Bringing Artificial Intelligence to Agriculture

How Evolutionary AI is helping optimize crop yield and transform the future of food production

Open Agriculture Case Study

Source: Open Agriculture Initiative

The Challenge

Many problems still disrupt the agriculture world including extreme wastefulness of water, non-arable land, and distribution challenges.

  • About 69 percent of freshwater is devoted to agriculture worldwide, with 60 percent of it being wasted via runoff into waterways and evaporation.
  • Only 10.9 percent of the land on earth is considered arable land, or land that can be farmed.
  • Food is usually not grown where it is consumed. An apple that you buy from a grocery store is often picked 11 months ago—at that point it is not much more than a ball of sugar.

What if you could create the optimal growing conditions for crops? Not only that, but what if you could contain those crops in controlled environments so you could not only measure every relevant nutrient and variable but also, eventually, grow optimized crops anywhere?

The Big Question

Can distributed artificial intelligence help reform agriculture to benefit the masses?

The Answer

Yes; here’s how.

In 2015, we realized that Evolutionary AI could be applied to model how plants grow under different conditions, and to optimize the growth recipes themselves. In collaboration with the Open Agriculture Initiative, these recipes were tested on “Food Computers”, i.e. computer-controlled contained growing environments.

The first test of this approach was to find recipes for growing most flavorful basil. The optimization algorithms tested elements like period of light, type of light, and amount of UV light. It found many significant, nonlinear interactions between recipe variables, including a negative correlation between weight and flavor, as expected, but also flavor improvements with 24-hour light periods, which was a surprise. Thus, the AI approach discovered insights that would have been difficult for Open Agriculture to find on their own.

For more information, read our research paper.