MuDG

Augmented Collaborative Idea Exploration

What is it?

A mind mapping tool that uses AI alongside other people to explore and learn; visualising the learning process.

Who is it for?

Anyone who wants to explore an idea in depth and see where it takes them. Also a chance to see how others think, seeing pathways over complex topics.

More explicitly - students, writers or people who want to have a clearer basis for their understanding.

Why should I care?

Much learning is personal, often we see the conclusions of others and keep unfinished or incomplete notes to our self. It was built with 2 principles, learning can be done better together (and AI can help), and that visualising the history of this thinking is valuable not only for ourselves but others.

Try it About

Want to know more?

Follow us on Twitter for updates, insights, and to join our growing community of collaborative learners.

Follow @TJCBerman

Why did I build it?

At a principle level

I have lost track of the number of times when I sit to listen to a speaker / lecture and within a few mins says something I believe raises a huge question. This casts a shadow over the rest of what they are saying.

We live in the Information Age. There is no shortage of information, but at the same time, there is a frustration at the general quality of discussion / debate. It has been called “Echo” chambers or motivated reasoning. Most discussions between opposing viewpoints completely talk past each other.

I believe this happens because, much like a magic trick, it happens outside of your awareness / the parameters of the discussion.

Different assumptions lead to completely different conclusions. For example debating if an action is ethical entirely depends on the ethical framework you are using.

Without the ability to question at word level and question these assumptions, then we miss out on productive exchanges.

At a specific level

The AI revolution and learning. Having a tool that can answer many specific questions and has a great deal of knowledge about wide range of topics is an unprecedented tool for learning. To make learning an active process, to ask questions that interest you and a very granular level is in itself an engrossing process.

I think it currently has a few drawbacks:

  • Linear walls of text make it challenging to take the conversation in several directions at once.
  • They are inherently single player - it is you talking to the tool. I think bringing in other people can lead to more interesting directions.
  • The conversations seem disposable. I often have very interesting discussions, but I rarely if ever return to a conversation. Meaning building bigger ideas becomes difficult.

Example

Exploring different ethical frameworks

The example shows a graph looking at different ethical frameworks. We can define different frameworks and explore their implications. For instance, we can explore the implications of utilitarianism, deontology, and virtue ethics.

Vs. ChatGPT

The conversation becomes a wall of text, to explore specific topics you need to scroll up and down, meaning the flow of the conversation is lost.

1) Idea graph

Visualise the paths through an idea

This screenshot is a graph exploring ethics. As we can see we have several approaches to ethical decision making. We can explore any path through the idea, whatever approach seems most interesting at the time.

We can come back whenever we want to other paths and pick up where we left off. Now we can explore each approach in detail, asking for an example, and branching through between the Pros and Cons.

Ideas have clear relation to others, with parents and children, showing how differing conclusions come from earlier decisions.

Ethics Knowledge Graph

2) Multiplayer

Real-time Collaboration

Each graph can be edited by any person on the site. Each change is shared with everyone in real time. Each node can be noted for later reading and to surface the most interesting points.

You can lock a graph when you feel it is a finished idea, for easy sharing to students.

3) Shareable

Share Any Node

Each node has a link associated with it. So you can have a unique URL for any node on the page, that can be shared with anyone else. This means you can return to any point and continue from where you left off.

4) Creative Inspiration

Explore New Ideas

This starts from the landing page. Not sure what interests you? The ideas pages give you a starting point, with many interesting topics to explore.

The interaction with the language model hopefully inspires new ideas and generates debate points for any topic or helps find the balance between opposing perspectives.

Ideas

About MuDG

Our Vision

Muti-user Dialectical Graph (MuDG) began with a simple observation: the revolutionary potential of large language models for education remains largely untapped. While today's AI assistants offer impressive capabilities, they exist primarily as isolated, ephemeral exchanges. Our vision is to transform these interactions into persistent, evolving idea structures that grow richer with each exploration.

The Knowledge Graph Paradigm

At MuDG, we believe the future of learning isn't linear—it's networked. Traditional educational paths often follow predetermined sequences, but human curiosity rarely operates this way. We naturally branch, connect, and integrate ideas across domains. Our knowledge graph approach mirrors this natural thought process, allowing learners to move freely between concepts while maintaining coherent connections.

Each node in a MuDG graph represents a discrete concept, question, or insight. Unlike traditional learning materials that might bury important connections in paragraphs of text, our graph structure makes relationships explicit and navigable. This visualization of knowledge helps learners identify patterns and connections they might otherwise miss.

Collaborative Intelligence

MuDG combines the analytical power of language models with the creative, contextual understanding of human minds. While AI excels at retrieving and synthesizing information across vast knowledge domains, humans excel at determining relevance, identifying novel connections, and applying insights to real-world contexts. By designing a platform where these complementary intelligences work together, we create learning experiences greater than either could produce alone.

Collaboration extends beyond the human-AI relationship to encompass peer-to-peer learning as well. When multiple users explore and expand a knowledge graph together, they contribute diverse perspectives and expertise. This multiplayer aspect transforms learning from a solitary activity into a social one, where discovery is shared and insights are collectively refined.

Technical Philosophy

Our technical approach is guided by several core principles:

  • Speed matters. Cognitive flow is easily disrupted by technical friction. We've optimized every aspect of the interface to minimize latency and maximize responsiveness.
  • Simplicity enables depth. Complex ideas don't require complex interfaces. By eliminating unnecessary elements and focusing on conceptual clarity, we allow deeper engagement with content.
  • Privacy is foundational. Learning requires experimentation and vulnerability. We've designed a platform where users can explore ideas freely without concerns about data harvesting or surveillance.
  • Persistence creates value. Unlike ephemeral chats that disappear, MuDG knowledge graphs persist, accumulating value over time as they grow and connect.

The Future of MuDG

We're just beginning our journey to reimagine how humans and AI can learn together. Future developments will include specialized tools for educators, expanded visualization options, and deeper integration capabilities with existing knowledge bases and learning resources. The goal remains constant: creating the most effective collaborative learning environment possible, where the boundaries between teaching and learning dissolve, and where knowledge isn't just consumed but actively constructed.

MuDG - today

This has been a personal project, that in many ways is a cumulation of my different interests. Understanding / learning, design, (quasi) mathematical, AI and programming.

The inspiration of the idea came from three distinct locations.

  • The rise of AI / LLM’s - back in my undergraduate degree (too many years ago!) I remember learning about neural networks, implementing an MLP in C and other less popular networks such as Kohonen Map (Self Organizing Map). To capabilities of the deep learning revolution happen in real time has been inspiring.
  • Dialectics - going back to the very birth of philosophy,  form of reasoning based upon dialogue of arguments and counter-arguments.
  • The Elixir community - Since working with Elixir and Phoenix LiveView, it has made building interactive websites a pleasure and made alot of what the project is possible.
Give it a try

Want to know more?

Follow us on Twitter for updates, insights, and to join our growing community of collaborative learners.

Follow @TJCBerman