What is the AI-Index?
The AI-Index quantifies an author’s content rendering across Large Language Models (LLMs) and AI Platforms (like ChatGPT, Claude, etc.).
Definition: Normalized square-root-log transformed LLM views for a content item.
Formula:
\[ S_c = 1000 \cdot \frac{\sqrt{\ln(V_c)}}{\sqrt{\ln(V_{\max})}} \]
Where:
- Vc = total LLM views for content c in the time window.
- Vmax = maximum LLM views among all author’s content in same window,
fixed global normalization constant (default: `50,000,000`, configurable) - ln = natural logarithm (base e ≈ 2.71828)
- Score for best-performing content (Vc = Vmax) = 1000
Rationale:
- Logarithm: Compresses wide dynamic range (handles 10 to 1,000,000+ views)
- Square Root: Further smooths distribution, reducing outlier dominance.
- Normalization: Division by √ln(V_max) normalizes scores against a fixed global V_max for cross-author comparability.
- Scaling: 1000× multiplier produces interpretable integer values (0-1000 range).
