EducAgent is a learner-facing demo that makes causality feel approachable. It introduces causal reasoning through short stories, visual examples, and lightweight checks instead of asking visitors to start with technical notation.
Live demo: vios-s.github.io/EducAgent
- A friendly homepage for a broad public audience: anyone should be able to understand the promise without already knowing causal inference.
- A mobile-first learning experience with compact navigation, responsive lesson layouts, and visual lesson headers.
- Study Mode as the main path into causality basics.
- A visible audience switch for previewing Everyone, CS student, and Healthcare tracks.
- Booth Mode for CHAIFest-style stand-up demos: one question, one hidden-path reveal, a causal graph, and a QR for the 3-minute phone challenge.
- A 30-second feedback page for booth visitors.
- Project and funding links with QR codes for VIOS Group and CHAI at the bottom of the homepage.
The first learner path asks a familiar question:
If students who attended tutoring scored higher, does that mean tutoring caused the higher scores?
The lesson introduces the difference between seeing a pattern and proving a cause. It uses a tutoring story to explain why groups may already be different before an intervention, and why random assignment can help us test causal claims more carefully.
The second learner path currently provides the shell for a DAG-focused course. It previews the structure for learning causal graphs, including nodes, arrows, paths, and why causal graphs cannot loop.
The healthcare bridge is a short, non-diagnostic example for CHAI audiences:
Patients who received a clinic text reminder were more likely to attend their appointment, but does the reminder itself deserve the credit?
It focuses on spotting patient engagement and life stability as possible hidden upstream causes, not on giving medical advice.
This is an early public-facing prototype, not a full course platform yet. The current demo focuses on:
- presenting the EducAgent concept clearly,
- testing a warm visual direction,
- showing one complete beginner-friendly lesson path,
- previewing a second DAG learning path,
- keeping the experience usable on phones.
Future features marked as coming soon on the homepage, such as hidden-path guidance and team-based learning, are intentionally shown as upcoming capabilities.
- VIOS Group - University of Edinburgh
- CHAI - Funded by CHAI
The demo is served from the docs/ directory. The root index.html redirects visitors into the GitHub Pages version of the app.