“I have been intrigued by machine learning and data analysis for a long time. As brewing beer is also a long-time hobby of mine, I started to wonder whether artificial intelligence could offer a helping hand in perfecting a beer recipe. This led to the idea to launch an AI-based beer concept together with a group of friends,” Lari Melander says.
But what is an AI-based beer concept?
Melander explains that the concept is based on collecting feedback from customers at scale. An AI algorithm, also developed by Melander, adjusts the beer recipe based on this feedback.
The original recipe was also created by AI, albeit not based on customer feedback but 150 or so existing beer recipes.
The first batch became available to consumers in the spring of 2021. The beer is brewed by United Gypsies Brewery in Lohja, Finland.
“Then the recipe begins to be tweaked based on customer feedback. Once there is enough feedback available, the AI algorithm processes the data and modifies the recipe for the next iteration. With the recipe being continuously improved, no two batches will be completely alike.”
An interesting part of the concept launch is seeing how eager customers are to become involved in the product development process, Melander finds.
“We are working to make it easy and simple for customers to give feedback to ensure access to as much data as possible. The online feedback form allows customers to decide the level of detail in their responses.”
As similar concept has not been tested in Finland before, what led Melander to the idea to launch an AI-based beer?
“I believe brewers and beer lovers around the world are looking for brews that constantly evolve in response to customer feedback.”
As Melander says, machine leaning has already been used, for example, to develop brewing processes, but mainly to make the production processes of breweries more efficient and not so much to fine-tune the actual product to suit consumer tastes.
“This type of approach with end users being fully involved could be more widely integrated into the development of a range of products. The AI-supported collection of customer feedback to improve product development may well become more widespread in the future.”
The key to designing the algorithm that refines the beer recipe was to allow AI to adjust the recipe but stay within the boundaries of the selected beer style. The beer currently developed by Melander and his team is an American IPA.
“We may test the concept with different types and styles of beer at some point in the future. As it is unlikely that there could ever be a single universally liked beer, our goal is to develop multiple versions of a single style of beer to appeal to beer drinkers’ individual tastes.”
It would be great to see the technology to become more widely used to improve all sorts of products that evolve over time.
Consumers will also be able to keep up to date with the evolution of the beer as it goes through multiple iterations. According to Melander, similar AI-based beer concepts have emerged around the world, but they usually keep their changing recipes secret. This time the process will be transparent, allowing consumers to see what they are consuming.
“I think people might be interested in seeing the direction in which the beer is evolving based on their feedback and finding out how others have rated the product.”
What the concept is all about is letting customers decide what kind of beer they want to drink and providing them with a personalised experience.
“We are working to create beers that are tailored to the personal tastes of the people consuming them. It would be great to see the technology to become more widely used to improve all sorts of products that evolve over time.”