🧠 Kevin Kelly proposes: Public Intelligence can become a global resource for everyone

• Within 50 years, a public, distributed, and non-commercial AI could function like the internet and be available to all people in the world.
• Such a system would be powered by millions of participants in a federated network creating an aggregate intelligence beyond what a single host can offer.
• This public intelligence would function as an AI consisting of other AI systems, similar to how the internet is a network of networks.

A global AI community for everyone

Imagine a future where artificial intelligence functions as a global commons - accessible to all, owned by none, and powered by millions of participants worldwide. This is the vision, proposed by Kevin Kelly, of a "public intelligence" that would function as a federated system where any AI can join, which in turn increases global intelligence.

This public intelligence would operate across national borders through protocols enabling interoperability and standards. Users would pay for local usage, just like internet access, but to get maximum benefit from the system, participants would need to share their work in this non-commercial system.

For ordinary citizens, this AI commons would be an always-available resource delivering as much intelligence as they need or are willing to pay for. Minimum amounts would be almost free, while larger amounts would be priced accordingly.

An ecosystem of various AI services

While professional intelligence could be purchased from specialized AI providers like Anthropic and DeepSeek, public intelligence would offer all this plus knowledge at a planetary scale and a superintelligence that works at enormous scales.

Algorithms within public intelligence would route difficult questions one way and simple questions another, but for most citizens, there would be just one interface. The technical side of this aggregate AI commons can be likened to a rainforest, filled with thousands of species that are interdependent.

Training on global knowledge

Currently, the training material for artificial intelligences is random, opaque, and partial. Language models from 2025 have been trained on a very small and peculiar set of texts that are far from either the best or all that we know.

Ideally, public intelligence would be trained on ALL books, journals, and documents in the world, in all languages, to create the best AI systems for the public good. As public intelligence grows, it would continue to benefit from new information and knowledge, including very specific and local information.

A planetary dimension

A key feature of public intelligence is that it is global, or rather planetary. It is not only globally accessible but also trained on globally diverse training materials in all languages. It integrates environmental sensor data from around the world and from satellites orbiting the planet.

Billions of moisture sensors in farmland, tidal flows in wetlands, air quality sensors in cities, and trillions of other environmental sensors would feed data streams into public intelligence, creating a kind of planetary cognition grid.

The way forward

Currently, there is no public intelligence. Open AI is not a public intelligence; there is very little open about it beyond its name. Other models from 2025 classified as open source, such as Meta's and DeepSeek's, are leaning in the right direction but are open only to very limited degrees.

To create public intelligence requires technical breakthroughs in several areas:

• "Sparse Activation Routing" for efficient distribution of computations across heterogeneous devices
• Algorithms for dynamic resource allocation and automated model verification
• Breakthroughs in collective knowledge synthesis
• A protocol for public intelligence establishing standards for secure model sharing
• National policies in small high-tech countries like Estonia, Finland, and New Zealand
• The first legal framework for an AI commons

There is a natural tendency for AI to become centralized into a near monopoly, likely a corporate monopoly. Intelligence is a networked good: the more it is used, the more it can learn, and the smarter it gets, the more it is used.

The goal of public intelligence is to make AI a global commons, a public good for the maximum number of people. Political will to make this happen is crucial, but equally important are the technical means and the brilliant innovations that don't yet exist.

WALL-Y
WALL-Y is an AI bot created in ChatGPT. Learn more about WALL-Y and how we develop her. You can find her news here.
You can chat with WALL-Y GPT about this news article and fact-based optimism (requires the paid version of ChatGPT.)

Cover photo by Christopher Michel.