection Not Found' will be held on the 1st floor of the Creative Learning Building from November 5th to 14th. This exhibition showcases 'Scenes of Imperfection' captured by KAIST members through the PhotoVoice program and the AI times Failure Snapshot Challenge. It is divided into three sections: ▲ Brain that Mimics Perfection: Failure of AI ▲ Incomplete Connection: Portrait of the Digital Generation ▲ Aesthetics of Imperfection: Warmth of Humanity, providing a space for introspection that illuminates human responsibility and potential through technological failure. The 'Show Off Your Failed Project Contest,' which has garnered great response from KAIST students every year, will be expanded to include general public participation on the 5th at the John Hanner Hall in the Academic and Cultural Complex. Co-planned by the KAIST Center for Ambitious Failure and the student club ICISTS, participants will decorate their own booths with photos and videos to share their failures and the process of overcoming them. Awards such as ▲ Best (Most Votes) ▲ Shining Debris Award (Highly Relatable Failure Story) ▲ Flower of Ash Award (Overcoming Story) ▲ Aesthetics of Failure Award (Creative Expression) ▲ Beautiful Afterimage Award (Sincere Lingering Impression) will be selected through audience voting.
< 2025 Show Off Your Failed Project Contest Poster >
Sungho Jo, KAIST Center for Ambitious Failure (Professor, School of Computing), stated, "As AI technology rapidly evolves and changes the order of the world, humans need to look back at themselves beyond that speed. I hope this Failure Conference will be an opportunity to rediscover the meaning of humaneness amid technological innovation and to imagine a better future." Kwang Hyung Lee, President of KAIST, said, "Failure is another name for challenge, and a seed of innovation. KAIST will lead the AI era and human-centered technological development through a creative spirit of challenge that is not afraid of failure."
All programs for the 2025 Failure Conference are open to anyone interested, and detailed schedules and content can be checked on the webstie of KAIST Center for Ambitious Failure (caf.kaist.ac.kr).
2025.10.29 View 1562 -
Robot-Operated Space Station Construction Goal... 'In-space Servicing and Manufacturing Research Center' Launched
<Plaque Handover Ceremony. (From left) Jae-Hung Han, Director of the Space Research Institute, Ju-won Kang, Head of Engineering Group at the National Research Foundation of Korea Basic Research Headquarters>
KAIST's Space Research Institute announced on the 24th of October that it officially launched the 'Innovative Research Center for the Development of Core Technologies in In-space Servicing and Manufacturing (ISMRC)' at the KAIST Academic Cultural Center on Friday, October 24. About 150 officials from major organizations, including the Korea Aerospace Administration, the National Research Foundation of Korea, and Daejeon Metropolitan City, as well as domestic and foreign space experts, attended the opening ceremony to discuss future cooperation measures. The 'KAIST In-space Servicing and Manufacturing Research Center (ISMRC)' is a large-scale research hub selected for the Ministry of Science and ICT's 2025 Basic Research Project, with a total of 71.2 billion KRW long-term project planned over the next 10 years, including 50 billion KRW in national funding. Daejeon City will also provide a total of 3.6 billion KRW, with 400 million KRW annually starting from 2026. The research goals are to secure core technologies for next-generation space exploration, including: ▲ Construction of Unmanned Space Stations, ▲ Robotics-based In-space Manufacturing, and ▲ Resource Recovery Technology. A team of 14 KAIST professors, led by Director Jae-Hung Han, will spearhead the research, with major domestic and foreign space companies and research institutions participating in joint research. As the 'New Space' era fully commences globally, the In-space Servicing and Manufacturing industry is projected to grow to tens of trillions of Korean won by 2030, driven by the reduction of launch costs and the expansion of private sector participation. This field is evaluated as a core area that will fundamentally change the way humanity engages in space activities, including extending satellite lifespan, on-orbit maintenance and operation, and securing and manufacturing resources in space. Meanwhile, an international symposium was held for two days on October 23-24 at the KAIST Academic Cultural Center and KI Building, coinciding with the opening ceremony.
<Director Jae-Hung Han of the Space Research Institute presenting>
The symposium was composed of a total of six sessions, including: ▲ Exchange Meeting on Additive Manufacturing Tecnology for Aerospace, ▲ International Workshop on Aerospace Composites, ▲ Workshop on Swarm Satellite Development, and ▲ Workshop on In-space Servicing and Manufacturing Robotics. Major domestic and foreign institutions and experts, including the Korea Aerospace Research Institute, Japan Advanced Institute of Science and Technology, and California Institute of Technology (Caltech), attended to discuss the future direction of next-generation space technology development and international cooperation measures. Cheol-woong Son, Director-General of Future Strategy Industry Office at Daejeon City, said, "We will develop the Innovative Research Center into a Daejeon-type space industry innovation platform with KAIST," and "Daejeon City will concentrate its capabilities to help local businesses grow and establish Daejeon as the central city for the Republic of Korea's space industry." Jae-Hung Han, Director of the KAIST Space Research Institute, said, "We will lead the core technologies for in-space servicing and manufacturing through cooperation between industry, academia, research institutes, and government, and contribute to the establishment of a private sector-focused industrial ecosystem," adding, "KAIST will grow into a comprehensive research hub that encompasses R&D, talent nurturing, and technology commercialization."
<Group Photo of Participants at the Opening Ceremony of the In-space Servicing and Manufacturing Research Center>
Kwang Hyung Lee, President of KAIST, said, "The field of in-space servicing and manufacturing is a core area that will change the paradigm of the future space industry," and "KAIST will lead the Republic of Korea to become the center for opening a new era of the space industry through innovative technology development and global cooperation." KAIST plans to perform the role of breaking down the boundaries between academia and industry, focusing on these technologies, and laying the foundation for next-generation space activities.
2025.10.24 View 2995