Related: sources · notes · metadata · Published Pieces

Education Should Be a Commons Deployment Path

Education should not be a captive market for AI vendors. It should be where society learns to operate cognitive infrastructure.

Education is too important to be trapped inside proprietary AI tutoring contracts.

If automatic radio becomes the personalized tutor people were promised, schools and libraries should have access to it. Universities should have access to it. Community groups should have access to it. Families should have access to it. But that does not mean the core business should become edtech.

Education should be a commons deployment path.

That means the educational version of the system should be open-source where possible, compatible with open-weight models where possible, and deployable on infrastructure controlled by schools, libraries, universities, municipalities, nonprofits, or local technologists. The role of the central platform should be to maintain the protocol, reference implementations, trust standards, artifact formats, interoperability, and public knowledge graph — not to turn every educational institution into a bespoke enterprise customer.

This is not charity as branding. It is the correct architecture.

Schools are not just customers. They are knowledge institutions. Libraries are not just distribution points. They are public memory institutions. Universities are not just buyers. They are research and teaching institutions. If a system exists to improve public cognition, these institutions should be able to run it, inspect it, modify it, govern it, and teach with it.

The worst version of AI education is another closed SaaS contract: a vendor sells a tutoring platform, locks schools into accounts, centralizes student data, tunes the product to procurement requirements, and claims to personalize learning while operating as an opaque commercial layer between the institution and its own students.

That is not good enough.

A better model is local capacity building. A school or library should be able to run a Choir-compatible stack using open models, local data policies, local moderation rules, and local educators. It should be able to host vtexts, automatic radio channels, source collections, student projects, community archives, oral histories, local research, and public-interest curricula. It should be able to connect to the global network when appropriate and remain local when needed.

This creates work for the right people.

Local technologists, teachers, librarians, students, civic hackers, and independent consultants can become integrators. They can help deploy the system, configure models, manage infrastructure, train staff, build curricula, create local content, maintain provenance, and adapt workflows to real communities. That is a better economic pattern than forcing the core company to become a services bureaucracy.

The educational commons path also teaches the right skills. Students should not merely consume AI-generated lessons. They should learn how the system works. They should understand models, retrieval, provenance, citations, versioning, logs, permissions, local inference, and verification. They should know that knowledge systems have architecture and politics. They should learn to ask: where did this claim come from, who cited it, what contradicts it, what model produced this summary, what sources were used, what was omitted?

That is education for the 21st century.

This approach also reduces the pressure to make education the profit center. Education is a high moral-value domain but often a poor business category: slow procurement, custom requirements, underfunded institutions, political fights, compliance burdens, and distorted incentives. If the core product is dragged into institutional edtech too early, the medium will be narrowed before it matures.

Let the main platform develop as a media, research, publishing, and intellectual-property system. Let the educational use cases flourish through open deployments, community versions, guest lectures, reference curricula, and third-party integrators.

This is better for everyone. The core platform stays ambitious. Schools get sovereignty. Technologists get business opportunities. Educators get tools. Students get access. Communities get local knowledge infrastructure.

The standard should be simple:

Education should not be a captive market for AI vendors.

Education should be one of the places where society learns to operate its own cognitive infrastructure.

That is a commons deployment path.