![](https://www.energetskiportal.com/wp-content/uploads/2024/07/Andreja-Stojic-photo-340x217.jpg)
Artificial intelligence (AI) is increasingly being used for environmental protection, offering ever-increasing possibilities. We cannot see and translate most of the connections that determine the state of the environment into equations. This is a prerequisite for understanding cause and effect. To acquire scientifically based knowledge, the solution lies in using sophisticated algorithms that underlie different areas of AI and can detect and quantify them. We spoke about this with Professor Andreja Stojić, PhD, senior research associate at the Institute of Physics in Belgrade and lecturer at Singidunum University.
Q: In which areas can AI play a role in overcoming environmental challenges?
A: AI’s role can be multiple in all segments related to the an adequate depiction of the context from a large number of angles. In this sense, perhaps the closest segment is air quality because it means data of sufficient spatiotemporal resolution. The data include concentrations of pollutants, meteorological data, and data on people’s mobility, as well as numerous political and economic indicators that show society’s pulse that affects air quality.
IN FOCUS:
- GREEN TRANSFORMATION OF THE REPUBLIC OF CROATIA
- EU SUPPORT FOR GREEN TRANSITION PROJECTS IN SERBIA
- WALK, RECYCLE, EARN TOKENS
Q: What is your evaluation of air pollution monitoring? Does it need to be improved, and if so, in what direction?
A: Society does not understand the complexity of the problem nor what needs to be done to improve the current situation. Let’s imagine a situation where the treatment is taken out of the doctor’s hands. Contrary to the need, environmental quality management is left to decision-makers who set up monitoring through regulations. However, the organization, expertise, ethics and ability of decision-makers result in rules that do not follow scientific research. Decades ago, research proved the presence of thousands of times more toxic, mutagenic and carcinogenic compounds than the number of types mentioned in the relevant regulations. This research recognizes the necessity of contextualizing air quality, including a large number of variables or applying concepts of data analysis adequate for the analysis of complex systems. There are tens of thousands of different pollutants in the atmosphere of the urban environment, with hundreds belonging to the hazardous group. However, the regulations recognize single-digit prime numbers. The representation of what we measure, the information based on which we acquire knowledge, is extremely small compared to what we would have to know if we adequately dealt with air quality. Improving the monitoring would have to involve increasing the number of types of pollutants that are measured and the number of measuring stations, adjusting regulations to modern scientific knowledge and concepts, establishing different ethical principles and strict application of regulations.
Q: How can artificial intelligence contribute to reducing the concentration of air pollution?
![](https://www.energetskiportal.com/wp-content/uploads/2024/07/daniel-sessler-LWC8vH5qgAo-unsplash-340x217.jpg)
A: Basically, the primary level of AI application involves the contextualization of phenomena, for example, the concentration of a certain pollutant in the air, using machine learning algorithms, then the explanation of the obtained models using explainable artificial intelligence, as well as different approaches to grinding and generalizing the results. In this way, we can understand the environmental conditions that lead to pollution, the scale of influence of emission sources, meteorological conditions, physical and chemical processes and many other factors that shape air quality. The intermediate level involves simulations through virtual experiments that enable scenario development and prescription. This way we can create a basis for preventive measures or planned action to reduce concentrations. The final level involves integrating data, results, and their interpretation with overtrained language models, which enables further research through communication in natural language, generalization, and simplification and lowering to the decision makers’ level. The application of AI in air quality management includes developing and implementing systems for continuous data analysis, emission source identification, pollution forecasting, automated decision-making support, and the creation of valuable data aggregations. Traditional air quality indices do not provide adequate information about the actual condition.
Interview by Mirjana Vujadinović Tomevski
Read the whole interview in the new issue of the Energy portal Magazine NATURE CONSERVATION.