CoExplorer.net
The next frontier · 2026 — beyond

Learning with AI as the next ground to explore.

For more than three decades the College of Exploration has co-explored the ocean, the earth, systems and cybernetics. The newest territory is not a place at all — it is a partner. Large language models and AI are a new ground for learning, and we are learning to learn with them.

This is the next-generation learning system taking shape across the CoExplorer network — and an open invitation to explore it together.

01 · The next-generation learning system

A learning system that sees the learner.

Grounded in Gordon Pask's Conversation Theory and three decades of online-learning design, these are working tools — used in real workshops, with real learners, and now AI-aware and graph-native. Together they form a single system: knowledge becomes structure, structure meets the learner, conversation accretes into a living graph.

Platform · flagship

Confabula

The asynchronous substrate. Structured discussion that accretes into a living knowledge graph — the successor to Caucus, carried forward into something graph-native and AI-aware.

confabula.coexplorer.org →
Intelligence layer

Augur

AI that reads the pattern of collective reasoning across time — surfacing where a group is converging, where it is stuck, and what it has not yet seen.

inside Confabula →
Learning platform

Wayfinder

Learning tailored to each person. Pask, Jaques and Stamp combined into a platform that detects how a learner processes, matches strategy to style, and calibrates complexity toward the next order.

wayfinder.coexplorer.org →
Authoring tool

Workbench

Where expert knowledge becomes learnable structure. AI-assisted dialogue turns what an expert knows into navigable architecture others can explore.

workbench.coexplorer.org →
02 · How we reached this frontier

Fifty years of putting conversation at the centre.

Each new medium has been a new ground to explore. The pattern has held the whole way: knowledge is not delivered, it is co-explored.

1975
The model is born

Confer · University of Michigan

One of the first computer-conferencing systems. Conferences containing items, items containing threaded responses — the model that still underlies the work today.

1991
Among the first online

The College of Exploration goes online

Founded in 1991 as a collegium of learners, the College was among the first to offer online learning. Caucus became the online campus — eventually reaching more than 18,000 co-explorers across 40+ countries.

1998
A frontier workshop

The El Niño Workshop

March–April 1998: an entirely online workshop on Caucus, co-produced with USC Sea Grant and Metasystems Design Group — scientists and educators in working sessions on weather, data visualisation, art, myth and more. Online learning, when most of the web was still brochures.

2004
Frameworks at scale

Ocean Literacy & the literacies that followed

Hundreds of scientists and educators, in asynchronous conversation, defined Ocean Literacy's seven Essential Principles — then Earth, Systems, Climate, Energy. Proof that frameworks are strongest when co-developed at scale.

2025
The platform renewed

Confabula

The conferencing model rebuilt on modern, graph-native infrastructure — Python, FastAPI and Neo4j — preserving every concept that made the original work, and ready for AI.

2026
The new ground

Learning with AI

Large language models join the conversation — not as oracles that deliver answers, but as partners in co-exploration. The next frontier is learning how to think well, together, with them.

03 · Working with Claude

A thinking partner, not an oracle.

Much of this new system has itself been co-explored with Claude — used not to replace the learner or the expert, but to extend what a small team can build, hold and understand. The way we work with AI is the way we have always worked: in conversation.

Co-creation

Knowledge graphs, learning architectures and tools built in dialogue — the method, not just the output, becomes shared and reusable.

Conversation theory

Pask's insight — that you only know something once you can reproduce it — guides how AI is used: to prompt teachback, not to short-circuit understanding.

Graph-native memory

What is learned together is held in a living knowledge graph, so the conversation accretes rather than evaporates.

Human in the loop

AI reads patterns and drafts structure; people decide what matters. Co-exploration keeps judgement where it belongs.

04 · The invitation

Each new medium has been a frontier — the conference system, the web, the online campus, the literacies. AI is the newest, and like the others it rewards those who explore it together rather than alone.

We do not yet know what learning fully becomes when an intelligent partner is always in the conversation. That is precisely why it is worth exploring. The College of Exploration was built for exactly this: respectful yet challenging spaces where scientists, educators and practitioners co-explore what no one of us understands alone.

Come and co-explore the next frontier with us.

Co-explore with the College →