Scottish or American? AI robot can distinguish between different WHISKEYS – and could soon replace trained sommeliers, studies show
They may have one of the best professions in the world.
But whiskey sommeliers could soon face competition for their work – from AI.
Scientists have devised machine learning algorithms that can determine whether a whiskey is of American or Scottish origin and identify its strongest aromas.
And they even perform better than human experts, the results show.
The aroma of a whiskey is determined by a complex mixture of odorous substances, which makes it very difficult to assess.
Panels of human experts are often used to identify a whisky’s strongest notes, but these require a significant investment of time, money and training – and agreement between experts is often rare.
A team from the Fraunhofer Institute for Process Engineering and Packaging in Germany assessed the molecular composition of seven American and nine Scotch whiskies.
These include the 12 year old Auchentoshan from the Scottish Lowlands, the 10 year old Talisker Isle of Skye Malt, Jack Daniels Tennessee Whiskey and Woodford Reserve Bourbon.
Scientists have devised machine learning algorithms that can determine whether a whiskey is of American or Scottish origin and identify its strongest aromas
They may have one of the best professions in the world. But whiskey sommeliers could soon face competition for their work – from AI (stock image)
To do this, they used algorithms, including a molecular odor prediction AI they developed themselves called OWSum.
The algorithms analyzed the whiskeys based on their detected molecules and identified each dram’s country of origin and the five strongest notes.
The authors then compared the results of the algorithms with those of a panel of eleven experts.
The results showed that OWSum could determine whether a whiskey was American or Scottish with more than 90 percent accuracy.
The AI identified caramelly as the most characteristic note of American whiskies, and applesy, solventy and phenolic – often described as a smoky or medicinal scent – as the most characteristic notes of Scotch whiskies.
Both algorithms were able to identify the five strongest notes of a specific whiskey on average more accurately and consistently than any individual human expert.
The authors believe that their approach could lead to rapid classification of whiskeys and identification of the most important notes in their aromas.
Their findings were published in the journal Communications Chemistry.