Explainable AI Analysis & Visualization: Applications from Brain Computer Interfaces to ChatGPT
summary
Shows an open source JavaFX tool called Trinity that visualizes hyperdimensional vectors from a 418‑sensor non‑invasive BCI projected into a 12‑D semantic space, where temporal trajectories reveal daily drift and retraining needs; applies UMAP projections and clustering to both BCI data and ChatGPT embeddings, using 50 permutations of masked text to build point clouds and convex‑hull manifolds, then measures Euclidean distances to flag likely machine‑generated content — not perfect but scalable and practical for automated screening.
AI-written summary · about
chapters (10)
- 0:00Explainable AI analysis
- 1:32Brain computer interfaces
- 7:09Neural semantics parallels
- 9:39Non-invasive BCI experiment
- 11:47Trinity BCI visualization demo
- 22:14AI-generated content challenges
- 28:33Permutations for AI detection
- 32:20Trinity ChatGPT demo
- 42:38Detection with matrices
- 46:07Trinity future potential
related talks
52mKeynote: Trinity Hyper Dimensionality Visualisation Tool
01h 00mData, visualization, and designing AI
01h 03mIt's All About AI
50mWhat Comes After ChatGPT? Vector Databases - the Simple & Powerful Future of ML?
35mTranslating Brain Electrical Signals into Actions through Models...
46mCan you trust your AI?
01h 01mKeynote: Transformers: The Rise of ChatGPT
31mDecoding AI: A Go Programmer's Perspective