Food is not a privilege but a fundamental human right for the continued survival of the human race. In Nigeria today, there is the astronomical rise in population, forced urban migration, desertification in the North, coastal encroachment in the South, pockets of crisis in the middle belt, lots of unsafe practices on the farms affecting the environment and the reality of climate change all mitigating against food security. The need to ensure we can produce food consistently in a way that is safe for the environment without breaking the banks – there is a need to adopt technology-driven production.
In a time where machine learning and artificial intelligence are helping every other sector increase efficiency, productivity, and profitability, why not agriculture? Machine learning and artificial intelligence for the average Nigerian sounds like buzz words, for others it means robots or some other exotic machinery. I feel this has derailed the conversation especially in the agricultural sector as people tend to feel – it means taking away jobs (true but not the whole truth).
I will try to paint a scenario to explain these terms and how they are helping us ensure food security. If I were living in Ojodu Berger (Lagos, mainland) and work in Victoria Island (Lagos, Island), there are two major routes I can use to get to work – I can take the Ojota axis (road A) or 3rd mainland bridge axis (road B). I assume I take the 3rd mainland bridge when I drive and use public buses (BRT) on the Ojota axis. Every morning I will need to make a decision on which route I will take but this decision is predicated on my observation of which is faster, safer, likely traffic, which days of the week I will encounter traffic, etc.
Every day I observe the world and make a mental model of what applies per day, so on a day I will close late, I know the BRT buses won’t be available so I drive and take the 3rd mainland bridge route. In my everyday observation of the world, I am collecting information (data) which I will use to form my model (between the world and my model formation, I am learning) – and from my model I can now make a decision easily.
Machine learning in simple terms is using machines to learn the scenario in the real-world to make a mathematical model, and when this model informs decision making, this becomes artificial intelligence. I will use another example of a farmer who sells tomatoes and wants to know the best selling price to ensure maximum profit. The farmer will begin by selling these tomatoes at various prices and keep records of sales made in the period. The documentation of this information is called data, so we differentiate data from observation in our field. (If I watch a movie and I come back to tell you the story, that is observation but if show you clips of the movie or even give you the movie, that is data, data is not bias). After a few months, a graph of sales versus price is drawn from the data to get a model (a mathematical model) that will guide my decision (maximum profit). So from the computed data, I can now infer:
Maximum profit (objective) = (sales * Profit) – cost
From the graph, I am able to see in my profit/price graph the best price to sell. This is the same way we are using machine learning and artificial intelligence in agriculture.
In our nursery for example, we understand that one of the factors that helps ensure good germination rate is temperature, light, and humidity. These we know from data collected over time, we are able to get the optimum temperature, light (intensity and duration) and humidity for the seeds to grow, so we create a small device which understands what is happening in the environment per time (using sensors) and adjusts the temperature (heat mat), light (create shade or remove shade) and humidity (allow irrigation) in the system.
Machine learning depends on statistics (data means nothing without proper interpretation and we decide the interpretation based on objective) as well as machine learning (using Python) and then we use this in guiding future decisions.
In summary, I will add Machine learning and artificial intelligence alone will not solve the issue of food security but it will play a big role as well as with other solutions to ensure together food security is achieved. We must always remember food production should not be seasonal because hunger is not seasonal, and in our push for food production we must ensure that the process and product of food production must be healthy for the farmer, the consumer, and the environment. The quest for the development of tools to aid food production must be such that it is not a burden on the farmer – if the farmer has to spend years paying back then they are indeed working for everyone but themselves, so we understand the need to ensure maximum efficiency and productivity without undermining profitability for the farmer.