Three ways in which AI can stop nature’s decline

Biodiversity loss has been dubbed one of the top global crises of this century. Nations have stepped up their commitments to protect and restore ecosystems, with some recently pledging to protect a third of their lands and oceans.

As artificial intelligence continues to gain ground, an increasing number of initiatives to use this technology for conservation are seeing the light of day. Here are three of such projects presented on Tuesday at AI for the Planet, a global virtual conference organised by the UN Educational, Scientific and Cultural Organization, the UN Environment Programme, Startup Inside and Microsoft.

1. Tracking where the wild things are

Over a million species are threatened with extinction, according to the UN. Their disappearance would have cascading, catastrophic consequences for ecosystems and for the services they provide for humans and the planet. But monitoring efforts are limited, small-scale and underfunded. This is according to Wildme, a non-profit organisation partnering with Microsoft’s AI for Earth project that supports organisations applying AI to environmental challenges.

The American organisation has launched Wildbook, an open-source software that allows researchers to track wildlife. Users can flag sightings of animals by uploading pictures to the platform. Through a computer vision algorithm, the software scans the images and digitally tags individual animals based on data such as unique markings, location, species, sex, etc.

What stands out. This project has citizen science at its core. By allowing anyone with a camera to participate, it can collect enormous amounts of data, which would otherwise require time and funds that conservationists do not have to spare, and at the same time raise awareness by encouraging public engagement.

“All you have to do is take a picture,” said Wildme director and co-founder Tanya Berger-Wolf at the conference.

2. A data-driven carbon market

As governments and businesses roll out net-zero pledges, carbon offsetting, or compensating for one’s CO2 emissions by buying someone else’s emission reduction efforts in credits, is becoming increasingly popular. But with little to no market standards, verifying the quality of those carbon credits can be difficult. Natural Capital Exchange, a data-driven market where people can sell and buy forest carbon credits, intends to bridge that gap.

The climate tech company SilviaTerra, also a partner of AI for Earth, used satellite imagery to create a high resolution forest map of the US that collects data not only on the number of trees but also on other characteristics such as species.

What stands out. The company has opened its market platform to all landowners in an effort to “[democratise] access to forest carbon markets”, as stated on its website. Landowners enrol their property, promise not to cut down their trees and get paid in return by corporations looking to buy carbon credits. SilviaTerra takes care of measurements and monitoring “to ensure that all of these credits are transparent, accurate, data driven and science-based”, explained Max Nova, founder of Silvia Terra.

“We're able to sort of connect the dots between the corporations that have these commitments and the millions of landowners across hundreds of millions of acres that actually own the forests, which can be an enormous tool in the fight against climate change,” he said at the event.

3. ‘Maps of hope’

Last year, the UN Development Programme launched a project to help nations in their conservation and sustainable development efforts. Kazakhstan, Costa Rica, Uganda, Colombia and Peru were chosen as pilot countries to map out their essential life support areas using spatial data. These so-called maps of hope are interactive web tools designed to help the countries identify areas that are critical for their environmental targets.

What stands out. These maps can be tailored to the needs of every country. “Governments can dial up and dial down different values, depending on their own national goals,” said Jamison Ervin, manager of the UNDP’s Global Programme on Nature for Development.

Costa Rica is using the map for their new Payments for Environmental Services Programme, through which landowners will be compensated for their conservation efforts. Uganda, which has been struggling with disasters including landslides and flooding, is focusing on identifying areas where rehabilitating nature can help reduce the impact of such disasters.

The downside of AI. As more tools are developed and become more powerful and sophisticated, conservationists need to think of the ethical implications, many panellists warned. AI-driven technology  could also be used for malicious purposes, for example, by poachers,  to spot animals.

Experts have also pointed to the risk of misinterpretation.  “The opaque nature of some ML [machine learning] algorithms means that the potential for unintended consequences may be high and this could have real-world consequences for people and wildlife,” an article from Nature Machine Intelligence states, citing as an example a species being wrongly assessed as extinct.

“AI in and of itself is not the solution. Only in partnership with humans to help provide insight and scale and accelerate the solutions, AI can truly be helpful,” said Berger-Wolf.