[Blog] Biofortis Research

Beverage Taste Testing with Artificial Intelligence | Biofortis

Written by Biofortis Research | Dec 6, 2022 9:00:58 AM

Artificial intelligence (AI) has been handling many human tasks for years. From automating manufacturing processes to serving us products and providing entertainment, & service recommendations, AI is advancing at a rapid pace.

It seems only natural that AI and related algorithms would eventually start to tackle tasks that involve human senses. Taste, however, seems like a difficult one. Before we look at how a machine can conduct a taste test, let’s look at AI in general.

What is Artificial Intelligence?

The debate for what constitutes true AI is ongoing. Most people think of AI as a machine that can think for itself, thanks to its depiction in countless science fiction books, movies, TV shows, & video games.

AI as it is used in most industries today, however, is more involved in managing data.  In particular, AI current-generation AI can process data exponentially faster than humans, making AI a preferred tool for mining massive amounts of data and culling the information that is key to a particular subject.  AI programs are valuable commodities.

A process called machine learning allows AI to ‘learn’ from historical data as well as hypothetical data, thus changing as it learns. Machine learning is a relatively easy way to get an AI algorithm up to speed and functioning in a useful manner.

How Can AI Taste?

The short answer is that AI can’t taste. Machine learning, however, has allowed AI programs to classify beverages by certain categories, based on composition and processing information, amounting to the AI version of a taste test. The most use of ‘taste testing AI’ currently concerns alcoholic beverages.

In the last decade, alcoholic beverages, specifically spirits, have been categorized by AI using models including:

  • Age
  • Cask material
  • Distillery
  • Variety

While this may seem simple, the AI has been able to use data gathered from these models to pinpoint where a spirit was made with astonishing accuracy. A study on tequila noted the AI was 89% accurate when it came to asking where a sample came from.

While the current state of AI in alcoholic beverage testing is exciting, it will still be a while before an algorithm can make recommendations based on individual tastes. However, the future holds great promise as chemical composition data is being compiled at a rapid rate, and AI will, someday, be able to use the chemical composition of beverages as well to characterize taste profiles. 

Non-Alcoholic Beverage Taste-Testing

While the use of AI taste testing for alcoholic beverages has steadily increased over the past decade, it has yet to see widespread use for non-alcoholic beverages. The exception is water-quality testing, a field where AI can test water quality with accuracy nearing 100%.

Taste-testing algorithms are being explored for non-alcoholic beverages such as milk and other dairy products. Research into using AI to determine the quality of milk is ongoing.

Taste testing, however, is not the same as testing the quality of water used in a beverage. The reasons for the lower usage of AI taste testing in non-alcoholic beverages is unknown. Potential reasons include the value of older wines & spirits, the authenticity of expensive alcoholic beverages, and the quality of the alcohol.

The Future of AI Taste Testing

While AI taste testing is making progress each year, it currently has a ways to go before it will be recommending drinks to people or creating popular new beverages. The chief concern of contemporary researchers studying AI & taste testing revolves around biometrics. That is, although AI can help classify beverages, understanding the science behind the differences in the biology of taste among different people is still in its infancy.  With the focus on personalized nutrition, though, research on how people taste, including the biochemistry behind taste receptors in the body, is advancing.  Researchers say advancements in robotics and biometric technology will eventually allow more accurate AI taste technology.

Technology related to human senses in general needs to reach higher levels before broader applications can be properly utilized. The current technology, however, is exciting and can be used in tandem with research data from human taste testers to classify beverages more accurately than ever before.

Research at Biofortis

Biofortis is dedicated to protecting consumer health worldwide by delivering a wide range of testing and consultancy services to the food, supplement, and nutrition industries. Biofortis supports this mission in two ways—through clinical trials and sensory and consumer insights testing. We specialize in clinical research targeting foods, ingredients, and dietary supplements that affect body structures, function, and overall health. Contact us with any clinical trial or scientific consulting needs.