Role of AI for Bioregional and Placed-Based Regeneration
* Report: Deep Dive on what AI can do for bioregional and placed-based regeneration.
URL = https://thebiofiproject.substack.com/p/can-ai-catalyze-the-regeneration
Summary
Felix de Rosen:
"On November 3, the BioFi Project organized a Deep Dive on what AI can do for bioregional and placed-based regeneration. We convened this conversation because we believe that learning how to relate to AI is one of the most important questions of our time. Its development is driving resource extraction, expanding opportunities for surveillance, and concentrating power over information, including knowledge drawn from local and Indigenous communities. But its potential benefits to bioregional initiatives are also undeniable, from tracking ecological change, to modeling watershed dynamics and informing where to focus and how to sequence investment in ecological regeneration.
These possibilities leave bioregional initiatives with practical questions: which tools are reliable enough to use today, what skills and resources do they require, and who controls the information they run on or that you may feed into them? For the November 3 Deep Dive, we invited representatives from Planet Labs, GainForest, Regen Network, and COBALT to share their work and examine where AI is useful, where it is overclaimed, and how these tools could support action on communities’ own priorities.
Building the field of bioregional economics and finance is a core priority of the BioFi Project. We convened this call to connect leaders in planetary AI with place-based economics and finance practitioners, identify needs on the ground, and examine how AI tools are already being used, and where collaboration across disciplines could make them more useful. This article carries that work forward by sharing the four approaches presented, the questions they raised, and opportunities for the field to pursue together.
...
There were far more questions in the session than we could answer, but I want to synthesize them into three buckets that have practical consequences for initiatives considering these tools.
First, when does AI add enough value to justify its costs? Given the availability of low-tech and no-AI methods, participants challenged the notion that AI should be a default tool. Audio analysis offers a clear use case: processing recordings at volumes people cannot manually review. But each application needs its own comparison: what improves, and at what cost to resources, our attention, and our precious ecosystems? That assessment also depends on how the model runs. Sharfy and Gregory highlighted the potential of open models running locally on small devices as alternatives to large cloud services.
Because most of AI’s energy and water footprint comes from large models running in cloud data centers, a small model doing one well-defined task on a local device can deliver much of the value at a fraction of that cost, while keeping the data on the community’s own hardware. The trade-off is that small models have real limits, and the devices still have to be made and powered, so local is not free. The useful principle is to treat a small, local model as the first option to consider, and put the burden on the cloud-scale alternative to justify itself.
Second, who decides what knowledge becomes data, and how it is used? GainForest’s experience makes this question specific: communities may permit publication of some observations while withholding others. Juliana Bohórquez of COBALT added a further distinction: some knowledge should not be converted into digital data, and access to information does not confer the understanding that comes through experience and relationship.
Sharfy raised a related challenge: information scattered across platforms could become more useful if connected, but how do communities’ conditions for sharing it carry across those systems? A community might allow researchers to analyze species recordings for a restoration project without agreeing to have the recordings published, used to train a commercial model, or shared with other organizations. Connecting datasets therefore requires more than technical compatibility: each recipient needs to know which uses the community has authorized, and further uses need a process for seeking permission.
Third, what enables communities to govern these tools? Participants challenged the assumption inherent in many tech-driven approaches: that communities already have the power and capacity to make transparent, accountable decisions together. Social fragmentation and unequal decision-making power can make that difficult. Anyone offering a tool to a community needs to ask who can participate, who reviews the results, and who has authority to act. Buying equipment or installing a model does not resolve those questions.
The discussion pointed to several efforts to support collective deliberation. Regen’s use of PACTO helps communities document and revise their priorities toward an agreement. BLOOM Project (not present on the call but close colleagues of ours) supports citizens assemblies, creating opportunities for communities to deliberate. Amy described how partners in an Amazon initiative agreed on shared values and principles before making implementation decisions, giving them a basis for assessing choices about infrastructure and data use."
(https://thebiofiproject.substack.com/p/can-ai-catalyze-the-regeneration)
Discussion
What needs connecting, and what we did not cover
For bioregional initiatives, using these tools involves more than choosing software. It requires connecting technical capabilities with local knowledge, existing projects, and community decision-making. The discussion highlighted two areas where collaboration could help.
Curation and connections across fields. Participants from Ojai, California brought a concrete ambition: rehydrate soils across the Ventura River bioregion as part of becoming better prepared for wildfire. They asked for help identifying relevant approaches, finding collaborators, and connecting with a cohort they could consult regularly. Their request exposed a gap between presenting individual tools and helping a community assemble what a project requires: ecological knowledge, suitable data, technical support, and people with experience applying them. This suggests a role for curation within existing bioregional networks: starting with an initiative’s priorities, identifying relevant expertise and tools, and connecting people around the work already underway.
But working at a bioregional scale is not only useful to communities; it can also be fertile ground for digital and AI tool providers: being close to a specific place grounds their models in real conditions, surfaces the local knowledge that makes data meaningful, and shows them what communities actually need before they build it. In this sense, we imagine the most useful collaborations not as a one-way transfer of tools, but as a mutual exercise in which the builders learn as much as the places they serve. In the words of Amy: “In our digital public good program Project Centinela, Planet learns what problems our imagery and analytics can help solve for conservation leaders, and where new tools, guidance, or capacity development can make the difference between whether a species persists or perishes or a community is rewarded for its good stewardship. We think of this program as a partnership between Planet and teams on the ground; together, we can have greater impact than we could in isolation. The result is better tools and a healthier planet.”
Connecting information without surrendering control. Sharfy described observations from information spread across different platforms: satellite imagery in one system, audio recordings in another. Combining them could help practitioners understand what is happening in a place, but how can permissions travel with the information? Gregory pointed to conditional-use knowledge commons and work on permissioned sharing. The challenge is to make information usable across systems while retaining communities’ authority over who can access it and for what purposes. It calls for developers and community representatives to design the technical connections and rules together.
Time limited our exploration of questions already central to bioregional economics and finance: how to keep resources circulating locally, map economic flows, identify opportunities for local production, and prioritize investment. We did not examine in depth where AI could assist with these tasks or with designing and managing local currencies. BioFi is particularly interested in whether these tools can help place-based initiatives assess investments and assemble portfolios that advance their regeneration strategies.
What stayed with me after the call was the hunger for collaboration. Bioregional regeneration spans biodiversity, culture, social justice, governance, technology, and finance, yet we often work within those verticals separately. We need more horizontal connections so that people working on different parts of a bioregion’s regeneration can plan and act together. That includes deciding what role AI should play in their work—a conversation made more urgent by how quickly these technologies are being adopted, often before communities have had a meaningful say.
AI offers powerful ways to interpret the volumes of data we now collect. But collective human sense-making also draws on lived experience, relationships, and knowledge carried across generations - ancestral intelligence that cannot simply be transferred into a model. We can ask AI to identify a bird call, summarize a meeting, and develop digital twins of bioregions. Deciding what a forest needs, whose concerns an agreement overlooks, or whether to use AI at all requires people to deliberate together.
For the BioFi Project, this exploration sits within our ongoing work to build life-aligned economies: helping bioregions organize capital, develop economic strategies, and learn from one another. Our partnerships and Community of Practice offer places to examine whether AI can support those efforts, under conditions communities help define. We approach these tools with curiosity, bringing their possibilities and limitations into the relationships and work already underway."
(https://thebiofiproject.substack.com/p/can-ai-catalyze-the-regeneration)