IAPP Artificial Intelligence Governance Professional (AIGP) — Question 50

What is the best method to proactively train an LLM so that there is mathematical proof that no specific piece of training data has more than a negligible effect on the model or its output?

Answer options

• A. Clustering.
• B. Transfer learning.
• C. Differential privacy.
• D. Data compartmentalization.

Correct answer: C

Explanation

Differential privacy is a method that allows for the training of models while ensuring that the contribution of any single data point is minimized, thus providing mathematical assurance against overfitting. The other options, while useful in their contexts, do not offer the same level of formal guarantees regarding the influence of individual data on the model's behavior.