In a modern greenhouse, a computer can now monitor temperature, humidity, lighting, carbon dioxide, watering and nutrients. Cameras monitor plant growth and fruit ripening, and algorithms can change some of the conditions automatically. This raises the question of whether artificial intelligence can grow better tomatoes than a human.
The short answer from expert Nikita Rumyantsev is: not completely. The entrepreneur and founder of SkyFarm Ecosystem told “Gazeta.Ru“ that the result of a change in temperature or watering is often understood after a few days, not minutes. Therefore, the algorithm cannot predict all the consequences of a given decision on its own.
“A person cannot be at every bed 24 hours a day. A computer does not have this problem“, says Rumyantsev.Where AI is already showing an advantage
Artificial intelligence is especially useful for tasks that require constant monitoring and processing of large volumes of data. It can compare readings from multiple sensors, detect deviations in the microclimate and support decisions about watering, lighting and ventilation. Systems can also recognize stress in plants, count fruits, detect signs of diseases and pests and predict yields.
A cherry tomato experiment at Wageningen University and Research Center shows the potential of this approach. For six months, five international teams remotely controlled five high-tech greenhouse compartments using artificial intelligence algorithms. A sixth compartment was controlled by humans. Within the set goal of combining yield, quality and resource consumption, all AI-supported teams achieved a better result than the human-managed greenhouse.
However, the result is from a controlled experiment, not proof that machines are better than agronomists in every situation. The researchers themselves note that automated plant condition control is still more limited than automated climate control.
The sensor does not know when it is wrong
The main risk is related to data quality. If the sensor provides inaccurate information about temperature, humidity or soil moisture, the algorithm can make a logical decision based on incorrect input data. In practice, this can lead to inappropriate watering, ventilation or nutrient dosing. A scientific review of smart greenhouses indicates that signal interference and sensor limitations are among the obstacles to reliable implementation of the systems in real conditions.
Rumyantsev sees the role of AI as a digital extension of the agronomist's experience. The system can record how a particular variety was grown, how much water and light it received and what yield it yielded. This allows the specialist to compare individual production cycles and find out which decisions have yielded results.
Such a model also corresponds to the position of the Food and Agriculture Organization of the United Nations, which supports the responsible and human-centered use of artificial intelligence in agriculture. For now, the most realistic scenario is not for the agronomist to be replaced, but to work with an algorithm that continuously monitors, calculates quickly, and warns in time.
Source: Gazeta.Ru