Fortytwomaru Shares Insights on 'Ensuring AI Reliability' at International Symposium

Fortytwomaru Leads TRAIN Symposium Tech Session
Director Park Ju-sik Shares Insights with LLM42 Case
Hallucination Mitigation and Enterprise-Specific Model Application... Ensuring Reliability

Generative AI startup FortyTwoMaru will share insights on securing the reliability of large language models (LLM) at an international symposium.

On the 22nd, FortyTwoMaru announced that Joo-Sik Park, Head of Overseas Business at FortyTwoMaru, will conduct a session titled "Trustworthy LLM: How to" at the TRAIN Symposium held at Gravity Seoul Pangyo in Seongnam, Gyeonggi Province.

The TRAIN Symposium is a global symposium hosted by TRAIN Global and supported by the Busan IT Convergence Components Research Institute (CIDI), Jeonju Information & Culture Industry Promotion Agency (JICA), and Korea Testing Laboratory (KTL), among others.

Over two days, on the 22nd and 23rd, the symposium will share movements to secure and strengthen AI reliability in Southeast Asian countries such as Vietnam, Thailand, and Indonesia, and explore global response strategies. Attendees include domestic and international government and public institutions, AI-related companies, associations, organizations, as well as legal and academic professionals.

On the first day, under the theme "Asian Solidarity for Trustworthy AI ? The Power of Information Exchange," presentations will be held across three sessions covering technology, education, and law, systems, and policy. On the 23rd, the TRAIN General Assembly will take place.

In the technology session, Director Park will address hallucination, security, and cost issues to consider when introducing LLMs, and explain ways to implement trustworthy AI by managing hallucination phenomena. He will share insights through real cases of FortyTwoMaru’s proprietary large-scale language model, LLM42.

FortyTwoMaru, a generative AI startup, has mitigated the hallucination issue?a drawback of LLMs?through engineering that integrates an accuracy-focused AI reading comprehension (MRC)-based QA (Question Answering) model. Additionally, it supports enterprise-specific modes, enabling safe use of ultra-large AI without concerns about internal corporate data or sensitive customer information leaks.

Dong-Hwan Kim, CEO of FortyTwoMaru, stated, "FortyTwoMaru has won the Good AI Awards, which evaluate language AI technology and ethics, for two consecutive years and received the 1st AI Reliability and Quality Award last year. We will continue to lead the generative AI market as a pioneering company that guarantees both technological capability and quality reliability."