OSAKA, Mar 30 (News On Japan) - A vibrant new line of craft beer has been created through an unusual collaboration: generative AI working hand-in-hand with seasoned brewmasters.
Revealed on March 28th, the project is a joint venture between NEC and craft beer maker Coedo Brewery. The result is a colorful quartet of beers—pink, blue, and other vivid hues—under the name "Jinsei Jozo Craft" (Life Brewing Craft), each brew representing the characteristics and values of a different generation, from people in their 20s to those in their 50s.
"This project is unique in that AI and craftspeople were able to collaborate and co-create through dialogue," said Ryuta Asami, a data scientist at NEC.
The recipes were developed using NEC’s generative AI technology. An AI agent was instructed to create beers inspired by people in their 20s through 50s. It analyzed the traits of each age group and drew on Coedo's internal brewing knowledge and global data sources to generate initial recipe ideas.
Craftsmen then refined the AI’s proposals through chat-based collaboration, resulting in four distinct beers tailored to each decade.
The beer for people in their 20s features a fruity, fresh peach flavor. The version for those in their 30s is strikingly blue and designed to soothe tension while offering a crisp finish from rice, allowing it to pair well with a variety of meals. For those in their 40s, the beer has a slightly spicy taste, capturing the depth and maturity of the decade. Finally, the 50s version is a deep red brew with rich aromas of malt and caramel, reflecting a life that has weathered many waves.
The use of the AI agent also led to a 40% reduction in labor compared to traditional beer-making methods.
“It really felt like a new team member had joined our brewing crew,” said Coedo Brewery President Shigeharu Asagiri. “The AI made proposals, and we discussed them together as a team, just like we would with a new colleague.”
The beer is set to ship starting in June and will also be showcased at the Osaka-Kansai Expo.
Source: FNN
