A Practitioners AI Wishlist for Agriculture in India
Artificial Intelligence (or AI) is poised to revolutionize our world. There is a palpable sense of excitement across all industries about what AI has to offer to it; and Indian agriculture is no exception. However, AI models, at least in the case of agriculture in India, are in early stages. Also, while much of the discussion is around functionalities that AI brings to the table, relatively little is being spoken on the kind of end use applications that these will spawn. But, this is a natural evolution of technology adoption. Consider Google Maps, initially perceived as novelty by most; very few would have then envisioned that this would go on to power a bulk of today’s gig economy businesses such as Zomato and Uber. Similarly, even as AI transitions from a novelty to a serious business tool, it is not unreasonable to expect an influx of innovative AI based solutions to our everyday problem. Given this trajectory, it is also probably an opportune time to outline an AI wish lists for applications that one would like to see. So here is a ‘Practitioners AI Wish List’ for the Indian agriculture sector.
The following are five tools that this author believes could significantly enhance efficiency for two key constituencies — farmers and the last mile delivery actors (hereafter referred as ‘Farmer Intermediaries”). The latter are individual or institutional actors such as agriculture extension workers, para professionals, Farmer Producer Organizations (FPOs), Community Resource Persons working with NGOs etc., who disseminate information, provide advisory support and also help farmers transact with the world.
AI Tools for Farmers
1) AI Powered Crop Planner. This author envisions an AI based Crop Planner that advises the farmers on their cropping decisions, i.e. what to sow and when to sow, by analyzing multiple data points like rainfall and climate forecasts, ground water availability, global commodity market conditions ( for example, forward prices on CBOT) etc. Currently cropping decisions tend to be intuitive and/or based on past experience. Such as, farmers sowing more cotton following a year of high cotton prices. Cropping pattern changes, if they happen, are based on trial and error over a few years. For instance, some farmers in Maharashtra, told this author how they had changed their sowing of black gram from monsoon to summer sowing; due to frequent untimely showers. But this change happened only after 3–4 cycles of continued losses. A AI powered crop planner, as explained above, could have helped these farmers make an informed choice, thus reducing losses.
2) Market Price and Deals Advisor. Could an AI application advise farmers on likely price trends ahead of harvest? Commodity Traders regularly use sophisticated models to forecast prices, so, an AI powered Market Price and Deals Advisory is not an unrealistic expectation. This tool could not only predict prices but also advise farmers on what markets to sell in (after factoring in all relevant data on product variety, distance to market, transport cost, taxes etc.) and when to sell. For example, imagine a ground nut farmer in Bundelkhand receiving an AI advisory telling her to “hold her crop and sell after three months, since prices were likely to rise” and “instead of selling locally, sell in Rajasthan, as there are arbitrage opportunities working out there”. Such advisories would help small farmers explore national market, instead of being confined to local ones.
3) Unstructured Query Translator. It is reasonable to imagine that farmers in the near future will use generative AI to answers their queries. And, they may not be type out a prompt (like we do now), but are more likely to use a ‘speech to text’ functionality. However, their questions may not always be very structured; and an average farmer is mostly likely to not know of ‘Prompt Engineering’. An Unstructured Query Translator could help translate a farmer’s query into a proper prompt for the AI to work on. Currently, human operators in Kisan Help Lines do this work whenever a farmer dials in with a question. An AI tool to decode what the farmer is saying, will go a long way in helping farmers use generative AI tools more effectively.
AI Tools for Farmer Intermediaries
All of the above tools would also be equally valuable to a ‘Farmer Intermediary’. But, this author believes that additionally the following two tools are likely to make a ‘Farmer Intermediary’s’ job much more efficient.
4) Route Planner & Advisory System. An AI enabled route planner cum advisory application could optimize an intermediary’s — like say an extension worker of Agriculture department — response to a crop related problem being faced. Currently, an extension workers route plan is as per a pre-determined schedule. So, if an extension worker has 20 villages mapped to her; she is at best visiting a village only once in three weeks. That means, her ability to promptly respond to a crop related problem is contingent on her having been at the right place in the right moment. But, if an AI application were to coalesce date from satellite, on-ground weather stations and IoT devices to flag off issues in a village to her, optimize an efficient route map for her as well as provide her with necessary knowledge resources, she would be able to help farmers arrest losses faster.
5) Intelligent Knowledge Bank. A large amount of data on seeds, cultivation practices, technology etc. — sits with universities and research institutions in India. But, this is not only difficult to access for a ‘Farmer Intermediary’, but also difficult to comprehend. An AI application that combines a ‘Recommender System’ and ‘Generative AI’ to give out relevant information to the intermediary in easy to understand summary, would help her bring this knowledge to the farmer. This knowledge bank should be so intuitive enough that an FPO CEO searching for solutions to the problem of wilt in chickpeas, should be able to get relevant prompts to her related information pieces, in the same way that an online eCommerce portal prompts a user with products related to her internet search. An ‘Intelligent Knowledge Bank’ will help bring the knowledge of the world to the farmer finger tips in form of an actionable input.
Actioning these AI applications would at the basic need investments in localized weather stations, IoT devices, linked satellites that are monitoring farms in India etc. This author, in an earlier blog had written about the Government of India’s intent to digitize the Fasal Bima Yojana by putting in precisely these kinds of infrastructure at the Gram Panchayat and Block level. That could be a good start. Additionally, it would need a well-functioning commodity derivatives market, good internet connectivity in even the remotest corners and a team of domain experts who are developing these AI models. Additionally, monetization models need to be developed to sustain these applications. If done, these applications could potentially revolutionize the way agriculture is done in India.
All views expressed in the above post are my own and do not represent those of the organization that I work with.
