By Dr Timothy Njagi
Unlocking Agricultural Transformation: Harnessing Artificial Intelligence to Shape the Future of Farming in Kenya
THROUGH the Agricultural Sector Transformation and Growth Strategy (ASTGS) and the Bottom-Up Economic Transformation Agenda for Inclusive Growth, the Kenyan government targets transforming the agriculture sector. This involves improving agricultural productivity and production, improving the incomes of agricultural producers, growing agricultural exports and reducing agricultural imports. Among the key strategies available to the government, artificial intelligence (AI) provides an extra arrow in the quiver to pursue these objectives.
AI involves multiple science disciplines to mimic the cognitive human ability to address developmental challenges. In agriculture, the use of AI has gained traction in the past decade to address outstanding challenges and focus on emerging ones.
For example, in the developed world, acute labour shortages and an increase in weather variability due to climate change amid declining natural resources motivated the use of AI to mitigate these challenges while at the same time raising efficiency and productivity. The key AI branches include machine learning, computer vision, natural language processing, artificial neural networks, robotics, and expert systems.
These are usually backed up by deep learning and modelling to improve the precision of analysis. The use of AI in agriculture is exciting because it is the least digitised sector. On the farm, AI technologies have helped identify pests and diseases and improve production efficiency through precise fertilisation and watering. Computer vision, coupled with machine learning, has improved the detection of pests and diseases, where AI algorithms have been essential in helping farmers take early action and minimise losses due to pest and disease attacks. Besides, AI has been critical in monitoring crop health, leading to healthier crops. AI algorithms have also been vital in combining soil, weather, and historical crop performance data to generate recommendations to farmers that optimise resource use, such as nutrients and water, thereby minimising wastage and maximising output. Besides, integrating AI algorithms with robotics has also been used to automate some of the production activities fully. Locally, the adoption of AI at the farm level faces several challenges. First, Kenyan agriculture is dominated by smallholder farmers.
These farmers need higher levels of digitisation and mechanisation. Furthermore, many smallholders act independently and only come together to aggregate produce. This makes the use of AI costlier. Secondly, we lack data. Publicly available data is more often at variance with farm-level observed data. For instance, the yield reported by the National Bureau of Statistics and the Ministry of Agriculture will be at variance with the actual yield data estimated using crop cuts.
This is mainly because the official data is self-reported or an expert estimate. Lack of accurate data affects the preciseness of AI models. Third, much of the infrastructure required to use AI effectively is not available, and the capacity of many of the actors in agricultural value chains to use AI is low. This makes building capacity, transferring knowledge and building partnerships to support continuous learning necessary for utilisation of AI in agriculture.
AI models of the farm have been used to provide resources to farmers, change how they market their produce and improve risk management of farming enterprises. AI financial solutions can generate data that assess a farmer’s creditworthiness in real time, reducing the time taken and costs of establishing risk profiles of farmers. Besides, predictive lending models can help tailored financial solutions and decisions for farmers. Another use of AI technologies off the farm is at the market level. Farmers lack real-time information about markets in terms of demand and prices.
Further, even when they access such data, they are unlikely to use it individually. Some of the AI-driven solutions are around helping farmers aggregate and meet specific demand and, in turn, get better prices while reducing the length of the value chains. These solutions also take advantage of the progress made in digital/mobile payments and provide an opportunity to structure a system from the current informal system. Lastly, for farmer to embrace farming as a system, understanding their enterprises and the risks they face is critical to inform decisions on the investments they need to make. Predictive AI models help estimate yield based on weather information and historical yield performance, can promptly model the expected performance, and can inform critical investment decisions. AI can be extended to provide muchneeded advisory services to farmers. Leveraging mobile phone applications, AI can be tailored to deliver real-time advice on production and marketing practices to farmers enhancing their ability to effectively manage their farms.
In addition, AI can significantly improve decision making for policymakers and enhance governance along agricultural value chains. AI models can generate scenarios based on decisions and aid policymakers maximise benefit from public policy.
Clearly, AI will play a critical role in the future. For this to be effective, the country must address the current challenges for effective adoption of these technologies. The following actions can greatly enhance adoption of AI technologies.
- Build the capacity of farmers and other stakeholders to utilise AI technologies. Digitisation of farmers is critical for enhancing access and utilisation of digital tools and AI. Despite the wide usage and coverage of mobile phone technologies, the use of digital tools is low, calling for comprehensive training and support to be provided to farmers. Such capacity building will be helpful in getting farmers to learn and adopt digital tools and AI technologies. In addition, it’s essential to ensure that such capacity-building efforts are inclusive to build trust and enhance acceptability.
- Increase public investment in infrastructure development to utilise digital tools and AI technologies. The penetration of digital tools and AI relies on internet and electricity connectivity, which are unreliable in many rural areas. Effective operation of AI systems rely on cloud computing and real-time data updates, while reliable energy in the form of electricity is necessary for powering devices and communication systems.
- Invest in generating high-quality, accurate and reliable agricultural data, which is essential for training AI models. Besides, it is also important to harmonise the data systems to ensure that data is highly interoperable and can be integrated with data from other sources, such as satellite imagery. This is due to the differences in data formats when pooling data from multiple sources.
- Enact policy and regulations that support use of AI and digital tools. Clear regulations and standards must guide the effective use of AI. The regulations and standards should address concerns over data privacy, data security, intellectual property, and ownership.
- Create incentives for investments, especially by the private sector. The government can provide subsidies, grants, and affordable f inancing options for hardware, software, and other infrastructure necessary for enhancing the adoption of AI and digital technologies. This will lower investment costs and afford a better return on investment for private investments, especially where there is no data on the level of return on investment.