AI2’s AI for the Environment Teams Innovate Toward a Sustainable Future
Harnessing AI and data to protect our planet
As world leaders gathered in New York City for last week’s United Nations General Assembly and the concurrent NYC Climate Week, there was a notable presence across borders and politics, transcending the confines of traditional diplomacy to tackle one of the most pressing issues of our time: the protection of our planet.
While AI2’s world-class research and engineering achievements are well-documented through the platforms, datasets, and innovative research we have contributed to advancing AI for the common good, our investment in the restoration and preservation of our planet might not be as widely known. Behind this commitment lie several dedicated teams, each accelerating the use of AI and data-driven solutions to confront our environmental challenges. From EarthRanger helping keep wildlife and habitats safe to Skylight’s state-of-the-art AI tackling illegal fishing, a climate modeling program that’s improving the way we understand climate change, and efforts to apply machine learning to wildland fire management, we delve into the work of these teams driving forward the technology needed to empower those protecting the planet.
EarthRanger
EarthRanger is a no-cost, open-sourced data visualization and analysis system that gives conservationists the real-time information they need to keep wildlife, habitats, and communities safe. The platform collects, integrates, and displays remote sensing data and field reports to effectively monitor collared wildlife, rangers, enforcement assets, and infrastructure within protected areas and landscapes.
The application of EarthRanger is diverse and varied based on the primary needs of the area or organization. Used at over 500 sites in 60 countries around the world, EarthRanger helps monitor vast areas, track and study wildlife movements and behavior, keep personnel safe, and proactively mitigate human-wildlife conflict — one of the biggest challenges facing wildlife today.
As we look to stem the alarming decline in global biodiversity, EarthRanger will be important in facilitating the reintroduction and restoration of diverse species and ecosystems, both of which are paramount for delivering crucial environmental and socioeconomic benefits on a global scale.
Skylight
Skylight is a no-cost maritime monitoring and analysis software platform providing analysts and protected area managers with a near real-time tool to identify potential illegal, unreported, and unregulated (IUU) fishing activities. IUU fishing isn’t only depleting our fish stocks, but also putting marine ecosystems, coastal economies, and regional security in peril. Each year, experts suggest the crisis of IUU fishing costs the world’s economy an estimated $23.5 billion. If that wasn’t enough, IUU fishing is heavily linked to slavery, human rights violations, and transnational organized crime.
By harnessing the power of advanced AI — including established techniques from computer vision and more recent innovations from large language modeling — Skylight empowers over 300 organizations in nearly 70 countries with an essential intelligence and analysis tool needed to safeguard and restore our seas for the long term. For the latest on Skylight’s activities, read more about the role AI and AI2 are playing in protecting the high seas.
Climate Modeling
The climate is changing fast. Even with a rapid transition from fossil fuels to renewable energy, all sectors of society must prepare for many decades of further warming and extremes of rainfall, drought, wildfire and snow/ice meltback, as well as sea-level rise that will disrupt human and natural systems. To address this critical need for high quality data and climate predictions, AI2’s Climate Modeling team is one of the first group to use machine learning (ML) to make climate models much faster, far more accurate over long time periods even at local scales, and broadly accessible, truly democratizing climate modeling.
Conventional climate models encode physically derived equations on a global grid to simulate changing weather patterns and land and ocean conditions for decades or centuries. They are more accurate when run on a finely spaced grid (e.g. 5 km horizontal spacing) than with a coarse grid (e.g. 50–200 km horizontal spacing). However, this is so computationally expensive that it is impractical to use for simulations longer than a decade, even on the world’s largest supercomputers.
AI2's Climate Modeling team is developing machine learning emulators based on these fine grid models and other data that are reliable, cheap, and easy to use by specialists and non-specialists alike. With collaborators at two leading US climate modeling centers and in other research groups, the team has already made strides toward this goal, and more is in the works. For further discussion of the project, and recent news and announcements, visit our Climate Modeling website.
Wildlands
While wildfires have long been a natural component of ecosystems, the combined forces of climate change and shifting land-use patterns have escalated their severity, as highlighted in this United Nations report. Unprecedented wildfires have erupted in regions as diverse as the Amazon, Australia, and even the Arctic Circle, unleashing uncontrollable and extreme blazes that are having devastating impacts on communities, wildlife, and ecosystems worldwide. Consequently, it is imperative that conservationists and scientists are equipped with the best possible tools for wildfire detection and mitigation.
The Wildlands program at AI2 — a team of AI researchers, engineers, and designers — is working together to apply machine learning and computer vision to empower land managers, wildland practitioners, and communities through the application of advanced AI-driven tools. By allowing land management professionals to use AI to zero in on ground-level fuels, which include the various materials that serve as potential fuel sources for wildfires such as grasses, shrubs, trees, and decaying leaves, the team is addressing a critical void with their Fuels Data app that current satellite technology cannot measure. These efforts are bridging a significant gap that might otherwise remain unaddressed. By taking a current low-tech methodology and applying computer vision, the Wildlands program will greatly amplify the impact of fire management. Read more about AI2’s Wildlands team and their new collaboration with FireWatch to increase community fire resilience.
As world leaders conclude their activities, marking the end of the United Nations General Assembly and NYC Climate Week, AI2’s commitment to the protection of our planet remains steadfast. With an eye toward the future, AI2’s teams continue to drive innovation, foster collaboration, and demonstrate unwavering dedication to paving the way for a sustainable future. In this vision, the balance of the needs and safety of humans and the health of our environment is preserved, enabling a path forward for the well-being of our planet and its inhabitants for generations to come.
Follow @allen_ai, @SkylightMarine, and @EarthRangerTech on Twitter/X and subscribe to the AI2 Newsletter to stay current on news and research coming out of AI2.
