At first glance, the title of Dauvergne's new book, AI in the Wild , seem disturbing. The wild and the wilderness is something we are used to thinking of as nature itself, the free and untouched, free of human infrastructure, not to mention high-tech surveillance. Surprisingly, the author does not even take the time to mention such scruples: The premise is that the natural systems around the world are in such an all-encompassing crisis that it is a matter of saving what can be salvaged. As in the case of gene therapy and brain implants in medicine, it is difficult to argue against saving lives, but what kind of world are we creating when everything is monitored, hacked and modified?
Drone war against nature
His first example is the underwater drone Rangerbot. It glides around coral reefs equipped with high-resolution cameras and advanced software, which enables it to recognize a certain type of starfish with 99 percent precision. The drone is also armed and shoots these starfish with a poison that kills them instantly without harming other species. Such a drone war against nature sounds like a cruel invasion of a natural ecosystem. The crucial point is that the Great Barrier Reef off Australia – the only ecosystem visible from space – is dying. Global warming has led to higher temperatures and more acidic water, which causes corals to die – and half of the reef is already lost.
Temperatures and manure pollution from land also bring the living parts of the coral reef into disarray, as the starfish, which are usually in balance with the reef, bloom in uncontrolled amounts and eat the remaining corals. Rangerbot is a desperate measure to save the reef, together with its sister drone Larvabot, which for more peaceful purposes plants new corals in destroyed areas. It is an unpleasant thought that more and more ecosystems are being patrolled by robots, but here we may have to give in to technology, as one advises a patient to submit to surgery when the crisis is severe enough.
Protecting wildlife is difficult, and I have repeatedly heard conservationists complain that the penalties for wildlife crime are too lenient, since no one has the resources to monitor huge jungles or deserted sea areas.
One solution the book presents is the system Rainforest Connection. Based on simple technology – mobile phones charged with solar panels in the treetops – this system, with a relatively simple software, can recognize the sound of a chainsaw within a kilometer. Since over 90 percent of all logging in the rainforests is illegal, these instantaneous alarm signals give rangers and locals the opportunity to pull out and stop criminal gangs of timber that are gradually eating into the jungle. Using machine learning and further programming, the sensor system can even detect the sound of a jaguar moving over the forest floor: not by hearing the sound of its soft paws, but by recognizing warning cries from birds and animals.
More and more ecosystems are being patrolled by robots.
But what happens when those who are looking to plunder coral reefs or engage in illegal jaguar hunting, get hold of the same technology, drones and sensors, advanced analysis tools? Strangely enough, Dauvergne does not address this urgent problem when it comes to nature conservation on the ground floor.
In return, he discusses sustainability at an overarching level. Economic and political analyzes make it clear: The more advanced and smart the technological and economic systems become, the more efficient and accelerated the use of nature and the extraction of resources becomes. But can artificial intelligence be steered away from dangerous use – or at least partially channeled into good intentions and sustainability?
A crucial reflection in 3Dauvergne appears at the end of the book: "To rely on artificial intelligence to ensure sustainability is to establish a management-technocratic model, which has long been shown to fail the environment."
We can applaud projects like Microsofts AI for Earth, but as Dauvergne points out, the resources spent are negligible compared to other economic and military advantages.
From a philosophical point of view of technology, it is an open question whether technology should be understood as neutral, or as harmful to embrace. To take an example from Dauvergne: Facial recognition programs have been used by animal welfare activists to recognize the faces of chimpanzee cubs that have been illegally posted for sale on the internet. But the usefulness of this measure is fading quickly in relation to the use of face recognition to monitor and suppress environmental activists and protesters in China and elsewhere.
The more advanced and smart the technological and economic systems become, the more efficient and accelerated nature utilization and resource extraction are doomed to become.
Although Dauvergne is probably right that artificial intelligence will, by and large, lead to more effective nature destruction rather than contributing to nature protection, we cannot opt out of new technology. Dauvergne thus encourages a selective use of high technology and informatics, and warns against their limitation and destructive tendencies.
Given the title, we could have heard more about wilderness and conservation and less about gadgets, socio-economics, smart cities and farms. Dauvergne also points out that his book is the first on this topic, and that the new technologies built on artificial intelligence have only appeared in recent years.
Game crime and illegal logging are desperately easy to get away with. Environmental groups and individuals who have tried to save rainforests by simply buying them, quickly notice that they often get poachers and illegal loggers involved. They need mapping and intelligence. If nature management has so far been poor, uninformed and powerless, artificial means of making it more informed and effective are most welcome: preferably not as fully automatic systems of armed "robot rangers", but as tools for critical and vigilant conservationists.