How Artificial Intelligence Could End The Harmful Tagging of Animals

Bryan Birrueta, Go Ogata, Shane Staret, Andrew Whitig

The Tagging Problem

Figure 1: This bird was just tagged with a Geolocator and is supposedly harmless to it.¹ However, it has significant impact on its reproductive success.

Animal tagging has become the most popular way to track animals and collect data of themselves and their habitat.² There is a standard rule that no biologging technology should weigh over 5% of the body mass of the animal in order not to significantly impact its behavior or health. However, recent studies have shown that the 5% rule has no grounds on which it stands on. As a result, more and more animals have been negatively affected by biologging technology, leading to impacts of survival, reproduction, and foraging behavior.³ Not to mention the obvious point being that no animal enjoys getting tagged.

Thus, we wanted to find a potential alternative to tagging animals for the purpose of tracking. We believe that cameras around the animal’s habitat is a much safer and less invasive way to keep track of them. An issue with this concept is that people may have trouble identifying these species.⁴

The AI Solution

Figure 2: AI could be a way to track wildlife better.

To help solve this issue, we decided to use a neural network AI architecture. Our intent is to get the ball rolling in this field of study by being able to design a system that specifically works to identify dogs apart from other wildlife. Over time, we hope others will expand on our work to include other animals and different variations of each one to help conservationists identify species.

The AI can be used to keep track of species and populations within particular areas without humans needing to constantly monitor the data the AI is processing. The automation of monitoring invasive, endangered, and locally new species can be critical for understanding an environment’s ecosystem.

Knowing how to track and identify animals has significant implications for ecology. For example, suppose you want to survey the amount of species in a nature preserve. A trail cam may end up having millions of pictures of the area. The only feasible way to manually look for any specific species is by crowdsourcing. However, using crowdsourcing and human analysis are usually very inefficient and prone to error, and it can be difficult to organize the data in meaningful ways.

With the capabilities of our AI, it can look into these datasets and have a much more consistent identification than compared to crowdsourcing. By extension, this can allow for a greater coverage of regions and make trail cams a conservationist staple in the field. Trail cams can be added in remote places like the African savannah to look for species that were not identified before in the region. Over time, biologging technology could shift to a less invasive approach and have the same information as if the animals were tagged.

The Side Effects

Figure 3: Hunters and Poachers will try to find a way to use our solution to their advantage.

The implementation of our AI is a more ethical and less invasive approach to animal tracking. However, as with all AI, there are ethical and moral complications to consider. For one, our AI could, instead of being used for conservation purposes, be used for the tracking and hunting of rare, endangered species. This is an issue that current biologging technologies have, and our AI is not immune to the possibility via misuse, hacking, etc.⁵

Figure 4: More data on animals could mean higher costs for insurance.

There is also the possible consequence that home insurance prices will increase. With greater wildlife tracking, insurance companies may use the information from our AI to increase premiums on homeowners insurance for those in wildlife “danger zones.”

To minimize the side effects and follow the ACM Code of Ethics, we recommend to only allow the current AI to try to identify dogs (until other adjustments are made to allow for other animals, species, etc.). For the expanded version of this AI, we recommend only allowing conservationists to use this AI and keep the data hidden from potential poachers and insurance companies. We do acknowledge that hiding this information from them implies hiding it from the public in general. However, in this case we feel as though the security of the animals is more important than the trust of the public, as the conservationist organizations that will be using our AI system are likely already trustworthy within the communities around them.

Another consideration is the fact that this form of tracking will unfortunately not end tagging completely, as researchers still tag animals to obtain data on animals’ physiological performance, energetics, foraging, migration, habitat selection, and sociality, as well as the environments in which they live.²

Conclusion

Image identification and analysis is an area of significance to ecological conservation. Animal tagging is an unethical and invasive form of tracking and shouldn’t be used to the extent it’s being used today. Efforts have been made to reduce the use of tagging, but they are generally inefficient and prone to mistakes. Our AI has the potential to change the biologging process to make it less invasive while still retaining similar data that tagging provides with better accuracy and efficiency than traditional alternatives.

References

[1] Ornithology, B. T. (2019, May 31). Bird tracking — a masterclass. Retrieved from https://www.bto.org/understanding-birds/articles/bird-tracking-masterclass

[2] Wilmers, C. C., Nickel, B., Bryce, C. M., Smith, J. A., Wheat, R. E., & Yovovich, V. (2015). The golden age of bio-logging: How animal-borne sensors are advancing the frontiers of ecology. Ecology, 96(7), 1741–1753. doi:10.1890/14–1401.1

[3] Gregg, J. J. (2019, January 25). The Ethical Debate Around Innovation in Animal Tracking. Retrieved from https://psmag.com/.amp/environment/the-ethics-of-better-animal-tracking

[4] Hooykaas, M. J., Schilthuizen, M., Aten, C., Hemelaar, E. M., Albers, C. J., & Smeets, I. (2019). Identification skills in biodiversity professionals and laypeople: A gap in species literacy. Biological Conservation, 238, 108202. doi:10.1016/j.biocon.2019.108202

[5] Mark Bridge, T. C. (2017, March 02). Poachers intercept tagging signals to hunt down endangered animals. Retrieved from https://www.thetimes.co.uk/article/poachers-intercept-tagging-signals-to-hunt-down-endangered-animals-q508z85np