Artificial intelligence in environmental management
AI techniques (Deep machine learning, Deep reinforcement learning, etc.), can recognise patterns in a seemingly chaotic mass of information and help to make decisions not only in the diagnosis of problems but also in their resolution. Their power lies both in being able to manage and interpret a large quantity of data, and also in combining and linking structured (numerics, logic, etc.) and unstructured (maps, images) information.

The bias seen in traditional interpretation can be reduced by managing diverse hypotheses in parallel during the modelling phases and trying out multiple possibilities. The process is completed by validating the hypotheses and therefore optimising decision-making in uncertain situations.
In the world of natural environment modelling and in particular the contamination of soil and groundwater, atmosphere, surface water etc., this can be a powerful tool in contrasting predictive hypotheses from traditional methodologies. For example:
- Applying machine learning to the resolution of differential equations in simulation models
- in the joint management of various types of environmental and industrial operation data (photographs, maps, non-stop logs, lithology, etc.)