Artificial intelligence has made a significant breakthrough in the world of whisky production, with researchers using the technology to predict the distinct notes of whisky and determine its country of origin—whether it's from the United States or Scotland.
Led by Dr. Andreas Grasskamp at the Fraunhofer Institute for Process Engineering and Packaging in Freising, Germany, the study marks a step toward creating automated systems that can predict the complex aromas of whisky based on its molecular composition. Traditionally, expert panels assess the woody, smoky, buttery, or caramel aromas of whisky to ensure consistency across batches. However, the AI developed in this study aims to enhance this process by adding efficiency and consistency to the evaluation.
"The beauty of AI is its consistency," said Dr. Grasskamp. "While expert noses are still required, this technology supports the process by bringing more efficiency and uniformity to the task."
Whisky’s aroma is a complex blend of chemicals, making it challenging to predict how it will smell from its chemical makeup alone. The interactions between these compounds can mask or alter particular scents, creating a distinctive aromatic profile that is difficult to replicate consistently.
For this research, the team analyzed the chemical compositions of 16 whiskies—eight from the US, including popular brands like Jack Daniel's and Maker's Mark, and eight from Scotland, such as Laphroaig and Talisker. The chemical profiles were then cross-referenced with the aroma descriptions provided by an expert panel of 11 tasters. The AI algorithms were trained to identify the five major aromas and distinguish between US and Scottish whiskies based on their molecular signatures.
One of the algorithms proved to be more than 90% accurate at identifying whether a whisky was American or Scottish. On average, the AI system identified the top five aromas of each whisky more accurately and consistently than any individual expert. The details of the research were published in Communications Chemistry.
Key chemical compounds helped the AI distinguish between the two types of whisky. For instance, menthol and citronellol were associated with US whiskeys, which tend to have a caramel-like flavor. On the other hand, methyl decanoate and heptanoic acid were linked to Scottish whiskies, which often have smoky or medicinal notes.
Beyond whisky, the researchers see potential applications for this AI technology in various industries, including the detection of counterfeit products by analyzing discrepancies in their smell and improving the blending process of recycled plastics, which can develop unpleasant odors.
Dr. William Peveler, a senior lecturer in chemistry at the University of Glasgow, noted that AI could provide greater stability and consistency in whisky production. "The chemical signature of a whisky brand could be checked from batch to batch to ensure a consistent flavor profile," he explained. However, he also cautioned that the study only involved a small sample of whiskies and that more work would be needed to see how the AI would perform with a broader range of products, particularly as flavors develop over time in casks.
The study highlights the potential for AI to revolutionize the whisky industry by offering a new level of precision and consistency, while also suggesting broader applications for this technology in other sectors.
No comments:
Post a Comment