By Aufa Ananda Rizqi Wibowo & Assoc. Prof Dr Hasmahzaiti Omar

Wildlife monitoring is conceived as a surveillance of natural environments or any of their components. As such, it is done by large fieldwork such as long-term observation and recording of parameters. Over time, researchers have started to optimise and innovate the way they analyse the large growing data and monitoring scale. One of the ways researchers optimise and tackle the challenge of wildlife conservation is through the use of AI technology.

In monitoring efforts, AI has significant impact in assisting data collection and image identification, as the machine learning algorithm can analyse and process data faster than doing it manually. In 2023, the World Wildlife Fund (WWF) used camera traps that had AI-recognition in Eastern Australia to initiate a wildlife monitoring project. During their research, they were able to capture more than seven million images of wildlife of various species, including koalas and wombats.

However, to process those vast number of images it would take years to analyse each of them one by one to gain information. Thus, WWF introduced an AI camera data trap called Wildlife Insights to automatically identify what species was in the photo as well as filtering out blank images. The Wildlife Insights was able to identify 1,300 different species, some just from features of the animals, like the tail and nose of a greater glider.

AI and innovative techniques in wildlife conservation

AI can also contribute to innovative techniques in wildlife conservation. For example, AI-powered drones can be equipped with thermal-imaging cameras, allowing researchers to gain valuable insights to behaviour, population dynamics and habitat preferences of various different species that the researchers couldn’t otherwise do without the AI technology. Applications include monitoring wildlife intrusions by mammals such as elephants, tigers, monkeys, wild boars, tapirs and palm civets, as well as managing smaller mammalian pests like rodents in oil palm plantations, paddy fields, vegetable farms and other agricultural areas. These challenges are particularly prevalent in Malaysia’s agricultural sector.

Besides, in 2022, researchers in the Ngoye Forest, South Africa, utilised drones with multispectral sensors to detect and attempt to save an endangered cycad plant species called Encephalartos woodie. The use of the AI drone surveys allows them to scan approximately 195 acres of the forest to find more of the plant species. While the search continues, the drone also helps to be proven effective in identifying and mapping cycad populations. This approach allows them to provide non-invasive means to monitor and protect endangered species in remote and inaccessible areas.

Despite the advantage AI offers to the wildlife monitoring scene, there are ethical issues regarding the practice of its usage. The reliance on AI training more often than not involves collecting sensitive data about wildlife and their habitats, potentially causing infringement on the privacy of local communities. It also may capture unintended human activities in the wildlife by the usage of constant camera or drone monitoring, thus leading to further ethical dilemmas regarding surveillance and consent. The issues above have to be prevented by incorporating data privacy. Transparency should also be prioritised, and the public should be informed that AI technology was used in the wildlife conservation efforts.

Bias in AI collection

Another biggest issue introduced by the AI system is the bias in AI collection. AI that is trained in incomplete or under-representative datasets might misidentify or overlook animals, as they will then lead to imbalanced conservation efforts due to some animals getting more attention than the others. This presents the question about the reliability and objectivity of AI-generated information in wildlife media conservations. It’s important to ensure accountability and responsible decision-making for the use of AI as well as maintaining multiple approach to the conservation efforts to mitigate bias in the data collection.

There is also the issue of efficiency and cost effectiveness of AI technology. The over reliance on AI in wildlife can lead to unintended consequences for the local community and economies, as the traditional methods of wildlife observation such as employing the locals and photographers may be overshadowed by AI technology. This can cause disempowerment in jobs and tourism, as the livelihoods of those who rely on media production for their income can be affected.

Despite the previous ethical concerns, it is crucial to consider the integration and potential benefits of AI into wildlife conservation. The integration of AI introduces both challenges and challenges that demand ethical considerations, thus it’s important to approach these technologies with balanced perspective. By addressing the ethical issues such as privacy, ensuring correct datasets and diverse representation and having transparency with the local communities, AI can further enhance our understanding of wildlife and contribute to future conservation efforts and projects.

-- BERNAMA

Aufa Ananda Rizqi Wibowo is an undergraduate student of Applied Sciences at UCSI University. Dr Hasmahzaiti Omar is an Associate Professor at the Institute of Biological Sciences, Faculty of Science, Universiti Malaya.