Digital Gaia
Public.icon
TL;DR:
Digital Gaia 2 is building a decentralized infrastructure for measuring and certifying environmental impact.
The infrastructure is based on distributed science and uses algorithmic, probabilistic, and transparent impact prediction and measurement.
It connects a global community of conservationists, experts, and data scientists with an incentivized knowledge feed.
It has an Impact Oracle that can back dynamic Proof of Impact NFTs.
Our view on Proof of Impact (PoI)
Tokenized impact certificates must be far more legitimate than legacy certificates. Most proposals assume such legitimacy is derived from proof by trusted governance institutions and have clearly defined immutable "face values," i.e., indicators of success (ideally based on pre-criteria) that are (implicitly or explicitly) assumed. This is reflected in Vitalik's own proposal1 for a Result Oracle for public goods funding. We believe this kind of model is insufficient for efficient pricing of externalities and particularly inadequate for natural impacts involving context-dependent, long-term interactions of global and local factors. In our proposal, the fundamental Web3 primitive is a dynamic, algorithmic evaluation of intervention impact. This judgment is achieved by autonomous (AI) agents that integrate available evidence and knowledge (provided by humans who are not fully open or trusted) using generally accepted principles of science and decision theory, such as Bayesian estimation and free energy minimization, and weight them. The source of its legitimacy is the scientific method itself. It is based on the developmental and transparent paper trail of inter-verified inferences from actual data, not on authority.
The impact tokens and the economy built on top of them creates a continuous incentive for the community to provide further knowledge, and creates a commons of impact assessment expertise. Ultimately, one can mint high-grade impact credits as derivatives (with a "99.99% certainty that this project removed one ton of CO2") to create the foundation of public goods investment with self-funding impact investing algorithms. \n> \n> About Digital Gaia\n> Our team and partner networks have expertise in both collective intelligence and deep-learning AI, data science, economics, community engagement, and climate business. We are building (1) the aforementioned primitive, (2) an MVP-contribution engagement app for conservationists and scientists, and (3) a Python SDK for data scientists. We have already built PoC for most of (1) and (2) and are verifying our architecture descriptions in Draft White Paper 2 with global leaders in AI, data science, and collective intelligence. \n> Asks\n> We are raising $200,000 to complete MVP development over the next 3-6 months. \n\n\n\sWill this become a Result Oracle?\n\s\sOf course, it will become a context of Retroactive Public Goods Funding.\n\s\sTherefore, I think we need more retroactive feedback from not only Degitak Gaia, but also more people. In this case, it's like a "group of evaluators". Previous companies, interviews, and appearances related to oneself.icon\n\s\s\shttps://www.digitalgaia.earth/\n\s\s\s\s>[Steward] optimizes, demonstrates, and monetizes the project's impact, unlocking the value of the circular economy and providing investors with clarity, reliability, and accountability\n\s\s\s\sIt sounds like it. tkgshn.icon*2\n\n\shttps://digitalgaia.notion.site/Natural-Intelligence-fa45119fa6224965b63c9cc2e0181dd8\n\s\shttps://gyazo.com/f0065075874d6510085ac437648887a6\n\s\s\s># Summary\n\s\s\s> The first white paper of Digital Gaia provides a technical overview of the **Natural Intelligence Network (NIN)**, a rooted (place-based) collective of artificial agents designed to pursue the regeneration of the Earth's environment and the prosperity of the economy. - NIN is based on the ActInf (Active Inference) framework for building Bayesian optimization decision agents, and its implementation makes extensive use of state-of-the-art product-level Bayesian modeling libraries. - Each agent of NIN is embedded in a real-world system, evolves with that system, and can independently inquire, learn, plan, and allocate resources online through active inference for its context.
- Agents interact with humans and other agents through a protocol of "quests," "claims," and "rewards" that encode Bayesian inference at multiple levels, providing principled support for unreliable inference, dynamic model averaging from cloud-sourced model worlds, and complex forms of inference and communication.
- As a context-dependent, distributed network of agent-based agents, NIN can gradually grow and evolve across spatiotemporal scales and domains, adaptively discovering global goal states and avoiding most known AI risks.
- All network activity can be fully stored in an additional public database with field-level encrypted signatures, allowing both humans and other agents to investigate and replicate agent inference and decision-making.
- NIN's agents function as "assistants" to support human and traditional organizational decision-making in the early stages, but may evolve into autonomous economic actors managing resources and organizations.
- Thus, a fully agentized NIN may play a critical role in the global transition to intentional positive-sum "Gaianomics."