Genie in a Bottle
How AI is influencing what we drink. By Fred Siggins.
On the coast of Far North Queensland, with the sapphire waters of the Great Barrier Reef Marine Park sparkling in the distance, a drone hovers overhead, its rotors humming like a giant mosquito. But this machine is not here to take Instagram-worthy videos of tropical vistas. Instead, it’s scanning a field of agave plants for spectrums of light invisible to the human eye. These spiky succulents have been planted by drinks company Top Shelf International (ASX:TSI) to be made into its Australian tequila-like spirit under the brand Act of Treason.
Across Australia, drinks companies are increasingly using technology such as drones to collect data about their crops, and AI-powered software to process those data into a fantastic array of useful information. That information can help with everything from estimating yields to analysing plant health to setting irrigation schedules, all in ways that were previously impossible or hugely labour intensive.
Chris Monsour is an agronomist with Prospect Agriculture who looks after the agave plantation for Top Shelf. “Traditionally, to monitor the plants you had to take physical measurements,” he says. “So you were limited by time and labour. It took a team of three people a full day to measure 200 plants. Now we’re able to fly drones over a block, and in about 20 minutes we’ve got health and size data for eight-and-a-half thousand agave.” The drones can also see things people can’t, meaning problems can be identified much earlier than before. “The plants reflect spectrums of light that we can’t see but the drones can sense,” Monsour explains. “So we can know what’s happening with the plants way before we’d be able to see it by eye.”
Once the data has been collected, the next step is processing it into a useful format, and that’s where AI comes in. “AI machine learning has been taught how to recognise agave leaves,” Monsour says. “So now we can use that software to measure all sorts of characteristics. Each plant is geotagged so we can look up an individual plant and have its entire life story.”
As well as having clear benefits for plant health and irrigation practices, from the point of view of consumers, the information provided by machine learning is set to have a big impact on the final product. “In terms of beverage production, it can really fast-track information about which agronomic practices lead us to the best product,” Monsour says. “We can also feed the info about individual plants into a harvester and set it to only pick the best of the best.” It’s the kind of knowledge that used to take generations of trial and error to attain.
Meanwhile, in the vast vineyards of the Barossa Valley, like a scene from a futuristic sci-fi flick, tractors with no drivers lumber back and forth along the rows of grapevines. Treasury Wine Estates (ASX:TWE), owner of brands such as Penfolds and Squealing Pig, is running trials of autonomous, AI-driven tractors in its wineries here as well as on its properties in the US and New Zealand. The tractors are equipped with cameras that collect real-time data on vine health and potential yield, used for things like the timing of irrigation, spraying for pests and harvesting. In the warehouses, Treasury has built its first fully autonomous barrel hall, where AI-equipped forklifts do the heavy lifting, literally, both moving and keeping track of thousands of barrels every day.
“The Australian wine industry is in trouble right now,” says Professor Rachel Burton from the University of Adelaide’s School of Agriculture, Food & Wine, one of the world’s most respected wine education programs. “The effect from Chinese tariffs [lifted on March 28], water [scarcity] problems and changing tastes means a lot of producers won’t survive.” As such, “There’s a massive amount of research going into water use, carbon sequestration and the microbiome in the soil, which is incredibly complex. We’re using imaging technology to map soil types, using drones to track vine health, using precision agriculture to make decisions around when to apply insecticide or fertiliser or even when to harvest.”
Professor Burton is involved in a joint project between the Australian Research Council, the University of Adelaide and Top Shelf International (the spirits producer) to apply academic research to commercial production of agave, hopefully with benefits for all. She says that these days, AI and machine learning are involved in every single agricultural product you can think of. “The technology is just going to get more and more sophisticated,” she says. “We can now use hyperspectral light to work out what’s happening within a plant without having to destroy it. In the case of industrial hemp, we’re able to tell what sex the plants are when they’re still very young, and there’s been a handheld monitor developed to measure the quality of wool samples. Seed quality and fruit quality are so important for so many crops, so imagine a point-and-shoot handheld scanner that can tell you if something is ready for harvest — it has massive potential.”
In terms of drinks production, agriculture is the primary frontier for AI and machine learning involvement. But expect to see other aspects of the drinks industry, from marketing to blending, influenced by AI in the very near future. As a key example, Swedish whisky distillery Mackmyra released the first AI-blended whisky in 2019. The distillery’s machine learning models, powered by Microsoft’s Azure cognitive services, were fed with existing recipes as well as customer preference and sales data to create the Intelligens series of single malts. As with all drinks, the quality is in the eye of the beholder, but according to whisky review website Master of Malt, the second edition of Mackmyra Intelligens is a “marvellous single malt Swedish whisky, made by an AI. The future is here, folks”.
This is an extract from an article that appears in print in our twelfth edition, Page 60 of Winning Magazine with the headline: “Genie in a Bottle”. Subscribe to Winning Magazine today.
Comments