The Agriculture Innovation Mission for Climate (AIM for Climate) partnered with the Enterprise Neurosystem, a research community of leading academic institutions and chief scientists of America’s top technology companies, to host the AIM for Climate Grand Challenge: Leveraging the Power of AI and Machine-Learning.

Introduction – The AI Neurosystem

The Enterprise Neurosystem is an open source AI research community, founded on the principle that all the species and ecosystems on this planet are part of a single system. And humanity has had an unprecedented impact on this system with the advent of climate change. A global-scale AI network is now required – to understand the multitude of contributing factors, and to generate new solutions to this crisis.

There are now 190 participants from over 30 companies and academic institutions who contribute their time to this non-profit open source research initiative. This includes volunteers from institutions like Stanford SLAC, UC Berkeley, EY, IBM Research, Intel, Meta, Reliance Jio, Seagate, Verizon and Yahoo!.

The Enterprise Neurosystem lends direct support to leading international climate organizations like the UNFCCC TEC and CTCN, and AIM For Climate, the international initiative for climate-smart agriculture and food systems, sponsored by the United Arab Emirates and the United States.

The Global AI Network

Hundreds of climate projects exist in the world today. Government initiatives abound, and advanced technology providers offer powerful technology platforms. These platforms cover a wide variety of areas – mapping methane with satellite imagery, tracking animal and plant populations, crop yield analysis, water levels in reservoirs and waterways, large-scale cloud infrastructure for nation-scale deployments, and so on.

To place this collective ecosphere back into balance, a global AI neurology is required to understand the health status of our planet. It would link all the relevant climate networks and data sources in one overarching framework, and conduct a higher order analysis across all points of reference in real time.

It would act as an early warning system – and monitor crops, drought conditions, air pollution, ocean temperatures and other impacts of climate change. And in terms of food security, it could detect patterns across the planet, and recommend optimal planting and harvest cycles, water conservation methodologies tailored to each region, and heat-resilient seed varieties.

It would also enable a missing link to the most critical data of all – allow the species that share this planet to communicate these impacts back to us, through sensors unobtrusively deployed in their habitats. Because humanity needs to draw closer to nature, not create more distance. It is necessary to intertwine a sensor network in nature itself, and observe thousands of species to better understand our effect on the planet. Not only do they share this environment with us, they support our agriculture, and can provide early warnings for climate events – for example, the ability of various species to predict earthquakes is well-documented. To better manage our impact, they must become our partners in this endeavor.

In addition, there are natural causes of climate change that have occurred over millions of years – for example, large scale volcanic eruptions can be a significant factor. The data from natural causes will be incorporated with the human-generated climate data, to create the most accurate ground truth for climate events across the planet. Accuracy of prediction and course correction requires this approach.

To be able to cross-correlate all this real-time and historical data, in turn, enables the deepest possible analytics and pattern recognition capability for the collective good of humanity. And to date, there has been no project to unify all these information sources and projects in a single global system. Individual initiatives range from sensors embedded in the oceans and forests, to government-funded research satellites. Yet
they are all still locked in separate silos of operation and analysis. There is often valuable actionable data that is simply out of reach, or unknown to groups that could both utilize and add to those data sets.

Creating a single open AI framework will allow all academics, companies and governments to plug into this system and drive deeper insights. This approach leads to the preservation of all the many excellent projects already in the field, while enabling a larger AI network effect across thousands of data sources. And this will be accomplished as a non-profit open source architecture, to best accommodate
multinational participation.

And an overarching set of directives need to be enacted, to ensure that all these systems operate in a responsible manner – drawing from carbon neutral energy sources, and built with green technologies and renewable components. In summary, we need to unify all these data sources and sensor networks. Use the capabilities of AI to look deeply across all climate data in real time, and contrast it to historical data, to better understand climate change dynamics. And to implement a system of this size and scope in an environmentally responsible manner.

What is required is an AI neurosystem – a single network that spans the planet, integrating all existing climate data sources and projects, and then cross-correlating this vast trove of information to search for deeper causes and conditions. And in turn, uncover and present new solutions, for restoring balance to our environment.

Join a Workstream

We established a number of collaborative workstreams to build this framework. Our community is working on a variety of projects, ranging from AI catalog architecture and networking infrastructure, to unique AI models for specific solutions. You are welcome to join where your own interests are.

Central Intelligence Platform

This will be the core framework where the AI models reside and operate. This workstream proposes a self-describing digital asset catalog as a foundation for community use, and will eventually lead to a cross-correlation AI engine for deeper pattern analysis. 

Secure AI Connectivity Fabric

This track is building secure connectivity between AI models, data resources and the cross-correlation engine. It will use application layer messaging techniques found in other open source projects, in conjunction with a new policy engine.

AI Signal Processing

The core of this code base was donated to the open source community by IBM Research. This workstream focuses on tools and solutions that can process high-rate signal data for acoustics, vibrations and other areas of analysis. For example, this is being used to study global bee populations, to help prevent their decline.