Selectstar announced that it should grow right into a ‘total AI service company’ that’s accountable for all stages of artificial intelligence (AI) introduction, from data design to large language model (LLM) verification. The core of this was the AI reliability evaluation field.
Kim Se-yeop, CEO of Select Star, said in his Latest 12 months’s address on the 4th, “This yr, based on data-driven artificial intelligence (AI) technology, we’re preparing to ▲ revitalize the AI reliability evaluation business ▲ hold the MWC Red Team Challenge ▲ expand the AI character business.”
First, CEO Kim said, “As generative AI is introduced throughout the business, securing LLM stability and reliability has grow to be essential,” and “Selectstar will give attention to dominating the AI reliability assessment market.”
To this end, we plan to launch LLM reliability assessment consulting and LLM reliability assessment software-as-a-service (SaaS) platform. This can be a customized service that designs an AI model evaluation index method by considering the service purpose and characteristics of every industry.
SelectStar participated because the organizer of the ‘Ultra-Large Language Model Reliability Benchmark Data’ hosted by the Korea Intelligence and Information Society Agency (NIA) last yr and has a history of constructing a benchmark dataset that may quantitatively evaluate LLM reliability.
Based on this, now we have been providing consulting support to domestic electronics and financial firms to introduce and utilize LLM and to secure reliability, and we plan to proceed to strive to supply consulting support to varied industries this yr.
Moreover, we plan to launch a SaaS platform that automates the evaluation process in the worldwide market. It includes functions to robotically generate evaluation questions and define evaluation indicators and standards to judge the standard and safety of LLM specialized for every company.
Next, a worldwide red team event shall be held on the Mobile World Congress (MWC) held in Spain in March. It is alleged to be the primary offline global red team challenge held by the Global Mobile Communications Association (GSMA) and Selectstar, and is receiving global attention.
This event is a mixture of the red team know-how accrued by Select Star and global partnerships. We plan to utilize Select Star’s red team technology experience and know-how, including scenario design and solution utilization.
As well as, we’re expanding our B2C business. A representative example is the recent application of AI character technology to the ‘Teeniping Cake’ and support for the production technology of video letters read directly by Hachuping.
Various characters and message options shall be added in the long run. In addition they plan to expand their business into character goods and K-pop AI goods incorporating AI technology.
CEO Kim Se-yeop said, “Evaluation and reliability verification of LLM are key issues in 2025.” He added, “It just isn’t nearly improving accuracy, but all-round verification that encompasses various issues comparable to meeting ethical and legal standards, eliminating bias, and strengthening data security.” “Work shall be required,” he predicted.
He continued, “Firms that use or provide AI will now haven’t any selection but to internalize LLM reliability verification and red team activities,” and added, “Through reliability evaluation and automation solutions that might be used on the worldwide stage, we are able to convey the worth of AI introduction to customers.” “We’ll deliver and achieve continuous growth,” he said.
Reporter Park Soo-bin sbin08@aitimes.com
