E. & J. Gallo Winery in California produces more than 7 million hectoliters of wine annually, almost as much as all of Germany. Gallo knows a lot about wine. It also knows it could do some things better. Gallo manager Nick Dokoozlian, for example, wondered about vine growth. Why are some vines ready to harvest yet others are weeks behind? Maybe a smart irrigation system could change that. Dokoozlian studied the issues and then collaborated with engineers, physicists, and data specialists. In 2011 Gallo equipped a 10-hectare test site with sensors that send data on soil moisture, temperature, wind speed, and groundwater levels to a computer. The computer was also fed satellite, weather, and climate data. The system used artificial intelligence (AI) to learn to recognize correlations and to provide suggestions for watering patterns. The result? The test site used 25% less water and yielded 30% more grapes.
AI refers to computer systems that collect and process information and automatically learn how to process it better. AI learns to recognize patterns in data more efficiently and, over time, can draw more meaningful insights from them. AI’s biggest challenge? To save the planet.
AI to the rescue
“We won’t win the fight against climate change without AI,” says Oliver Zielinski of the Research Center for Artificial Intelligence. That’s one of the reasons why policymakers are betting on AI. Germany, for example, has had a national AI strategy since 2018: “We want to leverage all of AI’s advantages and possibilities for environmental and climate protection," Germany’s Environment Minister Svenja Schulze says.
AI can indicate the greenest route through a city, propose a climate-friendly forest conversion, make industrial processes less carbon-intensive, even manage weed control. Topography, vegetation, movement, and weather data can be combined in order to detect forest fires and illegal fishing early. “Harnessing the Fourth Industrial Revolution for the Earth,” a study published by the World Economic Forum, highlights many such approaches. But there are two challenges.
First challenge 1: knowledge silos
Like Gallo’s Nick Dokoozlian, many experts know a lot about their own field but not much about AI. "Each discipline still works in isolation. People research, experiment, and develop in different areas," says Simone Kaiser, deputy director pf the Center for Responsible Research and Innovation (CeRRI) in an interview with enorm, an online magazine. Data scientists are experts at collecting, combining, and analyzing data; climate researchers, geoscientists, forest ecologists, and biotechnologists much less so. They don’t understand each other. What’s worse, they often don’t even communicate. “There’s a lack of suitable platforms and forums where different disciplines can meet and share ideas,” Kaiser says. But this is gradually changing. Experts from all disciplines understand reliable data are indispensable. And companies like Amazon, Apple, Google, and Microsoft are helping fund interdisciplinary collaboration. IBM’s Green Horizons initiative, for example, is behind Gallo’s project.
Second challenge: solution or part of the problem?
Saving the planet will require more conservation, including of energy. This also applies to computing power, which is already responsible for 4% of global carbon emissions. The U.S. Department of Energy estimates that data centers worldwide use around 200 terawatt-hours of electricity per year—more than Poland, Sweden, or Argentina. Computer and communications technology are expected to consume 8% to 20% of the world’s electricity in 2030. Data centers will account for one third of this amount. In other words, people who use AI to conserve resources are themselves consuming resources. Fortunately, AI can help here too. The Machine Learning Carbon Impact Calculator, for example, can roughly calculate algorithms’ carbon footprint.