Brewers have long supported the craft and passion behind brewing, but scientists believe that artificial intelligence (AI) and machine learning have the potential to revolutionize how the industry develops new beers. How will beer development look in the future? Here are some answers provided by BeverageDaily.

Brewed by Robots: An American AI-PA

Can AI help create better beers; will it save time and money for brewers; and where does that leave the traditional human brewer? Let's take a look at how AI is being used in the lab, in small breweries, and at a commercial mass scale.

Introducing AI into the brewery is St Austell Brewery in the UK. Founded in Cornwall in 1851, the brewery is looking to the future and debuted last month with a beer developed with AI.

"Hand Brewed by Robots" is a 4.2% "American AI-PA" created by an online generator, with assistant beer manager Barnaby Skerrett providing instructions based on the ingredients he wanted to use and the flavors needed.

While the recipe was computerized, the beer is still handmade, using St Austell Brewery's small batch kit.

For the brewery, the purpose of using AI was purely experimental rather than to save time or money.

"The idea for the beer came from working with AI tools in my general job with brewery automation," said Skerrett. "While talking to AI one day, I thought I'd ask it a brewing question to catch it off guard because I was interested to see the results.

"I told it to write me a recipe on some broad parameters of colors and flavors, and it came back with some ideas, so I decided to turn them into a recipe. There were a few tweaks to make it suitable for cask, as it didn't seem to know what a cask was, which was quite interesting."

The beer uses Willamette, Cascade, and Sultana hops to give an "AI-PA tropical juicy and resinous flavor." Skerrett also doesn't see any threat to beer production from AI in the future.

"In terms of beer, too much requires human intervention, whether it's checking ingredients and other quality checks that AI can't do," he said.

"It's the same when the beer comes out the other side and we do taste checks. An artificial intelligence can never tell you if that beer tastes fantastic or if it's not up to par."

What Does Your Beer Taste Like?

This is a question that Belgian researchers are exploring. In beer making, one of the biggest challenges for brewers is to describe—and then reproduce—the flavors and aromas in a beer.

As with any food product, comparing and ranking beer flavor profiles is highly subjective: guidelines often resort to generic terms like "fruity."

Scientists at the VIB-KU Leuven Center for Microbiology and the Leuven Beer Research Institute in Belgium have used AI to create a clearer and more precise way to describe beers. This information can then be used to analyze the flavor profile of beers, predict how the beer will be rated by consumers, and what flavor compounds can be added to improve it.

They started by chemically analyzing beers, carefully measuring the concentrations of hundreds of aromatic compounds. This was accompanied by an evaluation by a group of 15 trained people, who were asked to comment on a set of 50 criteria.

The project was a "truly Herculean effort," according to researchers, who quickly realized that the first 100 beers were not enough to capture the diversity of beer just in Belgium (they eventually analyzed 250 beers over the five-year project).

Once they had the chemical concentrations and detailed tasting reports for hundreds of different beers, scientists then turned to AI to connect the two. A model could predict key flavors and the final appreciation score of a beer without needing human tasting.

These results, in turn, were used to further improve the taste of an existing commercial Belgian ale by adding certain flavors predicted by the model to increase beer quality. The modified beer scored better in blind tastings than the original version.

"The flavor of beer is a complex mix of aromatic compounds. It's impossible to predict how good a beer is by measuring just one or a few compounds. We truly need the power of computers," explained Michiel Schreurs, one of the project's researchers.

As consumers seek to moderate their alcohol intake, non-alcoholic beers are booming. While brewers claim the success they have achieved in creating much tastier non-alcoholic beverages than their predecessors a decade ago, they still need to keep up with the flavor, mouthfeel, and quality of their alcoholic counterparts.

This is where AI could come in, said Kevin Verstrepen from the Belgian VIB-KU Leuven team.

"Our biggest goal now is to make better non-alcoholic beer. Using our model, we've already managed to create a cocktail of natural flavor compounds that mimic the taste and smell of alcohol without the risk of a hangover." Additionally, he sees the potential for the technology to be extended to food products.

An Artificial Beer?

In principle, artificial intelligence could be used to train an "artificial beer product," says Kevin Verstrepen from VIB-KU Leuven in Belgium—but it's not something that will happen soon.

Rather—at least for now—the role of artificial intelligence is as a tool for brewers to help them tweak existing beers.

"So far, the models make a connection between the chemical composition and consumer perception and appreciation. In our study, we tested the models by supplementing existing commercial beers with complex mixtures of flavor compounds predicted by the AI model to make the beer even better. And it works!

"However, since adding beer with additives is not a common practice, it is still up to brewers to achieve the desired chemical changes through the brewing process—such as adjusting ingredients, brewing scheme, or fermentation conditions.

"The good news is that this is possible—because we know so much about the flavor compounds in beer and their origins, knowledgeable brewers know perfectly well how to tweak the concentrations of major volatiles, such as, for example, fusel alcohols, esters, diacetyl, aldehydes, DMS, alcohol, and even very specific hop and malt flavors."

Where AI Goes Wrong

Getting it wrong is part of the learning process for humans, and that's also true for AI.

In 2018, IntelligentX, based in London, brewed an artificial intelligence beer: and noted that human brewers sometimes rejected AI beers—all part of the machine learning process.

The VIB-KU Leuven team didn't encounter this specific problem but found other limitations of AI.

"A specific shortcoming of the models we've created was that they are not good at estimating upper limits for flavor compounds, they will often predict that increasing a positive compound always leads to a positive change," said Verstrepen.

"This is, of course, false; there can definitely be too much of a good thing!

"The reason this happens is that we've trained our models on existing commercial beers and they don't have real flavors, where there really is too much compound. Therefore, the models can't learn very well what it means to have too much."

Here, again, human brewers can guide the technology. "For us, the problem is easily circumvented—we know what we're doing and what the normal limits are for each flavor compound and we stay within that natural range."

In the future, providing the models with a wider variety of beers—including lower-quality ones—could help them learn.

Reducing NPD Time

Sapporo Breweries—one of the largest beer producers in Japan—has already turned to AI for its beers and sees the potential to expand this in the future.

Unlike St Austell, it focuses on how AI can be used to reduce NPD time and costs on a mass scale.

In partnership with IBM, last year it launched the first AI-developed alcoholic drink.

While an RTD alcoholic drink rather than a beer, "Otoko Ume Sour Salty Plum" used AI to create the right flavor profile for the product.

Sapporo found that AI—called N-Wing Star (New Wing Star)—could "significantly" reduce the time needed for NPD.

It is estimated that it can reduce NPD time by about 50%: due to AI's precision in sorting potential prototypes in early stages.

And Sapporo highlights the technology's potential to surpass what brewers do: saying that AI has offered the opportunity to propose recipes and ingredient combinations that humans wouldn't have thought of. They are already looking to expand the technology across their RTD brands. (Photo: Dreamstime)