**The rise of **beer made by AI Customers love it Aug 27th 2025 WHEN BECK’S, a storied German brewery founded in the city of Bremen in 1873, celebrated its 150th anniversary in 2023 it decided to bring in a new brewmaster to mark the occasion: ChatGPT, an artificial-intelligence (AI) chatbot. The company asked it to whip up a recipe using only hops, yeast, water, and malt. The result was “Beck’s Autonomous”, a lager with a subtle sweetness, a hoppy** texture**, and quite a head. One Daily Mail reporter considered it better than the brewery’s standard lager. Beer and AI may seem an unlikely pairing, but Beck’s is far from the only brand to have asked for input from the technology. Atwater Brewery, an American firm, introduced an AI-designed citrusy India pale ale (IPA) in 2023 and last year St Austell Brewery in Britain used AI to create a tropical IPA dubbed “Hand Brewed by Robots”. In March Coedo Brewery in Japan asked an AI model to analyse the preferences of people in their 20s, 30s, 40s, and 50s, and then developed four craft beers, one for each age range. In general the response from customers, brewers say, has been overwhelmingly positive. “It gives us access to new recipes that we didn’t think about before,” says Prinz Pinakatt, boss of the beer business for Tilray Brands, Atwater’s New York-based parent company. Machine-learning tools can parse the minutiae of complex flavours, analyse the ingredients and equipment that an individual brewery has available, and then concoct new recipes while tweaking sweetness, acidity, hop level and other attributes to ensure the end product appeals to discerning customers. Beau Warren, who opened the Species X Beer Project brewery in America in 2021, knows this firsthand. In 2022 he started training AI models on a number of parameters—his proprietary recipes, different types of yeast, water acidity, various hops, the ingredients in the brewery cellar, the typical makeup of lagers, stouts and other beers—and, by 2024, began using it to guide the brewing process. In one instance, after being asked to make a new lager, the bot suggested mixing Maris Otter malt, usually found in stouts, with Belgian candi syrup. “I would never have thought of doing that in a lager, ever,” he says. “We brewed it anyway, and I thought it was one of the best lagers I’ve ever made.” His customers apparently thought the same: Mr Warren says patrons usually rated the AI-crafted beers better than any of the beers thought up by he and his fellow brewers. (That said, the AI beers at Species X are no more: the brewery closed down last autumn owing to financial difficulties.) Scientists are also intrigued about what bots might tell them about the chemistry of beer. In 2024 researchers from KU Leuven, a university in Belgium, analysed the chemical makeup of 250 Belgian beers, including lagers, blonds and West Flanders ales. They then trained machine-learning algorithms to model the effects of adding or subtracting different aroma compounds, such as glycerol and lactic acid, on the taste. “The models we develop help us to understand the complex relationship between the chemistry of a beer, its taste, and how consumers will like it,” says Kevin Verstrepen, a bioscience engineer who led the research team. Of course, it will take more than a chatbot to replace a human brewer. Ingredients must be poured, brew kettles must be tended and the beers must be tasted—whether they were made totally by human hands, or brewed, at least in part, by robots. “Yes, AI will become more and more part of the brewing process, but the brewing itself, the craft, is still the emphasis,” says Mr Pinakatt. “It will be very difficult to have machines make our beers.” ■ 當 Beck’s——這家創立於 1873 年、位於德國不來梅的老字號啤酒廠——在 2023 年迎來 150 週年慶時，他們決定邀請一位特別的釀酒師：人工智慧聊天機器人 ChatGPT。 公司要求它只使用啤酒的基本原料——啤酒花、酵母、水和麥芽——來設計一份新配方。最後誕生的是一款名為「Beck’s Autonomous」的拉格啤酒。這款酒帶有微微的甜味、濃郁的酒花香氣，還有豐厚的泡沫。根據《每日郵報》的記者試喝後表示，甚至比 Beck’s 的經典款還要好。 乍聽之下，啤酒和 AI 好像是八竿子打不著的組合，但 Beck’s 絕不是唯一嘗試的品牌。美國的 Atwater Brewery 在 2023 年推出了一款由 AI 設計的柑橘風味 IPA。隔年，英國的 St Austell Brewery 也用 AI 釀造出一款熱帶風味的 IPA，名字叫「Hand Brewed by Robots」。2024 年 3 月，日本的 Coedo Brewery 更是請 AI 分析不同年齡層——20、30、40、50 歲——的口味喜好，再釀出四款各自對應的精釀啤酒。總體來說，消費者的反應非常正面。 「這讓我們得到以前從沒想過的配方。」Atwater 母公司 Tilray Brands 啤酒業務負責人 Prinz Pinakatt 說。機器學習工具能精細分析複雜的風味、配料和釀酒設備，接著設計出新配方，並調整甜度、酸度、啤酒花的濃度等元素，確保最終的產品能打動挑剔的消費者。 美國的 Species X Beer Project 釀酒廠創辦人 Beau Warren 就深有體會。他在 2022 年開始訓練 AI 模型，輸入了自己的獨家配方、各種酵母、水的酸鹼度、不同的啤酒花、以及啤酒的種類數據。到 2024 年，他正式讓 AI 參與釀造過程。有一次，他要求 AI 設計一款新的拉格，結果 AI 建議把通常用在世濤啤酒的「Maris Otter 麥芽」，與比利時糖漿混合在一起。Warren 說：「我從來沒想過拉格會這樣做，但我們還是照著釀了，結果發現這是我喝過最好喝的拉格之一。」顧客的評價也很高，普遍認為 AI 釀造的酒比他和同事設計的更好。不過，儘管酒很受歡迎，Species X 最後還是在去年秋天因財務問題關閉了。 不只是釀酒師，科學家也對 AI 在啤酒化學上的應用充滿興趣。2024 年，比利時魯汶大學的研究團隊分析了 250 款當地啤酒，從拉格、金色啤酒到西佛蘭德艾爾。他們用機器學習模型來模擬不同化學物質——像是甘油和乳酸——對風味的影響。研究負責人、生命科學工程師 Kevin Verstrepen 說：「我們開發的模型能幫助理解啤酒化學成分、口感，以及消費者喜好的複雜關係。」 當然，AI 不會完全取代釀酒師。畢竟，原料要有人投入，釀酒槽要有人照看，啤酒最後也必須有人品嚐。Pinakatt 說：「是的，AI 會越來越多地參與釀酒過程，但釀酒的核心仍然是工藝與手藝。要完全靠機器來釀酒，是非常困難的。」 -- Hosting provided by SoundOn