Overview
This course explores the challenges and risks associated with machine learning systems not aligning with human values. The learning outcomes include understanding the alignment problem, ethical considerations, and potential existential risks. The course covers topics such as machine learning systems, training data, fairness, and ethical issues. The teaching method involves lectures and real-world examples. This course is intended for individuals interested in AI ethics, machine learning, and the societal impact of technology.
Syllabus
Intro
walter Pitts
Imagenet
The alignment problem
Machine learning systems
Training data
Labels in the wild
Representation and data sets
What can we do about this
TCAV
Fire Truck
Robustness
Open Category Problem
Examples
What makes a system fair
Predicting reoffending
Ground truth
Potential ethical issue
Criminal justice
Risk vs Needs
Reinforcement Learning
Facebook Reinforcement Learning
Coast Runners 3
Sparsity
Designing reward functions
Behavior cloning
Shadow mode
Inverse reinforcement learning
Taught by
The Royal Institution