The Govt. is looking at algorithms to fix delays in claim payouts under the Pradhan Mantri Fasal Bima Yojana (PMFBY), a major issue concerning the flagship crop insurance scheme.
The government is experimenting with big data analytics, artificial intelligence (AI) and machine learning to quicken assessment of crop damage, a lengthy and often disputed mechanism. If the PMFBY is to achieve its most critical goal — timely payouts to farmers — it can’t achieve it without a high-end technological fixes, experts say.
While analysing the scheme’s design in his recent work ‘Supporting Indian Farms the Smart Way’, economist Ashok Gulati predicted that, for prompt crop insurance settlement, India could even need a fleet of cloud-penetrating satellites for faster crop-loss estimates. Setting up such a constellation would likely cost Rs 2,000 crore, according to Gulati’s calculations.
The government has taken the first few steps in harnessing high-tech and the results are “optimistic”, an official said.This rabi season, the government partnered with CropIn, a Bangalore-based AI firm, to infuse big data analytics into a British-era method of estimating yield losses, called “crop cutting experiments”. The exercise was carried out in Kerala, Madhya Pradesh and Karnataka under the crop insurance programme.
India plans to scale up use of AI in a range of public programmes, according to a document of the Niti Aayog. In a letter, its CEO Amitabh Kant urged states and central ministries to explore use of AI in health care, agriculture and education sectors.
The agriculture ministry has issued fresh bids for tech solutions and similar trials to bring down the “unit” of the crop insurance programme to the level of “gram panchayat” or village during the 2019 June to September kharif season.
Launched in 2016-17, the PMFBY has been dogged by delayed payments to farmers. Prompt payouts to farmers suffering crop losses hinge on accurate estimates of yield loss, arrived at through crop-cutting experiments. Often, insurance companies tend to dispute yield loss data sent by states, rejecting or delaying claims. “This is where newer technologies can intervene and bring transparency,” an official said.
Norms require four crop cutting experiments at every village. This means the country must conduct about 7 million such experiments to estimate yields, an official said. This is the stage in yield-estimation process where high-tech is being brought in. For instance, the brief given to CropIn was twofold. One was to help reduce the number of farm locations for conducting crop cutting experiments to save time. Two, how could AI help give better idea of yield and output? CropIn used its remote-sensing tool called SmartFarm to glean information of the current and previous years to “identify homogeneity and heterogeneity of expected yield”, the firm’s head of R&D Richa Hukumchand said.
Smart Farm’s algorithms too can zoom in on farms through remote sensing and by reading pre-assigned digital signatures for specific crops, they can identify the crop, its maturity stage and other parameters. SmartFarm has estimated that crop cutting experiments can be reduced by 30%.