Jul 18, 2018 - Technology

Carlsberg turns to AI for brewing help

Renegade Brewing, Denver. Photo: Image: Andy Cross/The Denver Post/Getty Images

Since we’ve sequenced the human genome and created a tool that can edit it, the logical next step is Carlsberg’s Beer Fingerprinting Project, which uses machine learning to predict a beer’s taste profile — bitter, sweet, toasty, and so on.

The big picture: Brewing is chemistry, and computer-assisted analysis can help fine-tune recipes or processes that can make a lot of money for beer companies. It was only a matter of time before AI snuck into the brewery.

The details: This project from the centuries-old Danish brewer will allow it to iterate more quickly to come up with new types of beer to sell.

  • It's gimmicky, sure, but cutting down on costly, time-consuming brewing rounds can save a big brewery money.
  • The big question: Even if the system can eventually predict a beer's flavor profile — for now, it can only differentiate between types of beer — can it guess what tastes good to actual people?
"It may sound nice to have to taste a lot of beers every day, but we create hundreds of small microliter brews and beers, in such small volume that they’re not really testable."
— Jochen Förster, director of Carlsberg Research Laboratory, in a Microsoft blog post

Be smart: There is no industry that won’t try shoehorning AI into its work — whether because it makes sense or because it’s the buzzword du jour.

  • Förster told CNET: "We have had an unexpectedly positive marketing benefit from working with artificial intelligence."

Steve LeVine’s thought bubble: Do we really need more types of beer?

  • My thought bubble: Yes.

Drink up: Microsoft’s blog post describes how it helps analyze the data from Carlsberg’s sensors.

Go deeper