Generative AI Safety and Ethics

Study Large Language Models, AI safety alignment, prompt injection, copyright, and societal risks.

Questions

20, in 5 parts: Principles of Generative AI · Hallucinations, Bias, and Toxicity · Trick Questions · Intellectual Property and Copyright · AI Alignment and Long-Term Safety

Chapter

Artificial Intelligence

Game

Citadel

Pack language

English

A taste of what is inside

1. Artificial intelligence models capable of generating new text, images, audio, or code based on training patterns are:Discriminative AIGenerative AIRule-based expert systemsSymbolic AI
2. When a generative AI model outputs confident but factually incorrect or fabricated information, it is a:Data breachGradient explosionHallucinationOverfitting
3. Is information produced by a Generative AI text model guaranteed to be factually accurate?No, LLMs output token probabilitiesNo, unless written in PythonYes, AI connects to realityYes, LLMs output no false facts