CAMBRIDGE, MA – Early on a Monday morning in September, scientists, engineers, and problem solvers from a diverse collection of backgrounds trickled into the Samberg Conference Center on MIT’s campus. In the following two days, numerous presentations were given that centered on conservation-oriented, and technology-focused projects.

These two areas of interest may seem contradictory, but their coupling is the backbone of this unique workshop, Advancing Conservation through Artificial Intelligence (AI). The workshop was co-hosted by MIT’s ESI and its external partner Conservation International (CI) with sponsorship from the MacArthur Foundation.

In her opening remarks, Daniela Raik, Senior VP and managing Director at CI, stated that “the tools in the conservation toolbox haven’t changed much in thirty years.” However, the accelerations that are happening in artificial intelligence and machine learning could pave the way for new devices that will make processes like wildlife management and data collection more efficient and accurate for conservationists.

The workshop got underway with an address by MIT Media Lab Director Joi Ito, followed by a panel of speakers from Conservation International, moderated by Ito, to discuss technology needs within conservation. A key issue that emerged is the importance of addressing human behavior to enact any real change: technology alone is not a panacea, but rather is a key component of conservation solutions along with policy, economic incentivizes, and regulatory compliance.

Day One talks highlighted an autonomous robotic fish used to map the sea floor, neural networks that can identify different types of plankton and calculate biodiversity in a defined area of ocean, and drones used to capture images to detect species. These were followed by a breakout discussion that served as a catalyst for creating concrete projects that will intertwine AI with conservation.

On the second day, speakers added to the rich set of conservation science needs and opportunities for AI to fill those needs: modeling ocean waves, an existing restoration project in Massachusetts using sensors with AI capabilities to track regrowth, and deep learning within Amazon Web Services.

The workshop culminated with teamwork to brainstorm applications for artificial intelligence on novel conservation-aimed tools. In the end, seven project concepts were presented:

  1. Autonomous surface-water vehicles equipped with AI acoustic technology and soundscape libraries to monitor large scale marine protected areas
  2. High resolution spatial mapping of recruitment and mortality rate for profitable fish markets, such as scallops, in order to implement dynamic fisheries management
  3. Autonomous water vehicles to map shallow water habitats in isolated areas coupled with AI controlled adaptive sampling to monitor these critical ecosystems.
  4. A filter that can pick up micro plastics from polluted “super sites” throughout the ocean, and characterize the size and composition of plastics it is receiving with particle imaging, Rahman Spectroscopy, and chemometric analysis
  5. An app and/or camera sensor that utilizes image and pattern recognition software to accurately identify species by customs enforcement officers to stop illegal exotic pet trade
  6. AI-equipped sensors that will alert officials when poaching/illegal activity is occurring in order to increase monitoring efficiency
  7. A machine-learning based tool that takes sensor-collected wildlife data, from many different organizations who currently hold the data, and enables efficient data management, archiving and analysis of what is on an image or audio file and provide scientific results in a user-friendly format

This collaboration between AI engineers and scientists has opened up new doors that will benefit both fields. It is easy to see how artificial intelligence will advance conservation, and during the workshop, participants articulated a similar conviction that the implementation of these new technologies in the complex field of conservation will also improve AI.

Contributed by Camila Cortina; ESI Programs Intern