Steve Cubbage: Is Ag Too Complex For Artificial Intelligence?

October 8, 2018 10:36 AM
Steve Cubbage

Watch out, farmers. Computers are after your jobs, and they’re coming fast.

According to Tesla and SpaceX founder Elon Musk, artificial intelligence (AI) will beat humans at just about everything by 2030. We’re already seeing McDonald’s employees being replaced by an interactive kiosk.

Machines will be superior to us in translating languages by 2024 and writing school essays by 2026, based on University of Oxford research.

Within 10 years, computers will be better at driving a truck, and by 2031, they will be better at selling goods and, thus, might put millions of retail workers out on the streets. Farmers shouldn’t be all that surprised as tractors have been driving straighter rows than they can for more than a decade.

Don’t be too quick to turn over the keys of the farm to a silicon-based life-form just yet. Agriculture at the ground level might well be the final frontier when it comes to industries to be conquered by AI.

In simple terms, agriculture is complex. How complex, you ask? Mother Nature and agriculture have already set one of the planet’s largest computing companies, numerous high-profile agricultural firms and associated industry specialists back on their heels when it comes to AI.

In 2011, IBM through its research and development headquarters in Haifa, Israel, launched what was supposed to be a groundbreaking agricultural cloud-computing project. The project had one goal: to take volumes of academic and physical data sources from an agricultural environment and turn those ones and zeros into easy answers for farmers. In other words, the Big Blue computer would outthink and outflank Joe Farmer in making critical, real-time decisions for a growing crop.

It was the consensus of many IBM project team members that it was entirely possible to “algorithm” agriculture. Take it a step further and algorithms could solve any problem.

Why shouldn’t they think that? IBM’s “learning” supercomputer system named Watson competed in the game show Jeopardy against former winners and organic-based life-forms Brad Rutter and Ken Jennings. During the game, Watson wiped the floored as AI notched its first high-profile victory over human intelligence.

In the years that followed, Watson cracked the code that led to groundbreaking achievements in medicine. Behind the scenes, IBM’s agricultural computing projects were being cut back or shuttered entirely. Conclusion: Mother Nature wiped Watson’s floor and showed even the complex field of medicine is easier to compute than a single field in agriculture.

Musk’s prediction that AI will dominate agriculture by 2030 might be way off for two reasons:

  • Too many variables. Floods, heat and hail along with bugs and diseases, herbicide-resistant weeds and just bad timing, can turn a good crop bad with the drop of a hat.
  • Too little data. For AI to work, it must be fed mountain after mountain of data. Good data. As much as the precision ag industry would like to pat itself on the back for all of its advances, the truth is most of the practical field-level digital data collected to this point has as many holes in it as a slice of Swiss cheese.

Despite all this, the insertion of AI into agriculture is still coming. These are machines. They do not sleep. They do not get tired, and they will keep getting smarter and faster. Those who embrace AI, stay ahead of it and harness its power will likely benefit immensely. Those who don’t will likely be out on the street with the McDonald’s workers.

Back to news


Spell Check

No comments have been posted to this News Article

Corn College TV Education Series


Get nearly 8 hours of educational video with Farm Journal's top agronomists. Produced in the field and neatly organized by topic, from spring prep to post-harvest. Order now!


Market Data provided by
Brought to you by Beyer