The 1440 Daily Report covered an interesting study regarding Artificial Intelligence’s ability emulate sensory perception and perform pretty darn well!.
“A machine learning program was able to meet or outperform professional tasters in identifying whiskey notes, according to a new study. The demonstration marks a milestone in sensory perception using artificial intelligence. Data analysts tested their program on 16 whiskey samples (seven American, nine Scotch) previously categorized by 11 human experts. The program—using statistical models and neural networks—identified molecules in each whiskey and used them to predict the sample’s top five flavor notes. Researchers compared these findings against the top five flavor notes aggregated from human experts’ descriptions. This type of analysis has historically been challenging; while some chemical mixtures—like vanillin—are directly associated with a specific aroma, others—like whiskey—can rely on over 40 compounds. Notably, the program found molecules like menthol and citronellol unique to the American whiskeys and associated with “caramel-like” notes. Methyl decanoate and heptanoic acid, unique to Scotch whiskeys, were more likely to be associated with apple-like or solvent-like notes.” So is even better whiskey in our future? Will computers become alcoholics? – Jet Cannon |
Sent from my iPhone