Forget The Hops, AI Bots Could Brew Your Next Favorite Beer
Conversing with an AI-powered chatbot or trying to beat it at chess can be fun and all, but researchers, scientists, and engineers have much bigger aspirations for artificial intelligence. You know, like brewing the perfect beer, as it turns out. Before anyone scoffs, brewing beer is serious business, and also a good challenge for AI training.
Humans seemingly have the advantage for these kinds of tasks because of taste buds, and because we've been playing around with alcohol since nearly the beginning of the time. Even so, it's apparently possible to tap AI to become a brew master, as researchers in Belgium outline in a study published to Nature. In it, they discuss the "challenging" science of dealing with flavors.
"For each beer, we measure over 200 chemical properties, perform quantitative descriptive sensory analysis with a trained tasting panel and map data from over 180,000 consumer reviews to train 10 different machine learning models," the researches outline in their study.
"The best-performing algorithm, Gradient Boosting, yields models that significantly outperform predictions based on conventional statistics and accurately predict complex food features and consumer appreciation from chemical profiles," the researchers added.
From the outside enviously looking in, the experiment sounds like an excuse to sample beer after beer, while getting paid for it. Sort of like Norm Peterson when he scored a dream job as a beer tester in the third episode of season 11 on Cheers. However, the researchers are adamant that developing accurate AI models for flavor "would contribute greatly to our scientific understanding of how humans perceive and appreciate flavor."
Source: Nature Communications
For this research, they tapped 250 commercial Belgian beers across 22 different beer styles representing a variety of alcohol levels (side note: if you ever have a chance to sample an Apple Lindemans (formerly Pomme), I wholeheartedly recommend doing so—take it from someone who's on the Wall of Foam at Big T's in Lawton, Michigan). Each one underwent 226 different chemical property measurements.
In the end, the researchers highlighted limitations of different machine learning models and the overall science of having AI analyze beer for flavor, but the results were very promising.
"The predictions of our final models, trained on review data, hold even for blind tastings with small groups of trained tasters, as demonstrated by our ability to validate specific compounds as drivers of beer flavor and appreciation," the researchers noted in the beer study.
That will give us humans something to bond over with our robot overlords one day.
Top image created with Copilot (via DALL-E) and Photoshop
