Artificial Intelligence in Agriculture

Issues and Prospects

Naveen P Singh &

Ranjith PC

Technology is the key driver of 21st century world. From the trunk dial to video call, steam engine to metro, sprayer to drones, technology has revolutionalised the life of humans across the world. Besides, it also played crucial role to industrialize the countries around the world and formalized their economy in a significant way. The world is now ready to embrace the idea of 4th Industrial revolution which is likely to be led by Artificial Intelligence (AI). At this juncture it is imperative to give a big push to a rather neglected sector; Agriculture, by leveraging AI so as to revive and sustain the sector for the generations ahead.

Technology in agriculture is a decisive element in enhancing the scale of production and productivity. As per Food and Agriculture Organisation (FAO) estimates, India has to double its agriculture output by 2050 to feed the growing population. In the wake of fragmentation and subdivision of land holdings coupled with less emphasis on collective and cooperative farming, the arable land cannot be increased. The only viable option before us is to increase the productivity of the available land through improved technology. However, it is a established fact that technology adoption in farming sector is very slow due to several inherent reasons. In addition the required skill sets at individual farmer level to operationalize the modern agriculture technology is less than required. Apart from this, technology has to face behvioural test (Technology adoption) of individual farmers. With these problems and challenges on hand, the solution lies in our finger tips, that is to bring AI led innovation to agriculture, which is having less human interface in decision making and high adoption of needed technologies.

Input to policy shift

Improving agricultural productivity is not a simple task. It involves activities starting from crop decision to final harvest and marketing of farm produce. Usually, many of our farmers rely on experiential knowledge/ traditional wisdom to decide on ploughing, sowing seeds, applying fertilizers and pesticides and harvesting the produce. But, it is less viable  in curbing the wide array of challenges viz., pest, disease, monsoon etc. leading to low yields and thereby face economic hardships. Therefore scientific intervention in ensuring farmer to be informed is crucial to enhance farm productivity.

Hence, the solution for these problems and challenges is to have a self-guided automatic model that understands the behavior of various stakeholders for the benefit of all. Moreover, Artificial Intelligence (AI) is part of the solutions towards improved agricultural productivity. AI can be described as simulated algorithmic computer models that mimic the human behavior.  The process begins with imagining a robotic friend/ installed application that guides our farmer when to grow, harvest and sell his/her produce and link our agriculture resource

persons to track the most pressing problems faced by farmers through searches (Google analytics) made by farmers so that they can act/react immediately. It can also act as digital data for policymakers for exploring various effective programme/ policy options and interventions thereof. Moreover AI would be prudent option to leverage before policy establishments who are toiling to double the farmers income by 2022. The collection of massive data by way of digitalization of agriculture is to give a new dimension to develop and deploy AI in agriculture. Last year, farmers of Karnataka and Andhra Pradesh rely on the messages for sowing crops delivered using AI by Microsoft and International Crops Research Institute for the Semi-Arid Tropics (ICRISAT). Globally, AI has provided wider opportunities to farmers, organisations and governments. For instance in Tanzania, Google open source of disease data free AI discovered disease with 98% accuracy and deployed robotics to pluck the weeds, as also been evidenced in other nations. However, it is early to gauge the impact of AI in numerical terms but surely AI will tranform the agriculture package and practices landscape.

Market prediction and supply chain management

Marketing is the critical factor in agriculture to deliver the fruits of growth to farmers doorstep. Which is very essential in bringing income security and rural prosperity. In the modern marketing system, two fundamental things drive the farm income to prosperity: grading and price prediction. Crops grown needs to be graded according to physical para-meters (chemical composition in case of fruits and nuts) through automated quality analysis (AI led with less scope for biasness) of images both at farm level and at market yards. This helps in two ways; firstly by knowing the exact parameters farmers can upload their produce parameters in online platforms lik e-NAM, e-GraMS and others, thereby reducing the cost of marketing, wastage and problem of middleman significantly. Secondly it helps someone who is sitting far from the place of produce to initiate the process of buying and reselling. Further forecasting market price is important to receive remuner-ative price by producers and algorithmic models that use the price data of previous years tell us the possible price. Currently there is practice of manual price feeding to predict the price. With the help of AI models automated price feeding can be made thereby lessening the scope of price manipulation and errors due to false feeding. In a significant development, a private firm Markets and Markets valued artificial intelligence (AI) in agriculture market across the globe to USD 432.2 Million in 2016 and is further expected to be valued at USD 2,628.5 Million by 2025, at a CAGR of 22.5% during the forecast period (considering 2016 as base year).

Another important factor that could bring a paradigm shift in predicting price and streamlining the flow of agri commodities is supply chain management. It brings radical change in absorption of surplus commondity (peak season) provided by AI aided software that meticulously manages commondity exchange among the stakeholders in the market (through backward and forward linkages of commodity). It uses the data of various markets (e-NAM and Agri Census data) and retail demand schedule at a particular time to give matching figure of space availability to each crop before sowing/ deciding. Apart from this, integration of value chains-stocks/ storage space (currently no functional mechanism to trace storage availability), procurement and ICT enabled banking service helps to bring globalization in value chains. Possibly to aid this endevour of market prediction and price forecasting limited AI model which predicts things on evolutionary ladder would be helpful.

Soil and water management

Soil and water are facing severe threat under modern agriculture practices involving inorganic chemical to influence the crop productivity. It hampers the efficient resource utilization and technology adoption in the long run that may result in reducing the yields and posing severe threat to food security. The most preferable type of AI to explore and conserve the benefits of soil and water would be Reactive AI, where it basically works on perception of surrounding conditions and acts on what it sees. Thus, any deviation to the healt of soil and water would be easily recognized so as to ensure fertility status of soil and quality of water. Further AI led recognition and deep learning models can also use the integrated data signals from satellite and compare it with the local farm image thereby helping to know soil health and propose immediate action plan to restore the soil health. Thus, it reduces the high cost of laboratory testing infrastructure in monitoring soil health and paves way for spot distribution of soil heath card. Berlin-based agricultural tech startup PEAT developed application called Plantix which identifies potential defects and nutrient deficiencies in the soil. This particular software algorithms also correlates particular foliage patterns with certain soil defects, plant pests and diseases thereby immensely saving on costs.

Nearly 89 pecent of ground water extracted is used for agriculture inefficiently posing a serious question of water usage and its management. This situation has placed India to the net water exporter thereby raising question on future water availability with increasing uncertainity of monsoon. India being poor resource utilizer in terms of increasing the yield of cereals (major chunks in food security) compared to USA and China, has to adopt resilient resource strategy. As stated above, Reactive AI models store the crop specific moisture requirement and assess the moisture content in the field using the remote satellite and signals the farmer through text message. This can also augment laying of auto irrigation from borewell. This helps in curbing the indiscriminate use of water in crop production.

 As the Indian soils are becoming the storehouse of chemical residues causing significant imbalances in soil health on one hand and yield gaps across different crops and regions, it is imperative to leverage AI to holddown the healthy soil to feed generation ahead. NITI Ayog has recentry cautioned about severe water crisis that country might face in next decades as nearly 70 pecent is contaminated. This sounds clear signal to look for an alternative methods for judicious resource use.  

Agri-Startup

Agri startups become key in unleashing the farm potential by providing the crucial technology. Some of the AI startups like Intello labs, Aibono, thrithi robotics and Satsure are succsefully using the AI to predict yield, stabilize the yield, soil analysis and predicting economic value of future yield. Further giving emphasis and incentives to AI led innovation and startups will add significant number of  emplyoyment. Given the fact that startup become new age tech giant in India that could be trusted to achieve sustainable agriculture development on one hand and securing food and nutritional security to the people on the other hand. This may go a long way in achieving sustainable developmental goals (SDGs) effectively.

Takeoff: Current digital drive

Digitalization of agriculture is on its path even though it has been ranked on bottom in the industry surveys on the state of digitalization. In India digitalization through Digital India programme reaching the villages through OFC is essential input in leveraging AI techniques in agriculture. With the fleet of ~30 million smart phone owning farmers and an expected increase of internet usage in rural india to 315 million by 2020, its an easy way to connect, communicate and coordinate. Accenture study reveals that digital farming and connected farm services can impact 70 million Indian farmers in 2020, adding USD 9 billion to farmer incomes.

Initiatives to increase the digital literacy are another benefit before us to thrust AI as a game changer. Moreover, information availability during crop production and marketing will help to reduce the crop and income loss, thereby realization of doubling farmer's income at the earliest. Probably, in the right direction,  NITI Ayog and IBM have signed a Statement of Intent (SoI)  to develop a crop yield prediction model using Artificial Intelligence (AI) to provide real time advisory to farmers in aspirational Districts.

Technology in India has travelled several miles touching the lives and aspiration of 1.3 billion people. It improved the lives of many people and agriculture has shown unprece-dented progress begining from high yielding variety to algorithm trade strategies/models. However, sector  still remain underdeveloped in terms of bridging yield gaps on production fronts and ameliorating supply side constraints thereby giving market and price access. Assessing demand and supply situations, crop competitiveness and regional crop planning can come handy with AI. Startups are emerging in agriculture and the technologies developed by them are matching the global standards and has aspirations to solve these problems through AI. However, needed support of government in disseminating their technology via  extension network is warranted in the initial stages. The immediate reform needed is to bring the AI policy to involve AI deployment to solve the urgent and most pressing problems on priority basis. This policy has to touch upon most pressing sector in current and vulnerable areas in future. The early steps taken to adopt the AI led innovation would propel the second Green Revolution with income security to farmers. Finally, for realizing futuristic benefits of AI across all stakeholder in agriculture, policy push with synergy between central and state governments is a must.

(The authors are associated with National Institute of Agricultural Economics & Policy Research, New Delhi.

E-mail: naveenpsingh@gmail. com)

Views expressed are personal.

(Image Courtesy : Google)