Episode 1025 • • 2:26:06 - 2:29:39
1025: Blotto in Biloxi
AI Classifiers and Multi-Dimensional Linear Regression
A producer with a master's in AI explains that the systems used by Facebook and Twitter to identify hate speech are "classifiers" based on multi-dimensional linear regression. The AI is fed positive and negative examples to output a binary classification. The hosts simplify the concept as "trying a bunch of stuff" to see what sticks.