Turing's EMNLP 2025 Paper Accepted

Turing, a developer of inference AI agents, has proven its AI technology prowess with the acceptance of its paper at 'EMNLP 2025', one of the most prestigious conferences in natural language processing. Notably, it was selected for the Industry Track section, which emphasizes practical applicability, putting it on par with global leading companies such as Google, Apple, and Nvidia. The research paper, ‘Computational Blueprints: Generating Isomorphic Math Problems with Large Language Models’, led by the Turing AI Team (Team Leader Nam Jin-woo, Researcher Kim Jeong-hoon, Researcher Jo Geun-sik), proposes 'CBIT', a framework for generating isomorphic math problems using Large Language Models (LLMs). To overcome the limitations of applying existing LLM-generated problems in educational settings, the Turing research team focused on verifying the practical effectiveness of math problems generated through CBIT. In experiments conducted with students from 4th grade elementary school to 3rd grade high school, CBIT demonstrated superior efficiency by significantly improving problem generation accuracy and reducing costs compared to existing methods. Furthermore, an analysis of over 180,000 cumulative interaction records showed that the error reporting rate for AI-generated problems was 20% lower than for problems created by experts, objectively proving their quality and reliability. This suggests the feasibility of an AI system that not only recommends problems but directly generates and verifies high-quality problems. Based on this technological capability, Turing successfully launched 'GPAI', an AI agent specialized in STEM fields, in August, establishing it successfully in the Indian and U.S. markets. Turing aims to contribute to STEM learning through AI-driven personalized education, and further plans to develop as a research tool by adding AI-based engineering simulation and modeling capabilities. Choi Min-kyu, CEO of Turing, reaffirmed Turing's inference AI agent technology through the acceptance of this EMNLP paper, expressing his ambition to accelerate the company's global market penetration based on its AI capabilities specialized in logic, inference, and mathematics.