Human–AI co-learning, tested under friction.
Open research on learning, judgment, collaboration, and the conditions under which AI helps — or gets in the way.

What happens to human learning, judgment, and collaboration when AI becomes part of the cognitive process?

Pyragogy builds theories, protocols, knowledge bases, and open experiments to investigate that question.

We are interested in AI that can challenge, compare, expose assumptions, preserve disagreement, and participate in the work of thinking — not merely produce fluent answers.

But we do not assume that human–AI collaboration is automatically beneficial.

The work matters only if it can also reveal where our own ideas fail.

That is why the current Pyragogy ecosystem increasingly separates:

claim from source
problem from proposed solution
adoption from implementation
evidence from interpretation
agreement from independent confirmation

A result that weakens one of our hypotheses is still a useful result.

Calling AI a peer in Pyragogy is a design choice about its cognitive role.

An AI system may question, propose, search, synthesise, criticise, or participate in a learning process. That does not make it equivalent to a human participant in responsibility, authority, or accountability.

Cognitive participation is not a transfer of human responsibility.

The point of the peer framing is to make the interaction investigable: what changes when the AI is allowed to push back, hold a role, introduce friction, or influence the path of inquiry?

Likewise, cognitive friction is a research hypothesis, not a doctrine. Pyragogy studies when deliberate resistance improves thinking, when it merely adds cost, and when a different intervention would be better.

The current research ecosystem can be entered through four simple questions.

How should human–AI learning be structured if the goal is stronger judgment rather than frictionless completion?

The Pyragogy Syllabus is a living knowledge base for human–AI learning, cognitive friction, and epistemic autonomy. It organises concepts, risks, protocols, and observable markers as a connected knowledge graph.

It is where Pyragogy's educational ideas become inspectable structure.

→ Explore the Syllabus

How can new knowledge enter a shared graph without letting an AI silently rewrite it?

CIP-KGE is a human-gated protocol for turning tacit expert knowledge into bounded, reviewable proposals.

The AI may assist with extraction. It does not own the graph and it does not write directly to it.

The protocol is built around traceability, falsifiability, locality of change, and explicit human validation.

What happens when collaborative theory meets conditions it did not describe?

UnPeeragogy began by pressure-testing the Peeragogy Handbook against real friction. It now studies the distance between theory, practice, incidents, interpretations, counterevidence, and revision.

Its purpose is not to debunk collaboration.

Its purpose is to discover its boundary conditions.

Peeragogy → UnPeeragogy → Pyragogy
Continuation through friction, not rejection.

→ Explore UnPeeragogy

What actually helps groups work together — and where does it stop working?

Working Patterns studies organisational practices as interventions rather than slogans.

Instead of saying "async is inclusive" or "consensus works", it asks which intervention, for which problem, in which context, with which outcomes, according to what evidence, and at what cost.

It also contains a clearly separated research track for human–AI groups and agentic workflows. Those AI patterns are research questions, not validated answers.

→ Explore Working Patterns

These projects are not separate products. They form a research loop.

LEARN
Syllabus
   ↓
INTERROGATE
Cognitive Interview Protocol
   ↓
BREAK
UnPeeragogy
   ↓
TEST
Working Patterns
   ↓
revise the knowledge
   ↺

The Syllabus gives us structured ideas.

CIP gives us a disciplined way to propose changes.

UnPeeragogy looks for the places where theory breaks.

Working Patterns asks which interventions survive contact with evidence and practice.

Then the knowledge should change.

That loop is closer to what Pyragogy has become than any single model, tool, or manifesto.

Research writing, field notes, and ongoing reflections also live on the Pyragogy blog.

Pyragogy has changed substantially while being built. Earlier repositories are kept because they show that evolution rather than pretending the current method existed from the beginning.

We do not treat age as evidence and we do not silently rewrite the past. Older artifacts can remain useful while their claims are re-examined under newer standards.

• AI should be able to introduce useful disagreement, not merely optimise agreement.
• Evidence and interpretation should remain distinguishable.
• A plausible mechanism is not the same thing as demonstrated effectiveness.
• Human oversight must mean real authority, not ritual approval.
• Costs matter — including time, cognitive load, emotional labour, and who bears them.
• Unknown is a valid research state.
• Counterevidence improves the project.
• Theories, protocols, and AI-generated proposals must remain revisable.

This is the practical meaning of the Pyragogy idea:

AI that argues, not AI that agrees.

Not because disagreement is always better — but because unexamined agreement is a poor foundation for learning.

If your interest is learning with AI, start with the Syllabus.

If your interest is knowledge governance, read the Cognitive Interview Protocol.

If you want to study where collaborative theory breaks, explore UnPeeragogy.

If you care about how groups actually work, start with Working Patterns.

If you want the broader story, read the blog or visit pyragogy.org.

Useful contributions include code and prose, but also things that make the research harder to fool:

• evidence we missed;
• counterexamples;
• failed implementations;
• boundary conditions;
• incorrect citations;
• better explanations;
• replications;
• evidence that one of our conclusions is too strong.

Read CONTRIBUTING.md and CODE_OF_CONDUCT.md.

For discussion, use the Pyragogy forum.

General: info@pyragogy.org
Security: SECURITY.md
Web: pyragogy.org

An open Pyragogy research ecosystem.
Build. Test. Contradict. Revise.