Bioregional Applications of AI
Examples
Felix Rosen:
Planet Labs: combining satellite imagery with machine learning to track changes in forests
"Amy Rosenthal described how Planet Labs combines satellite imagery with machine learning to track changes in forests, habitats, and other landscapes. In Costa Rica’s Osa Peninsula, for example, Osa Conservation combines satellite information with field observations and species movement data to inform wildlife corridors and conservation decisions.
Amy emphasized that satellites cannot supply everything practitioners need. Understanding species and conditions beneath a forest canopy requires observations on the ground. Communities’ knowledge of a place helps interpret what the measurements mean. In the Osa example, the different observations help identify where species move and where protection or restoration could support those movements.
GainForest: placing small audio recorders in forests and using AI models such as BirdNET to help identify species
Sharfy Adamantine, speaking on behalf of GainForest, described placing small audio recorders in forests and using AI models such as BirdNET to help identify species in the recordings. The reason for using AI here is straightforward: the recordings accumulate faster than people can listen to them.
But a species detection does not establish that an ecosystem is recovering. Sharfy stressed the need for observations over time to assess ecosystem health and explained that models are less reliable where training data is sparse. GainForest also works with communities to decide what information may be published. Sharfy described communities identifying species they did not want photographed or shared. She did not need to understand every reason for a refusal to respect it.
Regen Network: helping communities shape conservation agreements
Gregory Landua from Regen Network described how, in Regen’s experience, AI’s greatest contribution has been helping communities shape conservation agreements. Incorporating community priorities into agreements that may last decades can be harder and more expensive than ecological monitoring. Too often, communities receive contracts already drafted for them. Gregory shared Regen’s use of PACTO (Participatory Assembly & Co-creation Through Technology) to support deliberation through AI-assisted transcription and summaries that participants correct and revise toward an agreement.
That deliberative process also raises a question: who controls the knowledge people share? Gregory proposed a knowledge commons with explicit conditions governing who can access information and for what purposes. Open technology can help protect sensitive information: locally run models let communities process it without sending it to external services. Open software need not mean public data.
To connect local control with ecological coordination across regions, Gregory proposed “nestedness”: keeping some information and decisions local while sharing knowledge and developing institutions at larger scales. Nestedness is an approach often used in bioregional organizing and finance. Communities should be able to participate without surrendering authority over their knowledge.
Cobalt: seagrass restoration in the Gulf of Maine
Finally, Glenn Page of COBALT brought a place-based perspective, focusing on seagrass restoration in the Gulf of Maine. Glenn described efforts to understand which existing meadows could supply seed for restoration, drawing on field observations such as flowering cycles, water conditions, and the reproductive potential of existing meadows. COBALT’s Living Atlas is meant to bring this fieldwork together with other sources into a shared platform for learning and decision-making, so practitioners and communities can better understand these systems.
COBALT isn’t using AI yet. The Living Atlas today is a system for connecting and making sense of data, and they’re partnering with an AI research group at Tufts University to add an AI layer on top of it later. Glenn emphasized that any digital representation of the ecosystem should help people return to and engage with the place itself, not stand in for it. The underlying work still involves getting underwater, observing the meadows, collecting seed with care, and listening to communities."
(https://thebiofiproject.substack.com/p/can-ai-catalyze-the-regeneration)