Key Findings
At ICML 2026 (International Conference on Machine Learning 2026), LG AI Research unveiled ‘EXAONE Discovery,’ an industrial AI application built upon its Large Language Model (LLM) ‘EXAONE,’ showcasing its groundbreaking real-world applications. EXAONE Discovery is an AI platform designed to automatically extract molecular structures from scientific papers and propose new candidate compounds in response to complex researcher queries. The system has already yielded tangible results, successfully discovering a new hair growth ingredient named ‘Rhamsydil’ and identifying an immersion coolant for optimizing AI data center efficiency. This platform holds the potential to dramatically accelerate R&D across diverse fields, including cosmetics, batteries, semiconductor materials, and novel drug candidates.
Technical / Clinical Details
The core of EXAONE Discovery lies in the EXAONE LLM’s vast scientific knowledge comprehension and reasoning capabilities. This LLM analyzes unstructured data (text, figures, tables, etc.) related to materials science, chemistry, and biology from global scientific literature and databases, extracting highly accurate information on molecular structures, reaction conditions, and properties. When researchers are seeking materials or molecules with specific functionalities (e.g., hair growth effects, high thermal conductivity), EXAONE Discovery designs and proposes novel molecular structures previously unknown, based on its existing knowledge base and learned patterns. For instance, in the discovery of the new hair growth ingredient Rhamsydil, the LLM identified molecules with specific structural features that could influence hair growth-related biological pathways from a multitude of candidates. Similarly, for the immersion coolant for AI data centers, the LLM proposed optimal liquid compositions by considering multiple physicochemical properties such as thermal conductivity, electrical insulation, and stability. These candidates subsequently underwent laboratory synthesis and validation processes to confirm their efficacy. EXAONE Discovery efficiently facilitates this ‘prediction -> synthesis -> validation -> learning’ closed-loop cycle, significantly shortening the material discovery timeline.
Background & Context
The development of new materials and drugs is a critical driver of technological innovation and economic growth in modern society. However, traditional R&D processes have historically been challenged by their high time, cost, and resource requirements, coupled with low success rates. Molecular design and material optimization, in particular, necessitate exploring a vast chemical space, which is often beyond the limits of human expert knowledge alone. Recent advancements in generative AI, especially LLMs, are beginning to provide powerful solutions to this challenge. LG AI Research’s EXAONE Discovery stands at the forefront of such AI-driven material discovery, aiming to accelerate competitiveness and innovation by resolving R&D bottlenecks across various industrial sectors. This technology will be an indispensable tool for further enhancing the international competitiveness of industries where South Korea excels, such as batteries, displays, and semiconductors.
Strategic Significance & Outlook
While EXAONE Discovery is still in its early stages of deployment, its practical applications demonstrate how powerful LLMs can be in industrial settings. LG AI Research aims to further expand the platform’s capabilities to address more complex material design challenges and multi-objective optimization problems. Additionally, through collaborations with more companies and research institutions, they plan to broaden its application scope and foster industry-wide innovation. This AI-driven discovery system is expected to play a crucial role in solving various societal challenges, including the development of new cosmetic products, performance enhancement of next-generation batteries, creation of innovative semiconductor materials, and even the discovery of new drug candidates for intractable diseases.
Get our weekly technology intelligence — free
Receive an infographic that lets you judge at a glance whether each field’s analysis report is worth reading.
Subscribe Free — Weekly Tech Intelligence
By subscribing, you’ll receive Troy-Technical’s weekly technology intelligence newsletter.
- Your email and selected fields are used only to deliver the newsletter.
- We never share your information with third parties.
- You can unsubscribe anytime via the link in each email.
See our Privacy Policy for details.
Takes about a minute · Unsubscribe anytime
Comments