To those of us who taste wine for a living, it was both fascinating and alarming to read last week’s news that researchers in Switzerland have trained an artificial intelligence program to identify Bordeaux wines from specific estates and vintages with 100% certainty.
In the study, published in the journal Communications Chemistry, scientists at the University of Geneva used machine learning to analyze the chemical composition of 80 red wines from 12 vintages between 1990 and 2007. “We were interested in finding out whether there is a chemical signature that is specific to each of those chateaux that’s independent of vintage,” says lead researcher Alexandre Pouget in the New Scientist. They asked a question at the core of wine knowledge: Does an individual chateau’s wine have a similar chemical profile—and therefore taste—year after year?
Pouget and his colleagues vaporized the 80 wines and used chromatography to separate and catalog the wines’ chemical compounds. Each wine’s readout, or chromatogram, had 30,000 points representing different chemical compounds. The researcher then used 73 of the wine’s chromatograms to train a machine learning algorithm, along with data on the estate and vintage. Finally, they tested the A.I. algorithm on the remaining seven wines—50 times each in various orders. The A.I. blind tasting algorithm identified the wines 100% of the time. The algorithm also was able to group the wines based on whether they were from Bordeaux’s Left or Right Bank.
You May Also Like: Producers Feared the 2022 Bordeaux Vintage Would Be a
This Article was originally published on Wine Enthusiast
