My young son is so freaked out from all the stories about AI that he overhears on NPR while I’m chauffeuring him around, he’s having nightmares about the robots obliterating the human race. While self-driving Tesla crashes (273 last year!) and chatbots that use machine learning to learn to act like Nazis and deepfake videos that will make us distrust our own eyes are all part of our impending Huxleyan technocratic state, it seems there’s a reason that — for now — no episode of Black Mirror has focused on the exciting, potentially frightening, new world of beer.
That’s not to say no one has explored AI’s potential applications to the realm of brewing. Brian Faivre joined Deschutes Brewery‘s brewing team in 2004 and worked his way up to being the brewmaster in 2011 before departing in 2021, in part to become the founder and lead programmer of The Brewery Pi Project. Faivre, who earned his B.S. in computer science, spent part of his time at Deschutes focused on machine learning applications in the brewing space. (One claim listed on his LinkedIn profile says he “utilized machine learning to reduce fermentation cycle times by 17%.”)
The fact is, we like calling beer “hand crafted” as if brewers actually handle mash paddles, but that primitive tool is hardly reached for these days. GoodLife Brewing uses Precision Fermentation’s BrewIQ, a combo of hardware and software that integrate fermentation tanks to streamline the process. And even fairly small Bridge 99 Brewing‘s brewhouse uses a Brewmation (Brew + Automation, get it?). Many brewers are able to manage their beers remotely from their smartphones. Given that brewing is highly mechanized, is it ready for machine learning?
In a “Craft Beer & Brewing” article in 2020, Faivre said, “A lot of these machine-learning concepts were now accessible to us, and were more mainstream. Open-source software was growing more and more accessible…In a handful of the transitions — fermentation, free-rise, diacetyl rest, and cooling — a brewer or a lab tech has to go out, get a sample, prep the sample, and do that analysis. If the fermentation hasn’t quite reached our expected parameters, then we wait (but that) extra time reduces our potential capacity.”
His research became the open-source Brewery Pi Project, which has yet to permeate the industry. Today, Tim Gossack, Deschutes’ brand new director of brewery operations, explains that there’s no AI at Deschutes at the moment.
“They tried to launch some AI work with Microsoft around 2019, ” said Gossack. “We do have a lot of data sources that provide us with the information that would’ve gone into AI to improve efficiencies in the brewhouse. We use (Brewery Pi) a bit but it’s basically data logging and using it to compare things across time to see how they’ve changed.”
While Hal 9000 isn’t running the brewhouse, Gossack does believe there are “absolutely” potential applications for AI. “When you talk about competitive systems that are used to do the same thing over and over again, AI could optimize them. There are systems being built that could support that kind of an operation.”
When Gossack worked at the then MillerCoors-owned (now Molson Coors-owned) brewery in Virginia’s Shenandoah Valley, he explains there were, “integrated systems talking to each other—in tanks, lines and motor control panels. They actually talk to each other and spit out work orders to humans. As this stuff advances, especially if the machines can integrate and find the data, I absolutely think that AI could run a brewing department.”
Gossack was quick to add, “No brewer’s gonna want to hear me say that.”
Faivre, who’s now the director of brewing and distilling operations at Atlanta-based New Realm Brewing, said he was recently invited to present to the annual American Society of Brewing Chemists conference about his research in beer’s AI applications. But it was virtually the same presentation he’d given six years ago.
“Look at ChatGPT. It seems like there’s more of an appetite now that (AI has) become more approachable for the general public.”
That said, Gossack points out, “Over time, with capital expenditure, you can have a system with (machine learning) built-in, but right now it’s cost prohibitive.” Plus, he adds of such machines, “At least they’re not drinking the beer.”
This article appears in Source Weekly July 27, 2023.
MAKE LOCAL JOURNALISM HAPPEN
Republish our articles for free, online or in print, under a Creative Commons license.
