Artificial intelligence and agriculture: if data is the new oil, AI can make jet fuel

Is oil actually valuable?

In 2024, crude oil was worth about $80 per barrel. That’s around 39 pence per litre. Compared to other liquids, like insulin and Chanel No 5 or even the relatively mundane extra virgin olive oil, crude oil is near-worthless.

But even the near-worthless can be lucrative when produced in great quantity, and oil and data are produced in mind-boggling quantities. Data: 403 million terabytes per day, enough to fill up the memory of three trillion iPhones. Oil: six trillion litres per year, globally; enough to fill the UK’s largest water body by volume, Loch Ness, every year. 

That’s not the end of the story for oil though. The value of oil is found in its use as bitumen for roads, petrol for cars or kerosene for jets. Refining the oil to these products, and in such great quantities, is where the value is made. Without a refinery, oil would be left in the ground.

The value of data

Back to data. For the most part, data is still ‘in the ground’; it’s unrefined and useless. In the past, we’ve used pen and paper, calculators and spreadsheet formulas to transform that collection of numbers into some kind of output, an output that informs our decisions and allows us to improve and create value and growth. 

Yield mapping is one agricultural example – ascertain which parts of a field yield less, understand why and take action accordingly. In theory, yield data will ultimately be transformed into greater returns or lower costs. Despite ag-tech expediting the process of identifying weak spots, the process is far from as fast as it could be.

Data’s refinery

In many areas, artificial intelligence (AI) has long since passed the point where it merely substitutes a human, performing the same task, to the same level of accuracy, in the same amount of time. We are now firmly in the realms of augmentation, performing the task quicker, cheaper and/or better than a human. Agriculture is no different. Be it interpreting spreadsheets, imagery or instrument readings, AI will be able to augment the process and, in turn, add value.

Consider soft fruit, a highly time-sensitive crop. Making accurate yield forecasts is essential for growers and retailers; forecast too much fruit and there is waste and financial loss. Forecast too little and growers lose business. FruitCast uses a camera to take thousands of pictures of soft fruit as it grows. From this, AI can determine the size, weight and number of fruit and deliver a more accurate forecast ahead of the competition.

Explore and experiment

The stumbling block is with the sector’s ability to feed AI sufficient data. Just as an oil refinery is useless without oil, AI is useless without data. Those who have kept records over time will be one step ahead. Those adopting ag-tech equipment, such as sensors and cameras, will also be in a better position. Deploying AI to process the data from these actions will yield value. 

It could be as simple as having AI analyse the farm’s financial records to pick up on trends and anomalies. It could be as complicated as having it work with cameras on a boom sprayer to target pesticide applications. The essential requirement is to start experimenting with the possibilities and thus add value to data that is already available. The first try might not turn your data into jet fuel, but it could create diesel and we know that costs more than 39 pence per litre.

Further information

Contact Joe Lloyd