Using AI To Reduce Human-Animal Conflicts
As many as 949,000 bats and 679,000 birds die every year in North America due to collisions with wind turbines. Furthermore, more than a million European starlings were killed in 2022 by the U.S. Department of Agriculture to protect crops from damage. As the global human population grows and our self-proclaimed territory expands more and more into nature, this report argues that conflicts between humans and wild animals are increasing. In it, the author explores the possible use of machine vision-based artificial intelligence (AI) to reduce these conflicts and how it might lead to widespread concern for wild animal welfare.
Commonly, human-wild animal conflicts are resolved without the welfare of the non-human animal in mind. This often leads to lethal measures such as killing animals viewed as “problematic” or ignoring animal fatalities altogether. Cases involving endangered or beloved species (like elephants) are an exception, since public concerns influence decisions in conflict mitigation. Unfortunately, the author points out that the most numerous non-human animals who suffer as a result of human-wild animal conflict are typically neither endangered nor beloved.
With artificial intelligence on the rise, the researcher highlights the importance of developing and establishing concerns for wild animal welfare. Although AI systems show promise for many areas of conservation, they may be detrimental to wild animals if they are only programmed for the most efficient or profitable solutions (from a human perspective). Implementing AI that can mitigate conflicts with both human and non-human animal interests in mind might be an important first step to establishing wild animal welfare as an area of concern.
The author gives two examples of human-wild animal conflicts that may benefit from machine vision technology, a version of AI that can help identify and monitor wild animals with videos and images. The first is wind turbine collisions. As the world increasingly turns to wind energy, the author worries that bird and bat collisions (and fatalities) will likely increase without any interventions. Machine vision AI may help to reduce turbine collisions by identifying and classifying birds from afar, then predicting their trajectory. One AI system uses this information to shut off wind turbines when there are relevant risks of collision. In doing so, it may replace human observers who are currently tasked with identifying large birds at risk of colliding with turbines. In fact, in a trial, the technology reduced eagle fatalities by 75%-89% compared to a site without AI.
The researchers also highlight a prototype AI that can be used in agricultural lands to deter starlings and other birds. Like other machine-vision-based AI, this prototype can identify potentially crop-damaging birds and send out deterring signals when a flock gets too close to the produce. This could be a cheaper and more effective way of deterring birds and other animals, as other acoustic solutions run the risk of disturbing nearby animals or causing starlings to become habituated to deterring sounds.
The author suggests that small-scale implementation of AI systems might lead to lower prices and thereby wider adoption. This could then broaden interest in the field, encouraging more AI research and development. As more cost-effective, animal-friendly solutions are developed, the relevant stakeholders may adopt them without realizing they are prioritizing wild animal well-being. Ultimately, as it becomes clear that the welfare of wild animals doesn’t conflict with human interests, public concern for wild animals — even the ones labeled “problematic” — may increase.
There are a few challenges to implementing conflict-mitigation AI on a large scale, including coordinating experts and stakeholders and increasing funding. Animal advocates can play a role in this by campaigning for financial support from companies, governments, and NGOs to support the cause. The author also calls for an interdisciplinary conference to bring together AI experts with professionals who work in conservation and animal behavior.
In addition, advocates can help center animals in human-wild animal conflicts by launching public awareness campaigns that increase concern for overlooked animals, such as bats. Bats are affected by wind turbines in larger numbers than birds, but most bat species don’t have legal protections in the United States. If policies are implemented to protect bats and other animals commonly affected by human-wild-animal conflicts, it may be easier for AI experts to adjust existing systems to account for them.