AI will pinpoint the origin of Chateau Bordeaux wines with 100% accuracy.

Machine learning has been able to determine the origin of 80 red Bordeaux wines with 100 percent accuracy by analyzing their chemical structure. Researchers from the University of Geneva in Switzerland, together with Alexandre Puget, examined the chemical composition of 80 wines from 12 years of vintages between 1990 and 2007. All the wines were produced at seven wineries in the Bordeaux region of France.

The aim of the study was to find out if there were chemical distinctive characteristics of each château, regardless of the year of production. This means that wines from the same winery will have very similar chemical profiles and therefore the same flavor from year to year.

Scientists used a machine to turn each wine into a vapor and separate it at thechemical components. In this way, they obtained an analysis of each wine, called a chromatogram, with about 30,000 points representing different chemical compounds.

The researchers used 73 chromatograms to train a machine learning algorithm, along with data on the winery's origin and year of production. They then tested the algorithm on seven chromatograms that they held.

They repeated the process 50 times, each time changing the wines used.

Even with only 5 percent of each chromatogram, the algorithm was still able to identify the winery. This suggests that.wine's unique flavor and mouth feel is not dependent on a few key molecules, but on the overall concentration of many molecules.

The algorithm was also able to separate wines into groups that are more similar to each other. It separated wines from the right bank of the Garonne, such as Pomerol and Saint-Emilion, from those from the left bank, known as Medoc.

This work provides further evidence that location, climate, microbes and production methods do contribute uniquely to the flavor of wine. However, the specific chemical compounds responsible for each wine were not addressed in this study.