A lab at the University of Toronto cooks with artificial intelligence. And by that, I mean they recycle carbon dioxide into more useful chemicals, these materials are discovered with artificial intelligence and supercomputers.
Ted Sargent and his team run a test kitchen that develops recipes, measures and mixes ingredients, and then sees what happens. As predicted, what is created turns out to be inedible but the flavor is not what they’re going for.
Sargent and his team have a goal to invent recipes by cooking with carbon dioxide to “upgrade” the greenhouse gas into useful materials. The pollutant that carbon dioxide releases could use renewable energy to convert it into raw materials they could sell instead of releasing it into the air or putting it underground.
An article from WIRED says that “one promising class of recipes involves electrically zapping carbon dioxide with other reactants to transform it into the six-atom molecule ethylene, composed of two carbon atoms and four hydrogens.”
Where the profit can come in is ethylene is a raw material used to make everyday plastics such as supermarket and Ziploc bags. Sargent says, “it’s about a $60 billion market.”
The true significance is that Sargent is cooking with artificial intelligence. They discovered new ingredients by using new AI and supercomputers that have become popular for scientists.
Sargent also teamed up with Zachary Ulissi of Carnegie Mellon University. Ulissi specializes in using algorithms to invent new materials, right up the alley of what Sargent wants to accomplish. Using a supercomputer, Ulissi simulated 12,229 microscopic close-ups of 244 different crystals to see which one could potentially make ethylene. Using a supercomputer to complete all 12,229 close-ups was not the logical way to go, so once a fraction of them was analyzed Ulissi trained a machine-learning algorithm to complete the rest.
With AI and supercomputers, it took the team about three years to identify catalysts, test them, and then publish results. Kristin Persson, a physicist at Lawrence Berkley National Laboratory, believes that the way to speed up that process is robotics. Using robotics can convert those testing stages from human-driven to robotic driven which can get results faster. Of course there will need to be human oversight with artificial intelligence and algorithms not being super precise yet.
Making these recipes requires a lot of electricity and the team has not found their winning ingredients quite yet. But Sargent, Ulissi, and their team are trying to make their design more economically friendly and decrease their carbon footprint while working towards finding their golden recipe.