Following the recent exposure of its private data, a report claims that Samsung Electronics has begun the extensive creation of its own Large Language Model (LLM) for internal applications. The project has received major manpower and resource expenditures, spearheaded by Samsung Research. Notably, the LLM, a crucial element in training such models, has been developed using all GPU resources available. Samsung wants to finish the first iteration of its LLM in two months.
Samsung’s AI Solution Aims to Slash Time Required for Software Development
Samsung’s LLM solution will mostly be used for software development, language translation, and document summarizing. Although the business hasn’t decided whether to make the AI solution accessible to everyone, its internal use is regarded as a priority. After an unexpected data leak caused by an earlier reliance on external generative AI solutions, such as ChatGPT, Samsung decided to forbid their use and start developing its own proprietary AI technology.
The time needed for software development and semiconductor design, according to Samsung, can be greatly decreased by implementing AI solutions based on LLM technology. Samsung investigated joint ventures with numerous well-known businesses, including Google, Microsoft, Naver, and OpenAI, but finally decided to create an internal AI solution at a top-level management meeting with Lee Jae-yong in attendance.
Numerous South Korean companies are also creating their own LLM-based generative AI products, such as KoGPT, ExaOne, HyperCloverX, and A-Dot. These companies include Kakao, LG, Naver, and SKT. The sketching and writing talents displayed by these solutions are human-like. However, there are still serious issues with data privacy and accuracy that must be resolved in the ongoing development process.