The question at the heart of this talk: When you provide an answer about pay, can you tell which part should be
calculated, which should be estimated, and which should be generated?

Compensation now runs on three very different kinds of Math. Deterministic calculations apply the rules (merit
matrices, bonus calculations, bonus pool allocations, vesting schedules) that must produce the same answer every
time. White-box models estimate what pay should be and explain why it differs, powering pay prediction and pay
equity analysis. Generative AI, the newest and most talked-about arrival, is increasingly used for explaining the
answers generated by deterministic calculations. When every outcome has an "AI" component, telling them apart
is what lets you use each one where it is strongest.

Key Takeaways:

• Explain the three kinds of math behind compensation: deterministic calculations, white-box regression models and black-box generative models.
• Discern which kind fits different compensation scenarios.
• Break down compensation outcomes into what should be calculated, what should be estimated and what should be generated.