Interview with Saad Chafki, Senior Vice President of Information Technology at Sollio Cooperative Group.
Cooperator: Firstly, to get up to speed, what is artificial intelligence (AI) and how does it serve in the agricultural sector?
Saad Chafki: Artificial intelligence serves to leverage the data of a company, specifically in agriculture. Using computing and primarily mathematical and statistical concepts, it helps understand and interpret data to assist humans, or a machine, in making decisions and formulating predictions. That’s the somewhat theoretical definition. It can be used in various sectors. For example, AI can facilitate, through algorithms, the detection of fraudulent transactions on credit cards.
Cooperator: In brief, how does this technology work?
Saad Chafki: There are three main steps. Firstly, it’s about understanding the situation, what is happening. This is the descriptive level. Let’s take the following case: someone uses a credit card to make a gasoline purchase at a local gas station. A few minutes later, 150 km away, another gasoline purchase is made with the same card. There’s a problem, obviously, and the algorithm will detect it.
The second step is predictive. Without a doubt, there’s fraud. This immediately leads to the third step, prescription, which is to refuse the second gas purchase and cancel the card. These three steps have been completed, without human intervention, in a matter of seconds. Depending on the situation, we can stop at the first (descriptive) or second (predictive) step. In other words, we don’t have to go through all three steps every time.
Cooperator: Why is there so much talk about it now?
Saad Chafki: One might think it’s because of the increased strength of algorithms, but not at all! We use the same mathematical principles as we did over 50 years ago. What has evolved exponentially is the availability of data, their storage cost, and the computing power of computers.
Today, in the span of two or three days, we generate as much data as we did in the past 40 years. How? Through our smartphones and the presence of numerous sensors. Our daily activities leave a phenomenal amount of digital traces, hence data.
But it’s all well and good to generate all this data, if I can’t store it, there’s a problem. In a decade, the cost of storing data has literally plummeted tens of thousands of times. Finally, the computing power of computers has developed exponentially. What took 18 hours to execute ten years ago now takes only fractions of a second.
All of this has contributed to making data more accessible to companies. And thanks to cloud computing, they don’t have to equip themselves with supercomputers, they just have to “rent” time to test algorithms, which is a huge advantage for startups and small businesses. IT equipment is no longer a barrier to entry.
Cooperator: Can you give us examples of AI usage in agriculture?
Saad Chafki: AI will use the power of data and computer processing capabilities to predict, for example, based on the size, characteristics, and past history of batches, the growth of a herd and the level of feed in a silo, which will have a huge impact on the supply chain and the quality of service offered to agricultural producers.
A supply chain that optimizes transportation by delivering feed orders from neighboring companies in a single trip will be less costly and generate fewer greenhouse gas emissions. Another advantage, and not the least, is that it will prevent producers from checking feed levels in their silos themselves, with the risks of falls that this can entail. Every year, accidents, sometimes fatal, are related to this practice.
Cooperator: In animal production, how could it take shape?
Saad Chafki: Take the example of a poultry farm. Computer vision could be used there, with cameras analyzing the birds, detecting abnormal behaviors (limping, crowding), and measuring their weight to predict the ideal exit date and time. This is currently being tested at a Sollio Agriculture farm. The potential is enormous. This same technology could also take shape in the dairy sector with the analysis of cow comfort, feed refusals, calving predictions, and lameness.
Cooperator: What about ChatGPT, another AI tool, and what are its advantages for agricultural producers?
Saad Chafki: The public ChatGPT from OpenAI offers much more depth than a Google search. That’s why the search engine quickly launched its own AI tool, Gemini, just like Microsoft with Copilot. Briefly, these generative AI tools (which generate content) improve through feedback, meaning they store what their users will consider to be good answers. They allow producers to obtain information in a domain they know little about, such as a new product, for example. These are good introductory or demystifying tools, but they won’t provide in-depth information in their field of expertise where they already know a lot. And let’s be clear: they won’t replace expert advice at all.
Lastly, as a word of caution, it’s important to know that the data entered into public ChatGPT, or any other public AI system, is probably no longer confidential. It’s important to be cautious and not share personal or sensitive information.
Cooperator: It’s also important to ensure that the recommendation actually has an impact on my business, isn’t it?
Saad Chafki: Yes. It’s important to be aware of the limits of recommendations. Are they supported by tangible evidence in my region, in my type of business, according to my growing conditions? All the more reason to confirm recommendations with my expert advisor. I also recommend using AI in a digital platform you already use to avoid scattering yourself everywhere and having a plethora of tools that don’t communicate with each other. It’s costly and inefficient. We need to limit technological fragmentation.
Cooperator: What influence will artificial intelligence have on the way agricultural producers and expert advisors work?
Saad Chafki: It will favor predictive analysis. With accumulated data (history of my crops, yields, weather, input prices, production costs, sales, etc.), AI tools will soon be able to make recommendations. Sollio Agriculture’s Lactascan program is a powerful tool for descriptive analysis in dairy production. It provides very valuable insights into operations. Data interpretation is very important. I repeat: we will always need the advice of experts in nutrition, health, and management. AI tools are an added value that will improve the work of experts and producers.
Cooperator: How do you see the future?
Saad Chafki: Tools will become more refined, the costs of their use will decrease, and their predictive capabilities, enriched by data, will improve. Artificial intelligence will play a major role in agriculture.
TRACKING CHICKENS
Ongoing trials in poultry farming buildings at Sollio Agriculture study the use of cameras to track the growth of hens and chickens. “We can observe them one by one,” explains agronomist Pascal Labranche, director of products and innovation at Sollio Agriculture. It’s not perfect yet, but the trials are conclusive. We’re inevitably moving towards closer monitoring.”
Density, weight, feed consumption, behaviors and movements of the birds, injuries, mortality… the possibilities for analyzing the data collected by the cameras are immense. ”
Photo : iStock.com | Fahroni
Article seen in Le Coopérateur
Text by Patrick Dupuis patrick.dupuis@sollio.coop
