Shiseido announced that, as a result of a PoC (Proof of Concept) utilizing Google Cloud's generative AI, it has been able to reduce the time required for ingredient discovery by 95%.
Shiseido Reduces Ingredient Discovery Time by 95% Using Generative AI, Replicating Researcher "Expertise" via Multiple LLMs
This article is a translation. Read the Japanese original
The company has been working on an initiative to replicate the "expert judgment" of ingredients—which researchers have traditionally performed based on years of experience and expertise—using multiple Large Language Models (LLMs). This system analyzes vast amounts of papers and existing knowledge to identify ingredients suitable for specific purposes.
Through this initiative, the ingredient discovery process, which previously required significant time, has been greatly shortened. Shiseido stated that it aims to further accelerate product development by deploying this technology across various research and development processes in the future.