Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
On the wetlands of Senegal, the researcher Alexander Delblink two drones for the aquatic birds: pelicans, flamingos, and savage. It flies the drone, but artificial intelligence analyzes the images of individuals in a herd, which speeds up the analysis of thousands of hours for each scan, as it is estimated. And the time from essence.
Since 1970, the numbers of wildlife decreased More than seventy percent. The world is in the midst of the biodiversity crisis, and according to some researchers, they are subject to it Sixth collective extinction. The planet previously carried five mass extinction events, with the last entry at the end of the Cretaceous era: the time of the impressive asteroid that launched the nuclear winter and the killing of dinosaurs. It was six ways before.
To save the species from the edge of extinction, you must first know what you have, and how much – which is often said is easier than doing, especially in the fields that many enjoy. Scientists appreciate less than 20 percent of the types of insects On the ground it was identified. After reviewing the artificial intelligence shots for the camera trap for only one week in Panama, the researchers say they are I found more than 300 species Unknown to science.
The hypothesis of artificial intelligence in scientific research is Not without critics. Supporters of high -tech in conservation cite the ability of artificial intelligence to analyze large data groups in seconds that may take months otherwise, in order to disintegrate artificial intelligence patterns in the reactions of species and their dispensed distributions, and to detect a continuous set of generators. Critics indicate its environmental impact, the possibility of bias, and insufficient ethical standards.
Many artificial intelligence work focuses on preserving the analysis of thousands of hours from the shots taken from remote cameras or weather investigative studies, but it is unlikely to end there. Currently, the researchers focus on processing clips with models of detection of objects, a type of artificial intelligence that can identify and identify objects within an image or video. These models are often built with tunnels (CNNS) and is trained to identify species, detect their existence or absence.
Projects that employ artificial intelligence “to save species” often generate the madness of the media. The researchers in South Africa generated a wave of major headlines asking whether artificial intelligence can save “The most lonely world plant.” Scientists have deployed drones on indisputable spaces from the dense Ngye forest in search of a female partner on male Cycad in Kww Botanical Gardens in London. Artificial intelligence has wiped footage of the signs of what is extinct in the wild, which researchers hope not to really extinct – just blocking under the canopy. But some say that these titles are exaggerated without looking at the consequences.
“There is a tidy wave of enthusiastic research on artificial intelligence applications and less important research that looks at costs, environmentally and socially.”
The training process for the artificial intelligence model, such as the LLM model (LLM). It consumes more than a thousand hours megawatts From electricity. Chaili Ren, whose research focuses on reducing the health effects of Amnesty International, says the less clear problem is water consumption for data centers.
Data centers contain the infrastructure needed to provide the treatment force for Amnesty International, and all technology must be cooled, usually through fresh water from local water supply. Because of the cooling needs, Amnesty International is expected to withdraw between 4.2 billion and 6.6 billion A cubic meter of water annually by 2027, many of which are lost to evaporate. The environmental impact is not equal, as technology giants export their databases abroad. The Google plan was launched to build new data centers in Latin America Huge protests in Chile and UruguayBiodiversity areas are already suffering from severe dehydration.
Ren said: “Data centers also create a general health crisis due to the air pollutants emitted, including the microscopic substance (PM2.5) and Nitrose.” The general health burden resulting from databases in the United States-is expected to cost primarily in low-income areas-will cost- Twenty billion by 2030.
“The models we run are not huge – they are great for us, but they are not like the huge data of the social network.”
However, the imprint of the work of most biologists from artificial intelligence, at the present time, is not mentioned. For his part, Delplanque has a local computer that treats images, and its Herdnet model – which helps in the population charges of extensively packed animals, such as elephants and antelopes on savanna – took about twelve hours of training, compared to LLMS that works on huge servers that work for weeks during the training process.
Laura Bullock, an assistant professor at McChel University, who aims to spread quantitative data aimed at spreading data aimed at spreading species that aim to spread species that aim to spread species that aim to spread species that aim to spread the species that aim to spread species that aim to spread species that aim to spread the species, “does not aim to spread AI -Indments Endisels:” We have this anxiety as scientists throughout Time: Do we really harm the environment we are trying to help?
But the world of computer environment, Tania Berger Wolf, says that current low -energy applications do not harness the full potential of technology, indicating the recognition of the images as “the artificial intelligence of the old school.” Perggar Wolf Wolok participated in writing a paper “”Unrealized capabilities From artificial intelligence to expand the knowledge of biological diversity.
“We want to go beyond the scaling and acceleration of what people already do for something new, such as generating test hypotheses or extracting non -visual patterns and groups,” says Berger Wolf.
“What we was doing with artificial intelligence so far is clear, which is all this quick disclosure of pictures and sound monitoring, but we must do much more: use artificial intelligence to ask the right environmental questions,” says Bullock.
One of the potential applications that generate attention, to both applause and condemnation, is the concept of artificial intelligence to decode animal communications. Earth Types Project AI and LLMs are used in the hope of building an translator to communicate with non -human life. There too CETI ProjectWhich focuses on the use of a similar approach to understanding sperm, which communicate through clicks similar to inherited symbols that can be dismantled, in theory, can be deciphered. Indeed, scientists were able to use machine learning It indicates that elephants address individuals In their family with unique names. But the biggest hypothesis of the disintegration of animal communications raises ethical questions and concerns about success. In other words: Will you work? Is it a waste of resources to try? Should we talk to animals at all?
“We have to choose the place where these models will make a difference, not just using them because you have a new shiny game,” Berger Wolf warned. Applications such as LLMS enhances a large environmental imprint, so it is not responsible for resource spending if the result of the research does not change. The data is a resource. ”
The models are good only like the trained data, which can lead to bias and poor preparation for conservation procedures. One of the most common problems includes spatial bias, where species are excessively represented in certain areas of data groups, and the legal bias, where charismatic species such as panda receives more funding, and thus more data are available on them easily from mysterious beetles. But Amnesty International can bias our perceptions and even the formation of the questions we ask. Written by paper On how to reduce LLMS environmental challenges.
He said: “There are much more options to provide Amnesty International to provide bias, resource extraction, and pay excess consumption more than the presence of conservation applications. “If we create the benefits of conservation for the effectiveness of the effectiveness of Amazon from artificial intelligence to make consumers buy more things, this is a largely uneven scale.”
In 2024, for its part, Google announced the publication of the artificial intelligence model to listen to coral reefs: Surfperch. Biochemistry plays a major role in assessing the stability of coral reefs – health coral reefs appear different – and Surfperch is analyzed sound signatures to measure the success of coral reef restoration efforts or define impending threats. As soon as the tool was published, Google also announced that it was less than the targets of the climate threatened Environmental demands of artificial intelligence.
“It is not hypocrisy to use artificial intelligence in preservation-it only should be used responsibly,” said Berger Wolf. But when it comes to organization, nor biological diversity or Amnesty International, they are accurately compatible with the geopolitical borders, I thought.