There are many ways of looking at the challenges being faced by the agricultural sector, and how AI & clean data can make a huge difference. There’s a stated vision of doubling farmers’ income by 2022 and we are less than two years away from the date, or we can spin a global perspective – there will be 10 billion of us by 2050 and food shortage is an impending disaster that we are staring at unless we act real quick. 

In India, approximately 100 million families depend on agriculture for livelihood so we are potentially looking at bettering the lives of 400 – 500 million Indians. The sector generates 49% of employment but contributes only 16% to the GDP. Clearly, there’s a wide gulf to be bridged. Farmers’distress has plagued the nation for 70-odd years and we need to change this narrative to bridge the inequality gap. Approximately 20 – 25% of farmers barely cross over the poverty line and about 40% or more need additional income through non-farm activities. And it doesn’t help that a large number of farmers own less than 2 hectares of land to classify under small-holding. Affordability & Accessibility are the cornerstones that will guide us on what’ll work and what won’t.  

Direct intervention is through precision farming where the farmer can avail advisory services that address a whole lot of issues – soil health, crop selection, how much water to be used, pesticides, etc. That, farm productivity has been very low, is a perennial problem and data-driven advice is required in a whole range of activities. The decisions that farmers make are essentially based on three parameters – production cost, profits and risk of crop failure. But these decisions are largely based on experience. 

Access to loans is yet another area that needs to be addressed. People who aren’t a part of the formal economy face great difficulties in getting loans. ML techniques based on behavioral patterns reveal an individual’s creditworthiness and even the willingness to pay back. These parameters can be used to serve the underprivileged sections by designing appropriate packages. 

Unlike some other industries, the “ecosystem” is very weak and leads to supply chain bottlenecks. Equally stiff, is the challenge of collecting reliable data. These models work well only if the data is clean. Data from one source may not always be reliable and that’s why the same data must ideally be captured from multiple sources to establish authenticity. And there aren’t any standardized formats of data capture. There are loads of successful PoCs but not too many have been taken to scale. Perhaps it’s time to consider a national programme for agriculture where the farmer is at the centre.    

When we build solutions for this sector, we must bear in mind the following: applicability, affordability, accessibility, achievability and sustainability. After 1960, we haven’t really faced a food security problems and yet, farmers don’t get the best price. The supply chain has to be demand-driven to fetch the right prices. Often, farmers don’t have a clear picture of the demand pipeline and end up producing more than required. 

Clearly, there are many areas where AI can play a role to improve the conditions of farmers, particularly those with smallholdings.

Image from pxfuel