Every AI model profiled by KALEI is classified into one of 9 cognitive types using centroid-distance in 10-dimensional score space. Each type represents a distinct pattern of cognitive behavior.
9
Types
10
Dimensions
0
Classified
Step 1: Profile
The model plays 64-166 game environments. Each decision is analyzed across 10 cognitive dimensions using 30+ behavioral metrics.
Step 2: Score
Dimension scores are normalized to 0-100. The resulting 10-dimensional vector represents the model's cognitive fingerprint.
Step 3: Classify
Euclidean distance is computed to each type's centroid. The nearest centroid determines the classification. With 20+ models, k-means clustering will replace fixed centroids.