On September 24, 2017 Bloomberg hosted the fourth annual Data For Good Exchange. According to Bloomberg, the conference enables data scientists from academia, industry and public sector actors from government and NGOs to build relationships, share insights and progress, and encourages them to work together on applying modern machine learning and data science methods to challenges in the public and non-profit sectors (Data for Good Exchange 2017). The event held panels that addressed novel methods for collecting data; collaborations to address the global refugee crisis; utilizing data to aid vulnerable and youth populations; and the effects of big data and judicial bias in the criminal justice system. Throughout the different panels, major themes appeared regarding the ethics of data collection and research. Specifically, panelists touched upon the possible unintended consequences of research. Although various panels and papers discussed these effects, an organization called Wildbook presented the most concrete consequences.

Wildbook is an initiative that merges crowdsourcing, computer vision and data science for conservation efforts to combat animal extinction. It aggregates photographs and videos of animals to provide information regarding endangered animal populations across the world. Wild book’s technology merges algorithms and machine learning to identify animals through their physical characteristics such as cheetah spots or whale flukes. According to Wildbook, the organization combines vision and user interaction design to create a quick and accurate edge mapper to identify animals by their individual characteristics (Services).

According to a data scientist at Wildbook, the team launched the organization in order more accurately count populations of endangered species. Prior to Wildbook, scientists estimated animal populations throughout the world utilizing satellite imagery, a measurement practice that has proven to be imprecise and costly. Now through the usage of Wildbook’s technology, tourist photography and YouTube videos can be scraped according to their dates and geo-location. The blending of algorithms and machine learning can then detect the animal’s specific physical characteristics to collect information on the animal’s network, travel patterns and population size.

Although some benefits of Wildbook include the ability to more accurately gauge the number of certain endangered species, other benefits involve tracking mobility patterns, animal social networks, and individual animal wellbeing. Through the use of embedded geo-tagging properties in cameras and videography, scientists can track a population’s or an individual animal’s movement patterns. For example, the software allows scientists to research and track humpback whale travel patterns across the Atlantic and giraffe migrations in Africa. Although beneficial for animal behavior research, capturing movement patterns also allows scientists to observe the effects of severe weather conditions such as hurricanes and floods on animal populations.

Wildbook also allows scientists to observe an animal’s social network. It derives the information from photographs and videos where animals appear together, as well as through similar geo-tag location. Collecting information on an animal’s social network may be beneficial for studying animals’ behavioral patterns and species interactions. Scientists also benefit by tracking an animal at the individual level. For example, Wildbook can identify the exact animal depicted in a photograph by name and can present the animal’s history, including when they first appeared on Wildbook, what other animals comprise its network and where it was last seen. Providing information on an individual animal allows researchers to check into how the animal is living and examine its wellbeing.

Although Wildbook provides a collection of beneficial information that can influence scientist’s ability to study animal populations, the organization discovered impactful unintended consequence for the animals represented through the software. Wildbook’s population estimation, location and network capabilities facilitate the tracking of these animals, a feature poachers have began exploiting. Due to the software’s ease of use, location precision and it’s benefit of being up to date with scraping occurring daily, poachers can more easily find the geographical location of the animals they wish to hunt through the use of the software. According to the Program for Ethnographic Research & Community Studies (PERCS), a dilemma may arise in which the software pits the interests of the researcher against the interests of the community (Merz, 1998). In this case, the “interests of the community” lie in the wellbeing of the animals themselves, who most certainly don’t care about being assessed or tracked, but do suffer due to poaching activities.

During the conference, Wildbook researchers stated that they had not yet devised a solution to poachers exploiting the software. Although the organization pride’s itself on being accessible to everyone through open-access software, this makes it easier for poachers to obtain geo-location information on the animals. According to PERCS, careful consideration should also be given to publication and distribution channels when presenting research and data collection. They state that as we think about methods of reporting, we must also think about the locations of that reporting (Merz, 1998) as they may be impactful to subject of the research.

Overall, although scientists generally choose to perform research in order to raise awareness or solve problems facing individuals or populations, the research they perform carries weight and might sometimes have unpredictable effects. In the case of Wildbook, the benefit the software provides is important, but the consequences to wildlife are significantly impactful and might create greater repercussions for the populations being studied—endangered species throughout the world. Wildbook should weigh both the benefits and negative implications of their research as well as their information dissemination methods in order to ensure that their research subjects are not being disproportionately disadvantaged.

 

References

Data for Good Exchange 2017. (n.d.). http://www.bloomberg.com/company/d4gx/.

Services. (n.d.). http://www.wildme.org/services/

Merz, T. (1998). The Ethics of Fieldwork. http://www.elon.edu/docs/e-web/org/percs/EthicsModuleforWeb.pdf.