AI Can Assist in Wildlife Protection but Data Gathering is Key

Artificial intelligence (AI) will make it easier to determine the best methods of protecting endangered wildlife in the coming years, with current methods making little use of technology in meaningful ways, an expert told Digital. 

“Over time, if you feed that data into an AI engine, you will get better and better information modeling or warnings and maybe even be able to start doing things like warning you ahead of time when you think that might happen. You can track their migratory patterns and kind of say, well, every time they do that, within the next 12 hours, they almost always get into a situation of danger,” Phil Siegel, founder of AI nonprofit the Center for Digital, told Digital.

“The technology is what retailers use to track employees to make sure they’re not stealing, and that kind of stuff in a store and gives you warnings and alerts,” he said.

As nonprofits and tech developers alike look at the many ways they can implement AI to solve the myriad problems plaguing the world, conservation and animal protection has arisen as one of the chief goals. 

The southern Indian state of Tamil Nadu has implemented the use of AI-based surveillance systems to catch when elephants cross the state. The state’s high court in 2021 ordered the forest department and railways to prevent further deaths after 36 elephants died over the past decade, the BBC reported. 

Tamil Nadu initiated a trial program with 12 towers along two rail tracks, each with a camera capable of thermal and visible light imaging – not to mention standard livestreaming of the scenes in question. 

The key to this technology is computer vision, which focuses on the visual component of possible AI use. This is the technology that will help automatic vehicles detect obstacles and avoid them, and really the means by which the rubber hits the road when it comes to robots and their abilities to interact with the world around them. 


The bulk of heavy lifting comes from the AI models, such as ChatGPT, but computer vision will provide more immediate and practical uses for AI. In Africa, Dutch tech start-up Hack the Planet and British scientists at Stirling University have tested out new technology that works with satellites to detect elephants in an area and alert authorities. 

The cameras also were able to help scare elephants into a village in search of food, preventing a potential conflict from occurring. 

The Tamil Nadu tech, part of an $860,000 project, has similar implementation but instead communicates warnings to train operators and officials so they can slow down their approach and give the elephants time to clear out before the trains resume their journeys. 

The system has picked up around 400 instances of elephants approaching the tracks, but the system will detect any animal in the area, potentially overwhelming the four workers who must continuously monitor the system. 

However, the state has determined the system is effective enough to try and roll it out to five other areas in order to expand protection efforts. 

Siegel noted that the technology will have wider applications than wildlife protection. In fact, it is the same technology that could prove useful in general emergency response situations, such as detecting a budding wildfire before it becomes a significant problem, potentially saving lives and preventing millions in damages. 

“If you don’t have a ton of data, you’re using human intelligence to do it, but over time, as you collect more data and you see more patterns, it’s harder for humans to see what those patterns are,” Siegel argued. “You can use the AI to do a better job at both warning and alerts in some areas to do a better job at surveillance and then do a better job at response as well.” 

“AI today is by far a higher level of us case or more use cases than the stuff that’s getting the most press, which is the large language models (LLM) like ChatGPT,” Siegel added. “Up until recently, there’s been a lot of AI used and it’s pretty much only machine learning types of algorithms.”

“That’s what’s been popular and very successful in a lot of areas, but, obviously, with LLMs, that’s kind of taking on a life of its own and started to overshadow just the machine learning stuff… but not at school and in companies,” Siegel noted. “Those things are still very important there and the bulk of the applications today.”