2023-12-05 01:32:50
Control a traffic light to regulate traffic, make an industrial process more efficient, prevent a fire and hear the chainsaws in a lush forest. These are some of the tasks that she has learned about artificial intelligence in recent years and that has made her an activist in the fight against climate change. This technology already accumulates a lot of experience in different sectors, but with the emergence of generative AI it is experiencing a true golden age. “It is thanks to the confluence of three factors: big data, low-cost computing and the development of complex learning models,” said Nuria Oliver, academic at the Royal Academy of Engineering and director of the ELLIS Alicante Foundation, a few months ago, in this newspaper.
The application of technology for decarbonization is one of the topics of debate at the climate summit in Dubai (United Arab Emirates), which is being held until the middle of this month. The capture and elimination of CO2 is the most cited, but AI makes it possible to reduce the emission of greenhouse gases and anticipate natural and environmental catastrophes that will leave companies with a bill of 2,660 million euros in 2022, according to the AON Foundation. Spanish companies and the destruction of 25,200 jobs.
This technology based on learning models allows optimizing innovation processes and that means optimizing energy consumption and reducing waste, but they are more useful in the prevention and detection of activities that endanger the environment. The NGO Rainforest Connect has put AI to listen and monitor the Carpathian forests in Romania. There, where Bram Stoker’s Count Dracula hid, illegal logging is depleting the ecosystem. According to several reports by Romanian NGOs, about 20 million cubic meters of wood are illegally cut every year in the country’s forests.
The project began with a smartphone attached to solar panels, now, almost a decade later, the project has improved. A constant record of sounds powered by solar panels and generating a huge amount of data and information. “Through a 5G connection, the information is sent to the cloud and there an algorithm detects if there are sounds of chainsaws,” explains Jorge Vega from Huawei Spain. “If it is detected, a geolocated alert is sent to the rangers,” he adds.
“False positives”
But AI is not only capable of listening or responding; She has also been trained to ‘smell’ smoke from fires. To this end, Grupo Sylvestris and the Repsol Foundation have placed dozens of sensors in thousands of trees in recent months to apply the maxim “if you see smoke, the fire is close.” Although its mechanism is not so simple, because “it can produce false positives if an old car that emits a lot of particles passes by or a barbecue is being held nearby,” say the promoters of the initiative.
The system, trained for this, is capable of analyzing the smoke captured by the sensors and if it detects four types of gases, an alert is sent to forestry agents that allows them to be on guard and begin a strategy in the event of a possible fire. Furthermore, thanks to this new technology and the data collected by satellites “it allows us to monitor and predict the progress of the fire, for more efficient management of resources,” they say.
This solution is already deployed in Las Hurdes (Cáceres, Extremadura), protecting a perimeter of 1,357 hectares in collaboration with the fire protection services of the Government of Extremadura.
The other face
Although artificial intelligence can help care for the environment, is this technology really sustainable? Precisely, the use of these energy resources is in the spotlight. “The greatest consumption occurs during the training of the algorithms,” says José Ventura, data & artificial intelligence tech sales at IBM.
A recent study found that the carbon footprint of training just one state-of-the-art deep learning model to perform natural language processing tasks is equivalent to the amount of CO2 an average American produces in two years.
Other research goes further and speaks of consumption equivalent to that of 136 Danish homes in one year. «AI can be used to optimize the use of water or energy in operations, discover more durable or less polluting materials, reduce CO2 consumption or improve recycling processes to bring companies closer to the circular economy. Although it is true that progress is being made in more efficient materials and architectures, these advances cannot compensate for the constant increase in demand,” explains Elena Gil Lizasoain, director of Artificial Intelligence and Big Data at Telefónica Tech.
Green algorithms to make AI sustainable
Artificial intelligence (AI) is present in practically all areas of life, whether domestic, work or business. However, the consumption of resources that it does is not so present.
«I do not have a direct measure of energy consumption, since my operation is managed by servers in data centers. The amount of energy used will depend on the infrastructure and energy efficiency of the servers hosting my model,” ChatGPT clarifies.
The consumption of this technology is based on its origin: in the energy needed by its data centers and in the water used to cool them. “It is also true that more and more data centers are looking for sustainable energy sources and that their manufacturers are investigating the use of new materials that allow savings and efficiencies in this field,” says Elena Gil Lizasoain, director of Artificial Intelligence and Big Data. at Telefónica Tech
#smell #smoke #fire