Belgian Researchers Use AI to Improve Beer's Taste, But How?

Belgian beers, popular for their diversity and rich heritage, have always delighted enthusiasts with their complex flavors.

Now, researchers claim to have unlocked the potential of artificial intelligence (AI) to elevate beer quality to unprecedented levels.

Can AI really change the game for the brewery industry this time?

AI's Impact on Beer Brewing

The meticulous process of beer brewing requires a series of steps that dictate its taste. Some beers literally age and when this happens, their taste improves the longer they are stored.

At present, a group of researchers from Belgium stumbled on a discovery that AI could further enhance its flavor and aroma.

Leading the charge is Prof Kevin Verstrepen from KU Leuven University, who believes that AI can revolutionize the complex world of aroma perception in beer.

Verstrepen explains that beer, like other food products, comprises numerous aroma molecules that interact with each other, influencing how our senses perceive them. It's important to further dig in on how AI can optimize beer flavors and it would be easier this time with machine learning.

What Verstrepen and His Team Found Out

According to The Guardian, Verstrepen and his team analyzed the chemical composition of 250 Belgian beers across 22 styles. They scrutinized various factors, including alcohol content, pH levels, sugar concentration, and over 200 flavor-related compounds like esters and terpenoids responsible for fruity notes.

A tasting panel of 16 individuals meticulously evaluated each beer based on 50 attributes, such as hop flavors, sweetness, and acidity, over three years. Additionally, the researchers examined 180,000 consumer reviews from RateBeer, aligning them with the panel's ratings to validate their findings.

How Machine Learning Contributes to Beer Enhancement

Utilizing machine learning, the team developed models to predict a beer's taste and appreciation based on its composition. They leveraged these models to enhance a commercial beer by incorporating substances identified as crucial predictors of overall appreciation, such as lactic acid and glycerol.

How Did the Results Fare

The tasting panel's feedback revealed notable improvements in both alcoholic and non-alcoholic beers, with enhancements in sweetness, body, and overall appreciation. While the models are primarily based on high-quality commercial beers, Verstrepen believes their significant application lies in refining non-alcoholic variants to meet consumer preferences.

"Tiny changes in the concentrations of chemicals can have a big impact, especially when multiple components start changing," Verstrepen said.

He added that some substances that are not usually added to beers could yield positive results if they are poured in lower concentrations. They could also complement other aroma compounds based on the findings.

Preserving Brewing Tradition

Despite the AI-driven advancements, Verstrepen highlights that the brewer's expertise remains indispensable. AI cannot easily replace the traditional brewing method, but it can be a great tool to explore other untapped areas of the process.

In short, while AI models provide valuable insights into optimizing beer, it is ultimately the brewers who bring these enhancements to life through recipe formulation and brewing techniques. This ensures that the tradition and craftsmanship remain integral to this unique operation.

The study titled Predicting and improving complex beer flavor through machine learning can be accessed on Nature.com.

Speaking of brewing, Japanese brewer Sapporo is also leaning on AI to develop new products.