Episode 28: LLMs Are Not Human Subjects, March 4 2024
Mystery AI Hype Theater 3000
Exploring Algorithmic Bias and Model Decomposition
Delving into the intricacies of algorithmic bias, the chapter critiques oversimplification and emphasizes specificity in addressing biases across different axes of domination. The discussion highlights the need to decompose models to uncover biases among various human subpopulations, touching on methodologies involving socio-demographic backstories and the importance of algorithmic fidelity in social sciences.
00:00
Transcript
Play full episode
Remember Everything You Learn from Podcasts
Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.
