A team of South Korean engineers from KAIST University and DGIST Institute of Science and Technology has created an artificial taste sensor that mirrors the way tongue receptors detect flavors, powered by advanced AI technologies. The discovery is published in the scientific journal ACS Applied Materials & Interfaces (ACSAMI).
The researchers employed four distinct sensors to gauge saltiness, sourness, bitterness, and sweetness, paired with a bespoke machine learning algorithm designed for flavor analysis. The system processes patterns across sensors to build a comprehensive taste profile for each sample, enabling precise flavor decoding without human tasters.
In tests, the scientists evaluated six different wines, building a dataset that maps the taste characteristics of each variety. Using what is described as an electronic language, an AI sommelier classified recommended drinks with over 95 percent accuracy and proposed options for wines with similar flavor fingerprints. This approach demonstrates how AI-driven taste analysis can assist in selecting and pairing beverages with consistent sensory quality.
The technology is anticipated to find applications across multiple industries, including food, beverage, cosmetics, and pharmaceuticals. It offers the potential not only to measure taste but also to support quality control processes, ensuring product consistency and safeguarding consumer experience across production lines.
Historically, similar innovations have appeared elsewhere, including an early Russian effort to develop an electronic nose aimed at quality control for packaging materials used with edible products. Such parallel developments illustrate the growing interest in AI-enabled sensory analysis for diverse applications, from flavor evaluation to packaging integrity and product safety. [Citation: KAIST DGIST study]”