Artificial Intelligence
263 views
Red Teaming AI
Quick Definition
Adversarial testing of AI models for vulnerabilities
Full Definition
Systematic adversarial testing of AI models to identify vulnerabilities biases and safety risks.
Examples
safety evaluation, bias discovery, adversarial probing
Related Terms
ai-safety
adversarial-attack
penetration-testing
More Artificial Intelligence Terms
AutoML
Automation of the machine learning pipeline
Zero-Shot Learning
Performing tasks on unseen classes without training examples
Neural Architecture Search
Automating the design of optimal neural network architectures
AI Fairness
Ensuring AI decisions do not discriminate against groups
AI Alignment
Ensuring AI systems act in accordance with human values
Reward Modeling
Training models to predict human preferences for RLHF