What we currently are as a civilization is the result of a temporally long process that encompasses, among other factors, evolutionary transformations, cultural and behavioral plurality, adaptive capacity, scientific discoveries, and technological development. Over the centuries, society has been adapting to different scenarios, conditions and, most importantly, transforming the world around it, in order to create the necessary conditions for its existence and perpetuation. This inventive capacity of man is, perhaps, responsible for the survival and prosperity of our species, as some more recent studies point out.

In a context of continuous evolution, transformations, sometimes slow, sometimes more accelerated, slowly shaped the society we are today. In this timeline, some milestones such as the discovery of fire, the domestication of plants, the invention of the wheel, the discoveries of some laws of nature and the industrial revolution, for example, represented dizzying leaps in this process. Perhaps the most recent leap is the development of information technologies, bringing with it a chain of evolutionary processes, which are no longer separated by centuries, but by a few decades or years.

Artificial Intelligence is part of this era of technologies and has caused a revolution in the way we deal with our surroundings and how we perform tasks with different levels of complexity. The emergence of Artificial Intelligence dates back to the early 1940s, when Warren McCulloch and Walter Pitts (North American scientists) published a scientific article reporting the creation of a mathematical model for an artificial neuron, based on the functioning of the human neuron.

From the 1940s to the present day, advances have been made in an attempt to build artificial systems capable of simulating and even replacing human reasoning. The areas of knowledge that have benefited most from such advances are industrial and communication. In forestry, in turn, the transformations promoted by information technologies are more recent and still discreet.

Artificial Intelligence tools have promoted a revolution in production fields. Through its insertion in various forestry operations, Artificial Intelligence can further increase the productivity of planted forests, helping to face the current challenge of meeting global demands in the coming years. Recent data from the Food and Agriculture Organization of the United Nations points to a demand of 3.1 billion cubic meters of wood in the year 2050, which represents a growth of more than 35% compared to current demand.

Defined as the ability of devices, systems and machines to simulate and reproduce human reasoning, Artificial Intelligence removes subjectivity from processes and allows the simultaneous processing of a large volume of data, learning and defining patterns from a massive amount of information. Artificial Intelligence accelerates the interpretations of natural phenomena and, more importantly, decision making. In forestry, Artificial Intelligence can be seen in autonomous machine systems, in sensors designed to remotely obtain information about crops, in controllers that, in an automated way, make decisions and make interventions in real time.

In addition to the applications described above, many others have been developed. The large number of these solutions based on Artificial Intelligence come from machine learning models, whether deep or not. Based on these principles, systems capable of learning patterns and classifying and/or predicting phenomena inherent to forestry production provide quick and reliable information, culminating in greater efficiency in the use of general inputs, optimization or replacement of human labor and greater effectiveness in management practices.

The use of machine learning algorithms follows a sequential process that begins with the construction of patterns and ends with the evaluation of the accuracy and precision of the models. With regard to the models used, these can be grouped into those of classification and regression. The purpose of each one indicates its applicability, with the classification ones being intended to identify features and/or phenomena in a sample universe, while the regression ones are intended to predict these patterns and/or these phenomena. Several algorithms are intended for both functions, only changing the way they are applied. An example of these “multipurpose” algorithms are artificial neural networks, whether simple or multilayer.

Using artificial neural networks, we developed a computational solution for predicting eucalyptus regrowth after the application of herbicides. The solution classifies the spectral response of the sprouts and compares them with the patterns defined in the learning process, identifying the variations that are suggestive of a regrowth and predicting when and where it will occur.

Another Artificial Intelligence algorithm widely used in forestry is Random Forest. When associated with high spatial resolution images, obtained with orbital or aerial sensors, this algorithm allows the identification and segmentation of targets in a complex environment such as a forestry production area. With this algorithm we developed a system for identifying weeds in eucalyptus production areas. This system has an accuracy of over 90% in classifying weeds and defining infestation levels and over 80% in segmenting plants according to their leaf morphology, separating them into “broad leaf” and “narrow leaf”.

Entering the deep learning environment, the complexity is substantially higher, however, with equally superior possibilities of use and application. With models of this class, it is possible, for example, to model trees in cropping systems and count plants and planting failures. Plant counting is a demand in forestry, as it is unthinkable for this operation to be carried out manually. Counting plants is essential for accurate establishment, for example, the survival rate in plots after transplanting seedlings.

The implementation of Artificial Intelligence in forestry is a continuous journey, since, with the digital revolution, models and systems are dynamic, with periodic updates and evolutions. The ability of companies in the forestry sector to understand the importance of Artificial Intelligence and, even more so, to identify its positioning in the different forestry stages will determine the course of transformation that the sector will experience.

In a scenario of increasing scarcity of natural resources, the use of technology is a viable way to reconcile productivity and sustainability. The evolutionary process of our time will lead us towards an irreversible scenario of digitalization, with greater connectivity, with increasingly sophisticated algorithms and intelligent machines. The lack of improvement, scaling and, mainly, detailing, will impose a new order, dictated by digitalization and permeating the use of Artificial Intelligence.