Scientists in Russia are now teaching artificial intelligence to tell the difference between sparkling wines—by analyzing the bubbles. Yes, the bubbles. Not the taste, not the smell, not the label that says “fancy,” but the fizzy little escape artists doing backflips in your glass.

Turns out, the researchers at the ITMO University (aka the Hogwarts of Russian tech wizardry) decided to borrow a method originally used for analyzing oil products—because apparently, the only difference between crude oil and champagne is whether it’s your boss’s birthday.

The method in question? Ultrasonic cavitation. This is a fancy term for “blast the wine with sound until it makes bubbles,” and then study how those bubbles are born, live a short dramatic life, and die for science. They recorded the bubbling drama on video, analyzed nearly 20,000 images with machine learning, and voilà—sparkling wine types could be distinguished with 84% accuracy, and the type of glass (plastic or glass, classy or sad) with 82%.

Why does this matter? Well, traditional methods for testing wine quality involve expensive chemical equipment or some guy’s opinion. Neither is exactly ideal. This method promises a more objective and scalable way to judge bubbly, and may even be used in pharma later on. What could be more medical than booze science?

As for the big picture:
If you’re just drinking wine to get drunk, the best wine is pure ethanol.
Next breakthrough? Proving alcohol is bad for you. But like… meh.