Generating Prompts and Testing Generative AI (Theory)

Arthur Kakande provides an introduction to Prompt Engineering and Adversarial Testing for Large Language Models (LLM). Kakande describes various prompt generation techniques such as zero-shot, few-shot, and chain-of-thought. Special attention is paid to LLM moderation to prevent malicious content and prompt injections. Adversarial testing, a method for identifying vulnerabilities in LLMs by deliberately introducing complex data, is demonstrated.