How does your artificial intelligence tool for conservation work?

Artificial intelligence (AI) is a term that refers to a broad family of models used to process and make predictions from large and diverse data sets.

We created a model using biodiversity datasets in addition to socioeconomic data. The aim was to discover optimal strategies for safeguarding nature. Our AI tool, Conservation Area Prioritization through Artificial Intelligence (Captain), uses a kind of AI called Reinforcement learning. This is a family of algorithms that optimize decisions in a dynamic environment.

The tool we developed was the results of years of labor by a global team with experience in biology, sustainable economics, mathematics and computer science.

The software we develop can take multiple kinds of data as input, including maps of biodiversity, species distributions, climate and predicted climate changes, in addition to socio-economic data corresponding to land costs and budget available for conservation activities. It then processes this information and proposes a conservation policy based on a specified conservation goal (e.g., including all endangered species in a protected area or protecting as many species as possible).

The tool’s environment is a simulation of biodiversity, a synthetic world with species and individuals that reproduce, migrate and die over time. We use the tool to look for probably the most appropriate conservation policy.

It works similarly to a video game, where the player (called an agent) is the “brain” of our software. The aim of the sport is to guard biodiversity and stop as many species as possible from becoming extinct in a simulated environment that comes with human pressure and climate change.

The agent observes the environment and tries to put protected areas on this environment in the most effective possible way. At the top of the sport, the agent receives a reward for every species he can save from extinction. It takes playing the sport again and again to learn how one can best interpret the environment and best place the protected areas. The model is then trained and might be used with real biodiversity data to discover conservation priorities that ought to maximize biodiversity protection.

Why did you test the tool in Madagascar? What have you ever found?

The Report on the State of the World’s Plants and Fungi showed that biodiversity is facing unprecedented threats and as much as 45% of all plant species are vulnerable to extinction. Along with climate change, that is considered one of the best challenges facing humanity as we depend on nature for our survival.

In a current one Paper We have summarized the extent of Madagascar’s extraordinary concentration of biodiversity, with 1000’s of plant, animal and fungal species. The project was led by Hélène Ralimanana from the Royal Botanic Gardens, Kew and the Kew Madagascar Conservation Centre.

By applying the Captain tool to a dataset of Madagascar endemic trees, we were in a position to discover a very powerful areas for biodiversity conservation within the country, corresponding to the realm within the Sava region where the Marojejy National Park has long been established.

Madagascar already has quite a few nature reserves and programs. Our experiment shows that the technology we developed might be used with real data. We hope it could function a guide for conservation planning.

Who do you’re thinking that can use the Captain AI?

We consider it could help policymakers, practitioners and businesses guide conservation and restoration planning. In particular, the software can use several types of data along with biodiversity data. For example, costs and opportunity costs related to establishing protected or restoration areas might be used. It may use future climate scenarios.

Is technology alone enough to preserve biodiversity?

Certainly not. Technology might help us analyze numbers and untangle complex data. However, there are lots of facets of nature conservation that can’t be easily quantified in numbers. There are facets of the cultural value of land and nature, in addition to social and political issues related to the equitable distribution of resources. These are problems that real people need to think about, not artificial intelligence programs.

Technology and science can (and will) help us make decisions, but ultimately the protection and preservation of the natural world is and should be within the hands of individuals, not software.

This article was originally published at