Brain Trust

03:24AM Sep 04, 2014
TP X3 X4 T14129

Artificial intelligence will make your farm smarter

When Khaled Rasheed wears a T-shirt promoting the University of Georgia’s Institute for Artificial Intelligence, people immediately start asking about robots akin to those in Steven Spielberg movies.

Humanoid robots are "intriguing," Rasheed admits. Yet the artificial intelligence (AI) he studies focuses on problem-solving and has greater potential for agriculture.

"Farmers need to keep up with the trends in smart technology because it’s going to be everywhere," he says.

In the next 10 years, computers with AI will let producers drive equipment more efficiently and market crops with greater precision. Machines have been engineered to model human biology, enabling them to solve problems in days.

"Evolutionary computing uses information from past trials to come up with new ideas like it to work well," says Risto Miikkulainen, a professor of computer science at the University of Texas. "You’re modeling biology, recombination and crossover. You mutate the idea."

In this way, computers will increasingly be able to parse massive Big Data sets. They will use simulators to develop efficient solutions such as minimizing a tractor’s fuel use in hilly field conditions without the aid of humans inputting specific desired outcomes. 

It’s already happening. In 2004, Miikkulainen and his colleagues used a simulator to develop a rocket that could fly 20 miles higher into the upper atmosphere than existing machines. They generated and tested solutions for a month. Accounting for fuel levels, attack angle and more, computers evolved thousands of scenarios until landing on the proper throttle controls.

"Today, it would train on your iPad or laptop in a few hours," Miikkulainen explains.

Such exercises will abound in the future. AI will guide product development and also be integrated into devices. The challenge will be how to process the resulting data.

"You may have to deal with it and learn what you can and then throw it away because you cannot keep storing it," Rasheed notes. 

Neural Trading. Farm offices can benefit from AI. Marketing precision might improve with computers that make sense of unknowns, from weather to political climate. 

Already, companies such as ExitPoints use AI to trade futures. Company founder David Register got his start working with NASA to measure the health of corn and other global crops using satellites, notes his colleague, Pat Aucoin. With the technology, researchers could spot drooping leaves, an indicator of crop disease, before farmers even noticed it. Register went on to program ExitPoints, founded in 2003.

"His system has the freedom to search for trends in historical data over multiple time scales involving many features, volatility and multi-market correlations among others, and make discoveries," Aucoin says. "His system is unemotional."

That lack of emotion combined with massive processing power and learning ability will drive innovation in the next decade. Take comfort, though: These machines won’t try to shake your hand. They’re just here to help.

A Singular Dilemma

Futurist Ray Kurzweil has predicted computers will exceed humans in intelligence by about 2045. Intelligent machines could theoretically craft advanced equipment and eventually decide to get rid of humans in a scenario known as the singularity problem.

"A lot can happen between now and then," notes Pat Aucoin, who works with the ExitPoints futures trading system. "It’s certain that robots will become more capable and more ubiquitous."

Don’t get too worried just yet, though, argues Khaled Rasheed, professor at the University of Georgia. "This is not going to happen soon, and I don’t think we’re going to be here if it ever happens," he says. "We’ll leave the next generation to worry about it."